Home

User manual v.2.0

Contents

1. ss 105 7 Data limitations amp system requirements ccccccononananannnnnnnononanananannananannncnnnononona nana nanrnnarannnannss 107 8 Troubleshooting ati did cisco 108 9 Tips for using the command Prompt sete cis ca 110 Part IV Tutorial 8 Examples 113 WEX rCISG d piano 115 2 XCICIS SD a a ae Dr it 117 SEXGEMCIS D inner e nette eu 118 L EXGICISS A inner nain R 120 D X FCISR D AU AR rt oo idos 122 D EXGICISR ONU ii ace Ne 123 PEXESO a o aa el ci 124 8 Exercises o ee 126 Index 128 O 2004 2008 Atte Moilanen Introduction 2 1 1 Introduction Aim amp purpose Zonation is a reserve selection framework for large scale spatial conservation planning It identifies areas or landscapes that are important for retaining habitat quality and connectivity for multiple species thus providing a quantitative method for enhancing species long term persistence Zonation produces a hierarchical prioritization of the landscape based on the biological value of sites cells accounting for complementarity The algorithm proceeds by removing least valuable cells from the landscape while minimizing marginal loss of conservation value accounting for connectivity needs and priorities given for biodiversity features species landcover types etc As a result a nested sequence of highly connected landscape structures is obtained with the core areas of species distributions remaining late
2. 4 ZIG2 Iterative cell removal done Showing removal rank Maps Run mode ee ete Calculate new solution pecies file list is in file C Load old solution rank file Removal rule Start of output file name Original core area Zonation Use cost file C Additive benefit function M Use incl excl mask Use SSI file Target based planning Generalized benefit function Use planning unit layer Use interactions file Warp factor moo 2 0250 Resampled species count Info gap settings Use info gap distribution discounting Info gap alpha 1 000 Uncertainty model Uniform error default C Proportional error List of error weight map layers load from u Cweights spp Output The effects of running the analysis with a mask file should be clearly seen from your output map The included areas should receive the highest values in landscape ranking where as the excluded areas receive the lowest values Run settings Species info Memo Landscape identification Solution comparison About SES y _splist spp transform species occupancy probabilities from logit values joutput_ m dat Annotate outputfile name cost asc maskl1 ras asc SSI ist txt PLU asc interact spp Settings for generating spatial aggregation into the solution M Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000
3. 0 008 1 01 0 015 1 01 0 015 1 01 The first column S1 is the focal species the map of which is transformed The number in column one is the ordinal number of the species referring to the order of species in the species list file in the example above the first row indicates that species map number 6 is transformed by connectivity to map number 1 using distance dependence beta B 0 008 Beta is equivalent to the alpha value that is given in the species list file Thus beta is calculated using the same formula O 2004 2008 Atte Moilanen ZIG The Zonation software 66 B 2x Ce l size in km Distance dependence km Input cell size where the species specific measure for landscape use refers to the distance to which focal species S1 in cell can interact with species S2 in the surrounding area of cell The fourth column type specifies whether this interaction is a positive e g resource consumer or a negative one e g competition see section 2 7 in methods The final fifth column gives the value of gamma which in turn defines how fast the value of connectivity between S1 and S2 saturates when moving away from the focal site For more detailed explanation about gamma see section 2 7 By default gamma should be 1 0 When including species interactions remember to select the Use interactions option in Run settings window and type the correct path of your interaction definit
4. ZIG The Zonation software 72 The third curve shows how the extinction risk of species increases as landscape is removed This curve is based on the species area ratio and shows the average extinction risk over all species assuming the exponent z given in the settings The fourth and last curve displays the proportion of distribution remaining for SSI species Species of Special Interest when landscape is removed Also here the red line represents the species with the lowest proportion of distribution remaining and the blue line represents the average over all species Note that unlike the other panels this panel will only be displayed when SSI species are included into the analysis The information of these four graphs is equal to the curves txt file that the program produces as part of file output Below the curve panels is a histogram the species specific habitat quality distribution This is the third basic output of Zonation in addition to the rank map and performance curves Essentially this is the histogram of habitat quality for a selected species in a given top fraction of the landscape enter fraction and select the species and the histogram will be displayed in a graph The respective information is also output into the Memo from where it can be copy pasted elsewhere for further processing The x axis of the histogram is linearly scaled from 0 to 1 where 1 represents the highest quality found in the original in
5. Distribution smoothing info gap uncertainty analysis Moilanen A and B A Wintle 2006 Uncertainty analysis favours selection of spatially aggregated reserve structures Biological Conservation 129 427 434 Basics of the information gap decision theory and accounting for distributional uncertainty Moilanen A Runge M C Elith J Tyre A Carmel Y Fegraus E Wintle B Burgman M and Y Ben Haim 2006a Planning for robust reserve networks using uncertainty analysis Ecological Modelling 199 1 115 124 Moilanen A B A Wintle J Elith and M Burgman 2006b Uncertainty analysis for regional scale reserve selection Conservation Biology 20 1688 1697 A quantitative method for generating reserve network aggregation Moilanen A and B A Wintle 2007 The boundary quality penalty a quantitative method for approximating species responses to fragmentation in reserve selection Conservation Biology 21 355 364 An extension of the BQP method to freshwater systems with different connectivity requirements upstream and downstream Moilanen A Leathwick J and J Elith 2007 A method for spatial freshwater conservation prioritization Freshwater Biology 93 577 592 Accounting for species interactions Rayfield B A Moilanen and M J Fortin 2008 Incorporating consumer resource spatial interactions in reserve design Submitted manuscript Replacement cost analysis Cabeza M and A Moilanen 2006 Replacemen
6. ZIG The Zonation software 54 parameters need to be written on separate rows with parameter names typed exactly as in the examples case sensitive E Run settings dat Notepad Alle File Edit Format View Help settings removal rule 1 warp factor 100 edge removal 1 add edge points 0 use SSI 0 SSI file name S5I_list txt use planning unit layer 0 planning unit layer file plu asc Use cost 0 cost file cost asc use mask O mask file mask ras asc use boundary ey penalty 0 BOP profiles file BOPcurves txt BOP mode 1 BLP use tree connectivity q tree connectivity file tree txt use interactions 0 interaction file interact txt annotate name O logit space 0 treat zero areas as missing data 0 2 0 2 resample species 0 Info gap settings Info gap proportional O use info gap weights 0 Info gap weights Tile UCweights spp Picture of settings file Settings removal rule Determines which cell removal rule will be used 1 Basic core area Zonation 2 Additive benefit function 3 Target based planning 4 Generalized benefit function warp factor Defines how many cells are removed at a time If warp factor is 100 it means that 100 cells are removed at each iteration Thus a lower warp factor leads to a finer solution but also to an elongated running time whereas a high warp factor keeps the running time short but might r
7. Va Fa framework and software Version 2 0 User manual Zonation Software for spatial conservation prioritization by Atte Moilanen Zonation is a spatial conservation prioritization framework for large scale conservation planning It identifies areas or landscapes important for retaining high habitat quality and connectivity for multiple biodiversity features eg species providing a quantitative method for enhancing species long term persistence Essentially this software is a decision support tool for all non commercial parties working on conservation issues As Zonation operates on large grids it provides a direct link between GIS statistical distribution modeling and spatial conservation prioritization The Zonation framework is presently under constant development and the next version of the software can be expected not too far in the future Thus keep an eye on our web site www helsinki fi bloscience ConsPlan Zonation User manual 2004 2008 Atte Moilanen All rights reserved USE THIS SOFTWARE AT YOUR OWN RISK THE AUTHOR WILL NOT BE LIABLE FOR ANY DIRECT OR INDIRECT DAMAGE OR LOSS CAUSED BY THE USE OR THE INABILITY TO USE THIS SOFTWARE CONDITIONS OF USE Zonation v 2 0 is freely usable for non commercial uses For any other kinds of uses contact the author for permission Do not use this software if you disagree with the disclaimer and conditions of use Even though the
8. output which will be used for each of the output files with a varying suffix output jpg output emf output curves txt output prop asc output rank asc output wrscr asc and output run_ info Jpg file An image of the map of the area illustrating the Zonation results ranked by using different colors to indicate the biological value of the site Here the best areas are displayed in red and the worst areas in black with the no data areas marked as white See section 3 4 1 Visual output for more detailed interpretation of the colors used in landscape ranking O 2004 2008 Atte Moilanen ZIG The Zonation software 74 emf file This is an identical image showing your output map but it has a higher quality compared to the jpg file Thus if you are using any of the images in publications it is recommended to use this file type The images can of course be redrawn in GIS from imported asc raster map files curves txt file A text file containing a list of species and the relative weights used in the analysis together with the initial sum of species distributions and the level of cell removal at which point targets for particular species have been violated The initial sum of distribution is simply the sum of each species local occurrence levels through the landscape E g if your species data is in probabilities of occurrence this is the sum of probabilities in all cells before any landscape has been removed If y
9. values Note that for output the factor has been multiplied by 100 Thus using factor 1 0 would result a suffix of S100 factor 0 1 results a suffix of S10 etc IGxxx uncertainty analysis included Again the following numbers show the info gap a value multiplied by 100 BQP BQP included BLPxxxx BLP included Again the following numbers show the penalty given for the boundary length multiplied by 1 000 l e using BLP 0 5 results a suffix of BLP500 BLP 0 05 results a suffix of BLP50 etc special option relevant for probability of occurrence models using logistic link functions Determines whether the biological values of cells will be transformed from logit space value 1 for processing In this case the raster files asc files should contain the values of the linear predictor part of a logistic link function If data is not to be transformed from logit space this parameter should be set as 0 treat zero areas as missing data This option changes all cells with no species resample species occurrences to missing data In other words if a cell has a value of zero for all species it will be turned into missing data This function might be useful in some cases for example if the missing data is in fact marked with the value 0 in your species distribution files due to some technical reasons Note however that there is a fundamental biological difference between species not occurring somewhere value 0 and us not
10. 0 2 fraction of the solution in file sol rank asc Compare solutions and output to overlap ras asc Y Show overlap on map ate fractions refer to a proportion of landscape on top of intial removal Colokkey for comparison map vee 10 Doth solutions Light gr en in present solution only Dark green in older loaded solution only T Fragmentation uncertainty analys12 nfo gap the effects of fragmentation ofi 02 fraction of solution file my_soLr ank asc with buffer radius in cells 5 Info gap solution 1G Iterative cell removal done Showing removal rank L OX Maps Run settings Species info Memo Landscape identification Solution comparison About 110280 PR Miscellaneous information about the progress of analyses DC CON Sum of probs 54141 5000002611 Info gapping given solution using presently loaded data SSI spp cor Remaining columas give the fraction of biological value lost at the respective IG alpha when 87 3 50 and 0 of neighboring cells remain SPIG IG alpha Femaimmes 3 Remamingi0 Remammel min prop over spp 0 matrix x dim 649 r Lo mn Le LA LA EA Em Go co ico matrix y dim lia ka mm a E Lu Er la dG 555 Laa T a Wm oS oo E ba F E I LA tas oo ca oo Al ca co oo MS sero oe oo oS ca EA os cells with 110282 a a a a a oo call too ba m p jt CS SS a le co co missing dati 249913 J ba Go Oh dis s
11. 7452531 O 7827750 0 8427305 0 8660162 0 90 1999 0 0134111 0 0207049 0 0369054 0 0248817 0 0250449 0 0250630 0 0261239 0 0291525 0 03504 4 Picture of ouput rank asc file prop asc file A raster file similar to rank asc file Here however the matrix shows the proportional loss of distribution for that species that has lost most of its distribution during the landscape removal process E g if a cell has a value of 0 7 it means that after removing that cell all species have at least 30 of their distribution left the value 0 7 indicates that one of the species which is doing worst after removing that particular cell has lost exactly 70 of its distribution wrscr asc file In addition to the rank asc and prop asc a third map is output automatically This is the wrscr asc file where wrscr stands for Weighted Range Size Corrected Richness This map reports for each cell the quantity Wrscr gt W Qi j where wj is the weight of species j and gij is the fraction of the distribution of the species in the cell The measure is simply a sum over species of the weighted fraction of species distributions occurring in the cell as measured from original input distributions To illustrate the cell could have many occurrences of widespread low weight species In this case despite high richness per se the wrscr value would be low compared to another cell which does not contain many species but does have a s
12. Calculate new solution E C Load old solution rank file Removal rule Start of output file name Original core area Zonation ES z Use cost file Additive benefit function Use incl excl mask Target based planning Use 55 file f Generalized benefit function Use planning unit layer Use interactions file 100 0 Z 0 250 Warp factor Resampled species count Info gap settings Use info gap distribution discounting Info gap alpha 0 000 Uncertainty model f Uniform error default Proportional error List of error weight map layers load from u Cwerghts spp Run settings Species info Memo Landscape identification Solution comparison About SEE Imy_splist spp transform species occupancy probabilities from logit values output dat Annotate outputfile name eostasc masa a 00 In Pear 8 8282s Interact spp Settings for generating spatial aggregation into the solution i Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000 Use boundary quality penalty BAP definitions file name BQPcurves txt BOP mode BOP aligns missing data f Potential habitat is species specific Use directed NOP PLU connectivity load from file tree Ext Boundary Length Penalty 10 000 Additional edge points 0 Picture of Run settings window 3 2 3 Batch run capability Your third
13. Examples 118 4 3 Basic Zonation Weighted Zonation 0 0 1 02 03 04 05 06 07 08 0 9 1 proportion of landscape lost 0 01 02 03 04 05 06 07 08 0 9 1 proportion of landscape lost proportion of distributiops remaining proportion of distributions remaining Solution comparison see section 3 6 3 is a useful feature that can be used for visually comparing differences between two solutions The figure below shows the overlap and differences in the best 15 of landscape between the basic and weighted Zonation runs Overlapping regions are shown in yellow and areas that are in only one of the solutions are in green dark green no weights light green weighted Figure showing differences in top 15 fraction between the non weighted dark green and weighted light green solution Exercise 3 Species of special interest Lets assume that in addition to the seven species we work with so far there is another two species which we should protect However these species are very rare and poorly known and we only have a set of occurrence points indicating where the species have been observed Unfortunately this data is not sufficient enough to allow us to model the species distribution to cover the entire study area Yet we want to include the species into our analyses and therefore enter them as species of special interest SSI simply using the point data we have To do this we have created two text files each
14. Species info window In the Species info window you can see several panels the first of which shows summary information about the proportion of distribution remaining across species when landscape is removed The red line represents the species with the lowest proportion of distribution remaining and the blue line represents the average over all species You can select from the dialog box a respective curve for each species which will be shown in black All this information can be found in numerical form from the curves txt file 1 ZIG Iterative cell removal done Showing removal rank Maps a seltings prop distributions remaining average SA extinction risk 02 03 04 05 06 OF 08 09 1 4 07 03 b4 05 06 OF 08 proportion of landscape lost proportion of landscape lost a E o w o 3 2 r kz a o de a 40 000 60 000 60 000 100 000 cost needed to achieve given conservation value The red lines are minimum and blue average over species select species to show select a species from the list to see the loss curve black for that species species asc Picture of the Species info window Below the occurrence curve is the cost curve showing how high costs are needed for buying the respective top fraction If no cost layer is used all cells receive an equal value of 1 0 and the curves thus show the number of cells needed for respective top fractions 2004 2008 Atte Moilanen
15. connectivity calculations on large landscapes Williams J C S ReVelle and S A Levin 2004 Using mathematical optimization models to design nature reserves Frontiers in Ecology and the Environment 2 98 105 Stochastic global search Stochastic global search includes techniques such as simulated annealing SA as in MARXAN and genetic algorithms GAs Input data In principle can be run on extremely large problems with few constraints on the complexity of the problem SA can handle larger problems than a GA because of the memory requirement for storing the GA population Output A solution to the problem typically of unknown quality In some cases it may be possible to devise an analytical method that provides bounds on solution quality as in Moilanen 2005 which then changes the method from a heuristic to an approximation Heuristic method for which the quality of results is unknown approximation method for which the maximum degree of suboptimality of the results has been quantified in a non trivial manner Optimality Degree of suboptimality will be highly dependent on 1 the size of the data 2 the complexity of the problem like does it have nonlinear connectivity components in it and 3 the details of the implementation of the optimization algorithm SA and GA are no way standard algorithms except for the high level meta algorithm They can be varied in endless ways in particular in terms of how they generate th
16. distribution u The denominator is the maximum connectivity any cell has to the distribution u Thus assuming Y 1 the fraction term scales from zero for unconnected locations to one for a maximally connected location If Y lt 1 then the here negative effects of connectivity saturate with a lower level of connectivity as in Eq 1 Possible analyses with species interactions Analysis variants that one might imagine doing using the interactions facility include 1 Modeling of predator prey resource consumer and host parasitoid interactions In these analyses the objective would be to ensure protection of those parts of the resource distribution that are close enough to be utilized by the consumer This corresponds to interaction variant 1 in which the connectivity of the resource to consumer is included in analysis 2 Application of Zonation to alleviating conservation consequences of climate change In this analysis the connectivity of the predicted future distribution of the species to the present distribution is used At least three maps would be entered for each species present distribution future predicted distribution and connectivity from future to present 3 Avoidance of invading species or sources of pollution Interaction variant 2 can be used to discount a distribution in the proximity to a known or potential source of an invading species In effect occurrence levels of the target species will be reduced at l
17. is the original number of neighbors The loss term in the curly brackets is divided into two local loss and loss in the neighborhood of the focal cell Local loss is the fraction remaining of the original value of the focal cell if many of its neighbors have already been lost the value of Qi S has been reduced Loss in the neighborhood is mediated via the loss of one cell from the number of neighbors which goes down from ny to nw 1 Note that the formula above is employed as it is only for the core area Zonation For additive benefit function and target based planning the formula includes few trivial differences but the concept behind BQP is the same in all cases max ilH s e LS JE i M 0 51 N LE d i The size of the neighborhood of a cell and the effects of habitat loss are defined separately for each species according to habitat models which themselves mediate the boundary quality penalty Because BQP ranks the cells based on the responses of species to fragmentation also species that actually benefit from the loss of surrounding habitats will be equally protected compare distribution smoothing 2004 2008 Atte Moilanen Methods amp algorithms 28 2 4 3 2 4 4 Boundary Length Penalty BLP This section is mainly based on Moilanen and Wintle 2007 The most common way of adding aggregation into a reserve network has been the boundary length penalty Thi
18. j e DA r i max gt expl d le Fm 1 which is the local resource density multiplied by the connectivity of the cell to the consumer population S using parameter 6 to model the foraging distances of the consumer dis distance between cells and n Thus locations with high R have both an abundance of resource and that resource is within the foraging distance to a relatively high number of consumers Eq 1 is the connectivity of the resource to the distribution of the consumer Y 0 1 is a species specific parameter describing how fast resource use is saturated If Y 1 R scales linearly with connectivity between distributions S If for example Y 0 1 R achieves its maximum value when connectivity is 10 of the maximum it gets anywhere in the landscape and so on Concerning parameterization in the two dimensional case half of the foraging would be performed with a distance of 2 B from the focal cell Knowledge of foraging distances thus allows setting a reasonable estimate for 6 Note that the size and unit of the grid cell needs to be accounted for when calculating B see section 3 3 3 9 Essentially B is calculated identically than the scale of landscape use for the distribution smoothing technique Eq 1 is a simple variant of a connectivity computation between distributions A general variant is Rus E rif Sen fax a 2 where function f describes an arbitrary more complicated function of
19. names Remember also to use decimal points not commas in all input files Boundary quality penalty definitions file A text file where different species responses to neighbourhood habitat loss are displayed as points of penalty curves each curve on their own row E BOPcurves txt Notepad Fie Edit Format View Help 1 0000 1 0000 Q O 9949 1 0000 1 0000 s s OB 6336 1 0000 1 0000 pl O 2460 1 0000 1 0000 9147 z F4 0 6656 1 0000 1 0000 O i 1 0095 Picture of BQP definition file O 2004 2008 Atte Moilanen ZIG The Zonation software 64 3 3 3 8 Here the first column indicates the number of row After row numbers comes the first column pair in which the initial state when no neighborhood habitat has been lost is represented Here the first number of the pair indicates the proportion of neighborhood habitat left and the second column indicates the status of biological value in the focal cell Hence in the initial state when no habitat has yet been lost and therefore the biological value of the focal cell has not yet changed the two parameters always have a value of 1 000 respectively The following column pairs describe the loss of neighborhood habitats and effects that this has on the biological value in the focal cell Note that the change in biological value can be either negative lt 1 or positive gt 1 depending on species preference to fragmented
20. 10 90 34608 0 31564 i E 5431 0 049 5431 0 049 lt SSPE SADA SE SH Se Se SAAS SRE SoS Se Se SAS SAD SP EE EE EEE Running ZIG Summary also automatically produces a standard format output raster file asc file which represents the hierarchy of the cells Here the value of each cell indicates how many times that particular cell has been included in the solutions as a proportion of all solutions Remember again that this file can be imported to GIS programs When importing select floating values not integers as the type of your data Obtaining species specific information about solution quality During different analyses Zonation produces a variety of species specific information about the solution quality The purpose of this section is simply to collect all these outputs to one list and thus assist users in finding species specific information from their results All these outputs have been described in more detail in other sections of this manual and are thus listed here with only a short 2004 2008 Atte Moilanen ZIG The Zonation software 106 description see references 1 Species specific remaining curves see section 3 4 1 These curves found in the Species info window show how large proportion of species original distribution is remaining in the landscape at each step of the cell removal process 2 curves output file see section 3 4 2 This output file contains the same information as the
21. 2 4 2 boundary quality penalty where the connectivity between sites is strictly directed such as in riverine systems This option also demands the use of planning units groups of cells which are removed as a whole instead of singular cells during the landscape ranking process In freshwater planning these units would correspond to catchments The value of a focal planning unit is influenced by the removal of other planning units upstream or downstream of the focal unit Following the philosophy of BQP also here the change in local value is based on species specific responses to nearby habitat loss However the computation times are relative to the size of the planning units Smaller planning unit size means longer computation times and vice versa If you are including aggregation into your analysis it is recommended to be careful with the simultaneous use of multiple methods due to difficulties in interpreting results There however is no technical reason why smoothing the BLP and the BQP could not be used in the same analysis Distribution smoothing This section is mainly based on Moilanen ef al 2005 and Moilanen and Wintle 2006 Instructions to how to use distribution smoothing in Zonation can be found in section 3 5 2 The first aggregation method described here is called distribution smoothing When using distribution smoothing planning is based on a connectivity surface computed from the original species distributions that
22. 2004 2008 Atte Moilanen ZIG The Zonation software messages or warnings 76 O 2004 2008 Atte Moilanen 77 Zonation User manual 3 9 3 5 1 Main analyses Basic Zonation and species weighting The basic Zonation run ranks the cells in a landscape based on their conservation value The theory and algorithm behind basic Zonation is explained in sections 2 2 and 2 3 Running basic Zonation For running the basic Zonation you need to have all the compulsory input files 1 Species distribution files one file per each species 2 Species list file 3 Run settings file only when operating the program from the command prompt You can run the program either from the command prompt or from the windows interface Note that as with any other analysis you can also write a batch file and use it to run the program In the command prompt 1 Set all options of additional analyses e g BQP uncertainty analysis etc to zero in your run settings file AND in the command prompt call to indicate that no additional analyses are used 2 Adjust your settings in the run settings file for the following options if necessary e warp factor e edge removal e add edge points e logit space Note that if you do NOT select the edge removal the computation times will increase significantly with large data sets 3 Adjust the species weights in species list file if you wish to stress the conservation of ce
23. Calculate new solution Load old solution rank file Start of output file name Removal rule fe Original core area Zonation Additive benefit function Use cost file Use incl excl mask Target based planning m Use SSI file C Generalized benefit function Use planning unit layer Use interactions file Warp factor 1100 TN 0250 Resampled species count Info gap settings M Use info gap distribution discounting Info gap alpha 1 000 Uncertainty model Uniform error default C Proportional error M List of error weight map layers load from UCweights spp Output my_splist spp transform species occupancy probabilities from logit values output_ IG dat Annotate outputfile name Settings for generating spatial aggregation into the solution Y Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000 M Use boundary quality penalty B QPcurves txt BQP definitions file name BQP mode BQP aligns missing data C Potential habitat is species specific Use directed NQP PLU connectivity load from file Boundary Length Penalty 0 000 i Additional edge points The analysis produces a standard map of the landscape where reddish colors indicate sites that have both high species occurrence and high certainty Depending on the amount of error in your data the differences between the basic Zonation solution and distributio
24. Moilanen ZIG The Zonation software 70 function variants generate landscapes with many species occurring simultaneously at potentially low occurrence levels and with high overlap between species Core area Zonation produces solutions with species occurring at higher densities but with less overlap between species Which kind of solution is preferable depends on the goals of planning but it is important to realize that there may be significant differences between core area Zonation and benefit function Zonation solutions 3 Special maps By clicking the Special maps button you can ask the program to display different types of maps of your study area Note that all these maps can be saved by double clicking the picture To return to the basic map click on the Remaining button Richness The Richness map highlights areas that have the highest occurrence values in terms of summed fractions of distributions in the cell across all species in your data Here the colors indicate the sum of all normalized species distribution values in a given cell In other words the black areas represent those cells which have the highest proportional occurrence for the largest number of species Rarity This option shows you the areas that are most important for any single species in other words the core areas of a species distribution These areas might have a low biodiversity low number of species but a very high occurrence of one spec
25. PLU 9 flows into PLU 8 and PLU 8 has no downstream component like it flows into the sea PLU 7 flows into PLU 3 3 into 5 and 5 is at root of the tree PLU 1 is unlinked to anything and it is taken as an independent entity The ending of a linkage line is always marked as 1 Note that the planning unit numbers need not be consecutive Warnings will be issued to the memo if linkage information is missing for a planning unit or linkage is confused like when having multiple downstream connections for one planning unit remember that a PLU can have several upstream connections but only one downstream connection 2004 2008 Atte Moilanen 65 Zonation User manual 3 3 3 9 D My tree file txt Notepad File Edit Format View Help Picture of directed connectivity description file In addition to the directed connectivity description file loss functions i e penalty curves completely analogous to those used in the BQP technique need to be defined for each species Just like with BQP the penalty curves represent the loss of biological value in the focal unit here the planning unit when neighboring units are removed The difference to the BQP is essentially that the neighborhood is not symmetric but directional and that for each species there are separate upstream and downstream response functions modeling effects of habitat loss upstream or downstream from the focal location hence in stead of one each
26. Summary program Note again that the file name extension of a batch file has to be bat If there is anything else after the bat suffix Windows cannot identify the file as a command file 3 Open command prompt and go to the directory where you have the ZIG Summary program Type the name of your bat file without the suffix and press enter 4 Another way to run the program is to simply double click the batch file from windows 5 You can edit the batch file using notepad or the command prompt edit command Output As the program starts to run a separate ZIG Summary window appears on the screen In the Image window the program first quickly displays all those solutions which are included in the analysis and then runs the comparisons After the program has finished it displays a map where most valuable areas are shown in grey scale In other words these are the areas which in most cases are included in your solutions Note that when using the grey scale settings most valuable areas are shown as black where as with colored settings most valuable areas are shown in white This map will automatically be saved as your output jpg file but you can save it again e g with a different name or to a different directory by double clicking the picture 4 Zonation landscape identification summary Image Memo About The Image window showing an example of the ZIG Summary output map when the layout settings have been adjusted to gre
27. appropriate when i the species are 2004 2008 Atte Moilanen 21 Zonation User manual essentially surrogates or samples from a larger regional species pool and tradeoffs between species are fully allowed and ii the hierarchy of solutions and easy weighting of species is desired e The targeting formulation is most appropriate when i it is accurately known what proportion of the landscape can be had and the hierarchy is not needed ii there is a definite set of species all of which are to be protected ili occurrences are additive and iv easy weighting of species is not needed In target based planning species weighting essentially needs to be done by giving species different targets core area Z onation proportion remaining 10 08 proportion remaining 00 10 proportion remaining 0 0 02 04 0 6 0 8 10 proportion of landscape lost The figure above from Moilanen 2007 illustrates some general differences between the core area Zonation the additive benefit function formulation and the targeting benefit function Here the lines show how large proportion of species distributions is remaining in the landscape as cells are progressively removed Overall the additive benefit function has the highest average proportion over all species retained dashed line but it simultaneously has the smallest minimum proportion retained solid line because it favours species rich areas over
28. compared to other reserve selection approaches In this section we comment on the differences between Zonation and other commonly used approaches to reserve selection This comparison is not meant to be exhaustive nor completely referenced but rather to give an indication of the most fundamental differences that we believe to exist between these methods This section will also be outdated when new features are developed into other conservation planning softwares 1 3 1 Zonation Input data Zonation is targeted for use with large grid based data sets This implies that species distributions used within Zonation might be produced using some predictive statistical technique using environmental layers as predictors for species presence abundance Data sets in the order of millions of grid cells can be analyzed Output Instead of outputting the optimal set of sites for achieving targets Zonation outputs 1 the hierarchy of cell removal throughout the landscape and 2 species loss curves This kind of output has multiple advantages 1 The result for a range of targets is immediately obvious ii there is an indication of the importance of all cells both inside and outside any given fraction ili the curves show how well relatively individual species do at any given fraction of the landscape and iv the curves indicate the relative value of the solution as well as the stability of the solution v the zonation output lends itself to e
29. data elements 13032 687685064 123 600003 111 000000 5 000000 649 555 DISTRIBUTION DISCOUNTING loaded error weights map from filesp2 UL asc species asc 555 rows read in total sum of non mis ing data elements 29142 6873970031 Used Info gap alpha 0 5 spw 1 Using Info gap weight matrix ispecies asc fraction of onginal occurences remaining after DISTRIBUTION DISCOUNTING info gap 0 77640585061 9251 123 800003 111 000000 5 000000 643 555 Completed load of species file speciese asc lt Picture of the Memo window when running distribution discounting Species interactions This option allows you to include species interactions to the cell ranking process This method makes it possible to value cells not only by the presence of species within the cell but also by its connectivity to important resources such as a population of prey species or to avoidable features such as a population of a competitor species The theory and algorithm behind species interactions is explained in section 2 7 Preliminary reference for the type 1 positive interactions is Rayfield Moilanen amp Fortin 2008 Incorporating consumer resource spatial interactions in reserve design Submitted manuscript Running species interactions To include species interactions to your analysis you need to first create the interactions definition file This file tells the program which species or other input features are interacting and to what extent Also
30. from fragmentation and the degree of sub optimality of the solution is unknown 11 The core area Zonation This method has so far only been defined algorithmically not in an objective function form the CAZ cell removal rule specifies a difference equation for conservation value but not the objective directly and the degree of suboptimality of results is unknown Then again no other implementation of this method is available 1 3 2 Integer programming Input data Can accept arbitrary sites as well as grid cells According to a relatively recent review Williams et al 2004 the data size limits of IP were at that time around 10000 landscape elements which as a grid is only 100x100 elements This is orders of magnitude less than the data limits of Zonation which can run landscapes of 10 millions of elements even when using the O 2004 2008 Atte Moilanen Zonation User manual 1 3 3 Boundary Quality Penalty Output Globally optimal set of sites achieving targets No prioritization through the landscape no performance curves Optimality Guaranteed globally optimal solution to a simplified problem The value of the global optimality of results is compromised by the requirement that both the objective function and constraints need to be linear or that they can be linearized In a sense you have the optimal solution to the wrong simplified problem Not applicable at least not easily to species specific
31. generalized benefit function as your cell removal rule with core area Zonation a dummy number e g 1 0 can be used 5 This column has three functions depending on which cell removal rule is used e If you are using additive benefit function as your cell removal rule this parameter is the exponent x of the species specific power function rjx that translates representation to value The power function determines the rate of loss of conservation value from the remaining landscape as cells are removed The exponent can be any positive number but zero is not a valid value Conservation value remaining 0 0 0 2 0 4 0 6 0 6 1 0 Proportion of landscape remaining Picture of power functions with differing x values e f you are using target based planning as your cell removal rule this parameter determines the target proportion from 0 0 to 1 0 of the species distribution which you require in the final solution 2004 2008 Atte Moilanen 93 Zonation User manual 3 3 2 3 e f you are using generalized benefit function as your cell removal rule then four extra parameters are needed for each species Essentially the final numerical column the single ABF TBF parameter is split into four numerical columns which in column order correspond to variables w T x and y in the two piece power function These parameters are used to determine the shape of the function as explained in section 2 3 5 The
32. give an additional suffix to your input files asc files This is the suffix for all basic GIS raster files such as species distribution layers cost layers mask file or uncertainty map layer All these files need to be exported from a GIS software e g from ArcView or produced with some other appropriate software When exporting remember to select ASCII as the file format During the export the suffix will be automatically added after the file name Spp files A suffix for species list files such as species list file or uncertainty analysis weights file In Zonation these files would contain a list of asc files with possible parameters You need to create all spp files yourself for example with Notepad You can use the tutorial files as templates When saving a spp file remember to add the suffix spp after the file name You could also name these files spp txt to emphasize they are text files containing species lists dat files This is the suffix used in the tutorial and examples for the run settings file Also this file needs to be created by yourself You can use the tutorial files as templates These files are also technically ascii files that can be created using any text editor including notepad Note that you could call spp and dat files anything e g myspecies spp or myspecies txt but using consistently a unique extension helps you to organize and find files of the required type 3 3 2 Compulsory files All these files a
33. graph displays the minimum and mean fraction retained across all SSI species Numbers for these mean and minimum curves are given in a special SSI_curves txt output file which is produced together with the basic output files whenever SSI species are included into the analysis In this file there is also information about the level of landscape removal when the last occurrence of that SSI species is removed Locations with SSI observations can be displayed in the Maps window by selecting this option from annotations it is worth checking that the SSI locations display correctly as errors in coordinates might otherwise easily go unnoticed Planning unit layer The planning unit layer file is a standard GIS raster file asc file containing integer numbers This file includes all basic raster information as explained in species distribution map files and a matrix where the number given for a cell identifies the planning unit that the cell belongs to Planning unit numbers must be positive integers but they need not be sequential they could be like 1 2 5 12 1010 Planning units may be used to model the situation when e g land ownership dictates that O 2004 2008 Atte Moilanen ZIG The Zonation software 60 3 3 3 3 certain groups of grid cells should be treated as distinct units Or with directed connectivity the planning units could corresponds to hydrologically linked catchment areas When the planning un
34. habitats As mentioned earlier the column pairs should be considered as x y coordinates on a penalty curve To draw a penalty curve or any curve at all you need to have at least two points Thus in the BQP definition file at least two column pairs are needed The two points could for example be the initial point when no habitat has been lost and the final point when all the habitat has been lost E g for species A the two points could be 1 000 1 000 and 0 000 0 500 meaning that when all the neighborhood habitat has been lost the biological value of the focal cell for the species has decreased by half Note that each of the penalty curve rows can contain a maximum of 20 points 2 0 1 9 1 0 0 5 Biological value of focal cell 0 0 0 0 0 2 0 4 0 6 0 8 1 0 Proportion of neighbourhood lost The figure above represents an example of different species specific penalty curves redrawn from Moilanen amp Wintle 2006 Note that the curve increasing over value 1 indicates a species that prefers semi fragmented habitats Directed connectivity description file A text file that contains a description of the linkages between planning units PLU This information is entered simply as file with two columns a planning unit number as given in your planning unit layer and the number of the planning unit downstream as shown in the picture below The numbers in the picture would be interpreted so that the PLU 10 flows into PLU 9
35. have been input into zonation The calculation that is applied to each species distribution is identical to the calculation of a metapopulation dynamical connectivity measure where the connectivity value is directly proportional to the number of migrants expected at that location in the landscape Technically the computation is a two dimensional kernel smoothing using a species specific parameter width of the smoothing kernel For practical purposes distribution smoothing identifies areas that have on average high occupancy levels for species The smoothing very effectively identifies important semi continuous regions where the species has overall high levels of occurrence although not necessarily in every grid cell In contrast relatively scattered occurrences in fragmented habitat lose value As the distribution of the species in the landscape becomes smoother populations in fragmented areas end up with less value relative to continuous areas with the same average probability of occurrence Note that distribution smoothing should be used with care if the data includes a species that lives happily as a metapopulation in a fragmented habitat smoothing should be narrow for this species at least if the habitat matrix is taken as partially suitable for the species Distribution smoothing is a convenient technique to apply because it can be run as a relatively fast preprocessing step before going on to the zonation analysis itself The appropriate l
36. is a list of some basic commands that are useful when operating the command prompt help cd cd md dir dir od dir jad edit notepad Copy rename type doskey call rem exit Prints a list of most common commands and their explanations You can also view descriptions for specific commands by typing help command name Shows the name of your current directory You can move to another directory by typing cd directory name Moves one level up in the directory tree Creates a new subdirectory Type the name of your new directory after the command i e md new directory Prints a list of all files in your current directory You can narrow the list to certain type of documents by typing an asterisk and the suffix after the command E g if you are searching for a text file type dir txt and only txt files will be shown Or dir a txt txt file that start with an a Prints a list of all files and documents in chronological order Prints a list of all subdirectories in your current directory Command to edit documents Write the name and suffix of your document after the command e g edit my document txt This is a useful command for editing your list or settings files Opens a respective document with Notepad program Write the name and suffix of your document after the command e g notepad my doc txt Copies one or several files to another directory For example command copy filel txt my docs fi
37. keeps computation times short and to some extent increases the connectivity of high quality habitats in the landscape structure Hypothetically however this option can have downsides in cases where a large area of poor habitat is completely surrounded by good habitats and the Zonation program should first remove all the good habitats from the edge to reach the poor area Naturally by not selecting the edge removal option the program would easily find all poor habitats deep inside the landscape but with the cost of lost structural connectivity and increased computation times To prevent any valuable areas to be lost and to keep the computation times short it is recommended that one should use the add edge points option together with edge removal By adding logical edge cells into the landscape the program can spot any larger poor areas surrounded by good habitats without the risk of removing valuable cells use SSI Determines whether Species of Special Interest SSI are included into the analysis The distribution data of these species is not given as maps but rather as a list of single point occurrences If there are no SSI species to be included to the analysis this parameter should be set to 0 SSI file name Indicates the file that contains the names and coordinates of SSI species use planning unit layer Determines whether a planning unit layer is used value 1 or not value 0 With this option cells are grouped into defined plann
38. landscape in cells 3 Sum of species distribution proportions In other words this value shows how large proportions of species original distributions the respective management landscape covers 4 Number of species which have more than 10 of their original distribution located in the management landscape 5 Number of species which have more than 1 of their original distribution located in the management landscape 6 Number of species which have more than 0 1 of their original distribution located in the 2004 2008 Atte Moilanen ZIG The Zonation software 98 management landscape de Liout_species Ext Notepad l E File Edit Format Help Most Important species in networks those occurring at a 14 3 538443e 307wel of original distribution Network Area spp_cdistribution_sum spp occurring at gt 10 21 0 1 gt 0 01 1 328 0 000 O O 0 2 6041 0 634 3 1 f f T Species speciesl asc 9 93 of full distribution Species speciesz2 asc 15 76 of full distribution species species3 asc 13 93 of full distribution Species speciesd asc 4 57 of full distribution Species species6 asc 18 44 of full distribution 7 Number of species which have more than 0 01 of their original distribution located in the management landscape 8 Number of species which have more than 0 001 of their original distribution located in the management landscape If the five last columns are marked as zero it mean
39. layer of an interaction multiple times then the layer models simultaneous connectivity to multiple different sources After interactions have been implemented Zonation proceeds as before Connectivity methods BQP and BLP operate as before also on interaction layers O 2004 2008 Atte Moilanen Methods amp algorithms 42 2 8 Assumptions amp limitations This section lists some known assumptions and main limitations of the presently available zonation implementation e The Zonation software presently only accepts data in grids and point observation lists in particular it does not accept vector based planning units This limitation is not practical not conceptual but it is unlikely to be removed in the next Zonation version e Zonation is presently for doing an implement in one go reserve conservation plan In particular it does not include any explicit mechanism for handling multi year planning with considerations of stochastic site availability and the possibility of site loss Speculatively it may be possible that suitably defined input layers or a clever use of mask cost files could be used for doing an analysis where approximate effects of habitat loss could be accounted for This limitation most likely can be removed at least partially e At present Zonation only has a single option per selection unit that is protect or not restore or not maintain or not and so on In a more advanced analysis one could en
40. map window for viewing the uncertainty layer remember to select the Wmap option Save by double clicking on the image All uncertainty layers are listed in the UCweights spp file where you can also find the species specific uncertainty weights In this exercise we want to stress the certainty of data equally for all species thus we give them an identical weight of 1 We use the same species list file as in Exercise 4a but the settings file needs to be changed to set_uc dat Again use the do_uc bat batch file to run the program or call the program yourself In this exercise where uncertainty analysis is done together with distribution smoothing the program first calculates each cell a new value based on the uncertainties in the data and then uses these values for the aggregation part O 2004 2008 Atte Moilanen Zonation User manual 123 Batch file do_uc bat Rank 1 0 Rank 0 15 Remaining 30 Top 15 Area 28 193 Area 16 543 BL A 0 194 BL A 0 273 Cost 28 193 cells Cost 16 543 cells av prop 32 9 av prop 20 1 min prop 29 9 min prop 18 0 4 6 Exercise 6 What about the costs As in many cases in conservation biology costs play a vital part in reserve planning So far we have been looking for sites based solely on their biological value and ignoring the possible costs except in terms of land area constrains But assume that in our study area land is most expensive at
41. multiplying species specific widths of the dispersal kernel a values the second column in you species list file This parameter helps you to produce multiple solutions with different dispersal kernel sizes without needing to change the kernel widths manually after each run All kernel widths will be multiplied with this factor To run more complicated solution groups e g with several different factors varying see section 3 2 3 for batch run capability e Zero or one indicating whether the program window will be left open O will leave the window open and 1 will close the window allowing the program to move on to the next run If a batch file is written for performing multiple runs it is important to write 1 at the end of each line or the program will not proceed to the next Zonation run 4 Press enter to initiate the computation Note that it is good to familiarize oneself with the command prompt and its functions but to use the program efficiently it is much more practical to use batch files instead of rewriting the commands for each run see section 3 2 3 In this example no aggregation has been included in the analysis However it is recommended that some aggregation methods e g distribution smoothing BQP etc would be used when running the final analysis to increase the true conservation value of the reserve network See also section 3 9 for basic commands that are useful when using the command prompt Windows interface T
42. not just display the solution but removes cells from the landscape based on the ranking order of the old solution Thus it is possible to test the performance of one network with different settings using new settings to evaluate the solution You can for example run the basic Zonation and then test how well the resulted network would perform if uncertainty or boundary quality penalty would be included in the analysis Full output will be produced from the loaded analysis but cell removal order will be enforced according to the rank asc file that is loaded O 2004 2008 Atte Moilanen 103 Zonation User manual To do this you need to load your old solution with new settings see section 3 2 4 for instructions If you are running the program from the windows interface you can adjust the settings in the Run settings window Then select the Load old solution rank file option and type the name of your ranking file on the field below If you are operating the program from command prompt type Ifilename as the second parameter of your call Remember also to enter the correct name of your adjusted settings file as the third parameter in your call You can see the results for example in the species info curves in the windows interface You can also compare the curves txt files of the two solutions which reveal how large proportion of distribution of each species is remaining when landscape is iteratively removed 3 6 6 Z
43. number of cells needed to get the target distribution for an individual species then core area Zonation may require fewer cells because it prefers the higher quality density cells Thus benefit function variants generate landscapes with many species occurring simultaneously at potentially low occurrence levels and with high overlap between species Core area Zonation produces solutions with species occurring at higher densities but with less overlap between species All these differences are such that they can logically be expected to occur in any data sets with the magnitude of differences depending on the nestedness of species distributions Differences would be largest when there are both i substantial regional differences in species richness and occurrence levels combined with ii a generally low overlap between species distributions In this case core area Zonation could catch cores of species occurring in species poor areas whereas the additive benefit function would concentrate the solution more towards species rich locations where cells have high aggregate value over species Core area Zonation and presence absence data Note that when using presence absence data all cells where a species is present receive an equal value of 1 Thus in P A data there apparently are no core areas of particular importance for the species and it might seem pointless to use core area Zonation as the cell removal rule But this is not the case First o
44. occurrence density in the selected area Note that species A does not occur in all cells in the landscape It follows that if species A is widespread getting the 36 could require for example 10000 cells from the selected area whereas if species A has a limited distribution 36 could mean 2000 cells Note that the 90 50 statistic reports the averages of these numbers across all species used in the analysis Figure illustrating the example Species A original distribution is marked with green cells values indicating the probability of occurrence in each cell The red line outlines the area required to retain at least 30 of all species distributions here including 40 of species distribution The yellow cells show the area of 90 cell count for species A totaling the 36 of species A distribution What is this statistic useful for then It actually quantifies differences between cell removal rules The trend is this to get a given minimum fraction across species core area Zonation requires more cells than the benefit function variants This is because benefit function variants take occurrences as additive whereas core area Zonation prefers the locations with very highest occurrence levels However if one investigates the number of cell needed to achieve a given target for an individual species then core area Zonation may require fewer cells because it prefers the higher quality density cells Thus benefit 2004 2008 Atte
45. of a map The SSI species input can be used for a species that either i has so few observations that the distribution of the species cannot be modeled or ii has been completely surveyed and all occurrence locations are known The idea in the SSI species is that if the species only occurs in a few locations it is wasteful to enter a full map for it a one million element grid map takes 3000 times as much memory to store as does a list of 100 observation locations population sizes Consequently a very high number of SSI species can be analyzed in Zonation Ordinary map species and SSI species can be mixed in the same analysis however it is not currently possible to run Zonation only with SSI species But one can just enter one zero weighted map of the landscape and all of the rest of the species as point distributions which amount to an observed distribution only analysis SSI species are treated exactly as map species in the Zonation process the marginal loss following the removal of a cell is based on the fraction of the distribution of the species residing in the cell However there is the difference that distribution smoothing and boundary quality penalty connectivity methods do not operate on SSI species Note that connectivity requirements for a SSI species can be implemented indirectly by entering buffering locations for the SSI species around the actual occurrence locations Overall it can be expected that full distributio
46. of species distributions could be covered with 10 of the landscape See Moilanen et al 2005 The data for generating these curves and the respective species specific curves is output into a curves txt file under the name you have specified for your output O S N Oo 0 01 02 0 3 04 05 06 0 7 08 0 9 1 proportion oflandscape lost proportion of distributions remaining Figure showing how the average and minimum proportions of species distributions are declining as landscape is removed 2004 2008 Atte Moilanen 117 Zonation User manual 4 2 Exercise 2 Weighting of species Now we have gotten started However two of our target species species 2 and 3 are endemic and can not be found anywhere else in the world Thus we want to enhance the protection of these species This can be done by giving each of the species a weight of 2 The weights are determined in the species list file thus we now use a different file called splist_w spp where the weights have been changed naturally you could also use the same species list file as in Exercise 1 and just change the weights manually Also the output files need to be renamed output_w txt so that the program will not overwrite our earlier solution The settings file used here is the same as in Exercise 1 Use the do_w bat batch file to run the program or call the program yourself Just remember to rename your output file so the pr
47. planning units 2004 2008 Atte Moilanen ZIG The Zonation software 86 3 5 5 map of planning units can be displayed via the special maps window which can be accessed via the Maps window Similarly a map of river basins shows linked planning units by the same color this map is useful for visually checking that the linking of planning units is correct Make sure to repeat the display a few times especially when using grey scale color as units could accidentally end up having a similar shade making them hard to separate from each other River basins displayed on a gray color map Including distributional uncertainty Distribution discounting Distribution discounting is a method to include uncertainty analysis UCA to reserve selection process With distribution discounting you can compare the performances of different solutions as uncertainty increases Essentially this method seeks a solution that has high robustness in achieving a given conservational target despite the uncertainty The theory and algorithm behind distribution discounting is explained in section 2 5 1 This method is also described in Moilanen et al 2006 Conservation Biology 20 1688 1697 Running distribution discounting To include distribution uncertainty in your analysis you first need to define the scale of uncertainty As explained in section 2 5 1 this can be done by giving values to 1 the uncertainty parameter
48. possibility is to run the program with batch files A batch file is a simple Windows DOS command line file that can be used to give commands to Windows Batch files can be created for example with notepad This is done by writing the program call same as when using the command prompt on a new notepad document and saving it with the batch file extension bat The file name extension of a batch file has to be bat like do_zig2 bat If there is anything else after the bat suffix Windows can not identify the file as a command file It is much more practical to use batch files instead of writing the commands in the command prompt for each run There are several reasons e You will save time e The information of your analysis which input files and settings has been used will be saved in the batch file and you can review them later e With batch files you can run multiple analyses without needing to adjust the setting between the runs A batch file is run simply by either double clicking the file icon or writing the name of the batch file in the command line in the command prompt and hitting enter picture below You can edit batch files with notepad or with the command prompt edit utility Command Prompt C cd Z onation D Zonat ion do_zigz 2004 2008 Atte Moilanen 47 Zonation User manual A batch file is useful for example if you want to experiment with different levels of distribution smoothing un
49. potentially both downstream and upstream from the focal location depending on the requirements of the species 2004 2008 Atte Moilanen 29 Zonation User manual The NQP technique was originally developed for freshwater planning in riverine systems Note however that the technique could be suitable for quite different situations as well the NQP method is based on a bidirectional linking of planning units These linkages could correspond to hydrological flow But they could also correspond to other kinds of biological linkages including i other spatially continuous connecting landscape elements such as hedge rows or ii spatially discontinuous functionally linked planning units such as areas on migration routes of birds or iii they could approximate connectivity at marine areas where very strong flows generate a situation analogous to a river system In the end an appropriate aggregation of cells to planning units and suitable linkages and loss functions allow modeling of relatively variable situations Following closely Moilanen et al 2008 the present version of the NQP technique is technically specified by the following modification of the marginal loss value used in the cell removal rule __ gt local neighborhood __ local upstream downstream 0 0 0 yP down up ij jo ij ij j o J down ij ij up down up down _ 4 gt py Po jon Ty py Ty pion kj A P
50. producing a sequence of landscape structures with increasingly important biodiversity features remaining It is emphasized that the result of a Zonation analysis is not a single set of areas Rather it is 1 the nested ranking of cells and ii a set of performance curves describing the performance of the solution at the given level of cell removal _ r 00 o gt h N prop distributiogs remaining o 0 0 1 02 03 04 05 06 07 06 09 1 proportion of landscape lost Sample figures illustrating the ranking and the curves The Zonation meta algorithm can among other things answer two questions frequently encountered in conservation biology e which parts of the landscape totalling x of landscape cost or area have the highest conservation priority ranking or e which part of the landscape includes at least y of the distribution of each species proportional coverage selection Whether the Zonation algorithm makes any sense at all depends on the definition of marginal loss i step 2 in the algorithm above This definition is done by a separate cell removal rule which is described in the next chapter see 2 3 The Zonation method can thus be divided into two parts the Zonation meta algorithm and the cell removal rule definition of marginal loss which should not be confounded The cell removal rule should be seen as a separate component with several alternatives that have different interpretations Note th
51. relative error measure with exactly the same interpretation and operation as for the map species If uncertainty analysis is used distribution discounting will be applied to the population size or to any other form of information given for the location Note that the fourth column can be omitted for SSI species if so the uncertainty error measure will be taken as zero and any uncertainty analysis will not influence population sizes given in column number three A _ coord_ species1 txt Notepad Ale Edit Format View Help 294220 0 6283664 6 294220 2 6283664 294220 0 6283664 294220 2 6283664 Picture of the species specific coordinate file The first two columns give the x and y coordinates of the record The third column shows the biological value of that record any non negative integer or decimal value and the last column is the relative error measure When including SSI species into your analysis remember to type in to your Run settings file use SSI 1 SSI option selected and SSI file name my SSI list txt name of your SSI species list file If you are running the program from windows version go to Run settings window select the Use SSI option and type the correct path to your SSI species list file if the file is in the same directory with Zonation program only the name of the file is required Output with SSI species Output for SSI species is given in the Species info window where a
52. species has two penalty curves Note that the use of NQP also changes the interpretation of the species list file With the NQP there is no species specific radius like with BQP the neighborhood is the set of linked planning units Rather instead of as number of response and radius the third and fourth columns in the species list file are interpreted as the row number of the penalty curves for upstream and downstream losses respectively The curves are specified in the same input file as in BQP see section 3 3 3 7 A map of planning units can be displayed via the special maps window which can be accessed via the Maps window Similarly a map of river basins shows linked planning units by the same color this map is useful for visually checking that the linking of planning units is correct Make sure to repeat the display a few times especially when using grey scale color as units could accidentally end up having a similar shade making them hard to separate from each other Species interactions definition file A text file which defines what interactions between species or other input features are to be included in the analysis This file is only needed if you want to include species interactions into your analysis The structure of the interaction definition file is as follows BP Sp_interaction txt Notepad File Edit Format View Help Sl focal species S2 connectivity to beta type gamma 0 008 1 0 1
53. species specific remaining curves in the Species info window However it also tells you the initial sum of species distributions together with the relative weights used in the analysis and the level of cell removal at which point targets for particular species have been violated 3 Histograms of habitat quality see section 3 4 1 Also in the Species info window is the histogram of habitat quality for a selected species in a given top fraction of the landscape The respective information is also output into the Memo The x axis of the histogram is linearly scaled from O to 1 where 1 represents the highest quality found in the original input map for the species 4 Landscape identification see section 3 6 2 Landscape identification produces an output file where you can information about species occurrences in each management landscape The file tells you which species are present in each management landscape but also how large proportion of species original distribution is covered by the landscape or by all the management landscapes as a whole O 2004 2008 Atte Moilanen 107 Zonation User manual 3 7 Data limitations amp system requirements Hardcoded limitations which may be alleviated in later Zonation versions All analyses e Maximum number of species any biodiversity features 4 000 e Maximum SSI spp 4 000 thus total maximum is 4000 4000 Distribution smoothing e Maximum size of rasters 16 mil
54. the reserve network When running the analysis the program goes through all cells and calculates them a value 0 based on that species that has the highest proportion of distribution remaining in the specific cell and thus represents the highest biological value to be lost if the cell is removed The cell which has the lowest 0 value will be removed The critical part of the equation is Q S the proportion of the remaining distribution of species located in cell for a given set of sites the set of cells remaining S When a part of the distribution O 2004 2008 Atte Moilanen Methods amp algorithms 18 2 3 2 of a species is removed the proportion located in each remaining cell goes up This means that Zonation tries to retain core areas of all species until the end of cell removal even if the species is initially widespread and common Thus at first only cells with occurrences of common species are removed Gradually the initially common species become more rare and cells with increasingly rare species occurrences start disappearing The last site to remain in the landscape is the cell with the highest weighted richness This is the site that would be kept last if all else was to be lost Note that Eq 1a can alternatively be expressed as Moilanen et al 2005 Jani an 1b where q is the fraction of the original full distribution of species j residing in cell according to data and Q S
55. the settings file set_bqp dat Because BQP substantially O 2004 2008 Atte Moilanen 121 Zonation User manual increases computation time the warp factor has been increased to 500 in production runs we recommend warp factor of 100 or lower We also need to use a new species list file which gives the program all species specific responses to fragmentation needed for running BQP Thus use the splist_bqp spp file as species list file You can use the do_bqp bat file to run the analysis or call the program yourself Batch file do_bqp bat Rank 1 0 Rank 0 15 Remaining 30 Top 15 Area 30 875 Area 16 543 BL A 0 563 BL A 0 706 Cost 30 875 cells Cost 16 543 cells av prop 37 8 av prop 21 6 min prop 29 9 min prop 15 4 As you see the two solution are fairly different To get a better picture of the main differences use the solution comparison You can either use the top 15 fraction of the two solutions for the comparison see figure below or you can check from the results how large fractions were needed to protect the 30 of all species distributions and use these fractions for the comparison Comparison between top 15 areas selected by Zonation with distribution smoothing and Zonation using the BQP Overlapping areas are shown in yellow light green areas are present only in distribution smoothing solution and dark green areas only in BQP solution O 2004 2008 Atte Mo
56. they have been modified To return to summary maps just click any of the map buttons Rank Remaining or Special maps 5 Annotations For printing your maps you can change the settings of the map layout in annotations by clicking the A button You can adjust the colors of your maps add scale line O 2004 2008 Atte Moilanen 71 Zonation User manual and panel ID or just outline the landscape to distinguish it from the white background when using gray scale After adjusting the layout settings select again any of the map options Specie specific maps Rank Remaining Special maps and your new annotations will appear in the map window Note that the point locations of SSI species Species of Special Interest can be displayed by selecting the Show SSI location on top option The locations will then appear as green dots on the map To save any of the images just double click on the pictures and a saving window appears 6 Wmap If you are including uncertainty in species distribution in to your analysis this option shows you how the uncertainty of your data is distributed in the landscape Activate the Wmap option and select a species from the dialog box 4 to display the species specific uncertainty maps Here the black color indicates high uncertainty in the data whereas white color indicates high certainty Note that these maps can only be displayed when uncertainty layers have been included to the analysis
57. those areas that might be significant for the existence of one species but that otherwise are species poor Core area Zonation has a high minimum proportion combined with a relatively low average because it retains the most significant areas of species the core areas till the end even thought these areas might be unsuitable for all the other species The targeting benefit function does well in terms of finding the highest level of cell removal without having any species specific targets violated However 2004 2008 Atte Moilanen Methods amp algorithms 22 2 3 5 when further away from the target it does relatively poorly in terms of the minimum fraction over species retained The problem with the targeting benefit function is that it is aimed at good performance at one particular set of targets but the hierarchy of solutions is missing in the sense that good overall performance at other levels of cell removal especially at a level where targets have been violated cannot be guaranteed There are also differences between the cell removal rules on how much area they require for achieving a set conservation target To get a given minimum fraction across species core area Zonation requires more cells than the benefit function variants This is because benefit function variants take occurrences as additive whereas core area Zonation prefers the locations with very highest occurrence levels However if one investigates the
58. to Run settings window select the Use incl excl mask option and type the correct path to your removal mask layer file if the file is in the same directory with Zonation program only the name of the file is required Note that use of removal mask layer is likely to result a biologically non optimal solution as the program is not allowed to remove cells only based on their conservation value See r eplacement cost analysis section 2 6 for the analysis of the suboptimality of masked solutions 3 3 3 5 Distributional uncertainty map layer A standard GIS raster file asc file of uncertainties in species occurrence These files are needed If you are including the uncertainty in species distributions in to your analysis where you need one uncertainty map layer for each species used in the analysis The file includes all basic raster information as explained in species distribution map files and a matrix of species occurrence uncertainties in each cell parameter wsc in the info gap uncertainty model The species cell specific uncertainty value wsc can be any measure of error in prediction or any uncertainty about whether the species will persist there or a combination of those as long as the data of all species is in same format For example if confidence intervals are available for the probabilities of occurrence of species in a given cell the uncertainty value for species A in that same cell can be the size of the lower half of t
59. to operate as the skeleton of the corridor Because the skeleton is masked in and the BQP is used it becomes advantageous to expand areas around the skeleton Whether the solution is good will very much depend on the choice of skeleton areas so they should be chosen well 2004 2008 Atte Moilanen Introduction 12 Outlook for the next version e V 2 1 main feature The main feature of v 2 1 is going to be ability to use Zonation on community modeling data which utilize data models about species richness and community similarity dissimilarity Essentially this builds an ability for the use of environmental based surrogates in Zonation Look out for the new feature which can be used in combination with species based analyses O 2004 2008 Atte Moilanen 13 Zonation User manual 2004 2008 Atte Moilanen 15 Zonation User manual 2 1 Methods amp algorithms References The basic Zonation analysis and distribution smoothing Moilanen A Franco A M A Early R Fox R Wintle B and C D Thomas 2005 Prioritising multiple use landscapes for conservation methods for large multi species planning problems Proceedings of the Royal Society of London Series B Biological Sciences 272 1885 1891 Moilanen A 2007 Landscape zonation benefit functions and target based planning Unifying reserve selection strategies Biological Conservation 134 571 579
60. we use the original species list file splist spp To execute the analysis you can use the batch file do_nqp bat 0 0 1 02 03 04 05 06 07 08 09 1 proportion of landscape lost proportion of distriQutiogs remaining As you can see the use of planning units changes the results quite radically Also species performance curves change from smooth to staggered lines and in general all species are performing notably worse as compared to e g the basic analysis in Exercise 1 This is due to the fact that the planning units used here are relatively large and as the program removes units from the landscape the risk of loosing valuable areas inside the units is very high Therefore it is important to understand the use of large planning units will automatically cause a decrease in the quality of results and thus the sizes of planning units should be carefully selected 2004 2008 Atte Moilanen Index 128 Index A adding edge points 53 additive 20 additive benefit function 17 18 aggregation methods 24 25 28 79 81 83 aim amp purpose 2 algorithm 16 alpha value 50 annotations 53 67 approximation definition 5 assumptions 42 107 B base analysis 6 basic core area Zonation 17 batch file 8 batch files 44 46 batch runs 46 benefit function 22 50 BLP 24 28 boundary length penalty 24 28 bo
61. working with your own data you can use the example files to create your own input files The installation package includes the following example files For running basic Zonation species1 asc species2 asc species3 asc species4 asc species5 asc species6 asc species7 asc splist spp splist_abf spp splist_tbp spp cost asc set dat set_abf dat set_tbp dat set_cost dat do zig2 bat do_abf bat do_tbp bat do_load bat For weighting species splist_w spp do w bat For using SSI species SSI_sp8 txt SSI_sp9 txt SSI_list txt set_ssi dat do ssi bat For including distribution smoothing do ds bat For including BQP BQPcurves txt splist_bqp spp set_bqp dat do bqp bat For including BLP set_blp dat do blp bat For including uncertainty analysis sp1_UC asc sp2 UC asc sp3 UC asc sp4 UC asc sp5 UC asc sp6 UC asc sp7_UC asc UCweights spp 2004 2008 Atte Moilanen Tutorial amp Examples 114 set_uc dat do uc bat For running the replacement cost analysis mask_rs asc mask_towns asc set_maski dat set_maske dat do rs bat do towns bat For including directed connectivity NQP plu asc rivers txt NQPcurves txt splist_nqp spp set_nqp dat do _nqp bat Other files needed for tutorial and exercises set_costds dat set_costbqp dat load_costds bat load_costbqp bat do cost ds bat As a final word this tutorial does not include all variants of everything When working with your own data remember
62. 0 0 0 1 0 1 41 Vi 0 0 0 1 0 1 Proportion of landscape remaining A i Linear 1 0 1 0 NA dummy 1 0 Il 1 0 lt 1or gt 1 INA dummy 1 0 1 same order at inclination gt 1 lt as W point e g 1 x w Power function ABF W ae w 111 Mild sigmoid O 2004 2008 Atte Moilanen Methods amp algorithms 24 2 4 iv Steep sigmoid step w same order at step gt gt lt lt 1 imitation as W e g 1 x Ramp at step NA A 1 0 Ramp with linear at step a The parameter definitions are suggestive and the exact shape of the function can easiest be determined by plotting it To use generalized benefit function as a cell removal rule the parameters of the function need to be given in the species list file Inducing reserve network aggregation Fragmentation is an undesirable characteristic in reserve design as it has been concluded in many studies that species persist poorly in small and isolated patches Also implementing a fragmented reserve network may be awkward and expensive In this section we introduce three different aggregation methods that can be used when running analyses with the Zonation program These methods produce relatively more compact solutions Note however that aggregation always involves trade offs There is usually an apparent biological cost in more aggregated solutions because in many cases it is necessary to include lower quality habitats into the reserv
63. 00 0 0000 0001 0 0000 0004 0 0025 CODO 0001 0005 O O O O a O O O O O O O O 0 0000 0 0000 0 0000 0 O O O 0 O O 2004 2008 Atte Moilanen 99 Zonation User manual 3 6 3 Note that setting the maximum distance between cells to zero allows you to view the statistics of every single spatially distinct patch in the landscape but also increases the running time A larger maximum distance leads to fewer management landscapes Note that the program presently only allows identification of up to 30 000 landscapes Solution comparison Solution comparison calculates how much two solutions overlap with each other and what is the average difference in the cell removal order The comparison is always made between the present solution and an older solution by using the rank asc files of both solutions as input files Running solution comparison The solution comparison can only be done from the windows interface 1 First go to the Solution comparison window and select how large proportion of each landscape is compared Compare fraction of present solution to fraction 2 Next define the name of the rank file to which the current solution is compared Thus you must have a solution computed or loaded before you can use the solution comparison utility 3 Finally give a name to your output file e g overlap1 ras asc and press the Compare solutions button 3 PG lheretive cell rem
64. 04 2008 Atte Moilanen 79 Zonation User manual 3 5 2 T iu i Be N k it i del EN E La Li vu 1 El o Ae TE e a T a Tm o 7 mE s ass dr J LE e Y Picture of solution when one of the species has a weight of 3 0 Distribution smoothing Zonation Distribution smoothing is a two dimensional kernel smoothing where the width of the smoothing kernel is determined by the estimated dispersal ability or scale of landscape use of the species in question This option results a much more compact solution where small isolated patches have been removed Using distribution smoothing increases computation times marginally if at all The theory and algorithm behind distribution smoothing is explained in section 2 4 1 This method is also described in Moilanen et al 2005 and Moilanen amp Wintle 2006 Running distribution smoothing To get distribution smoothing done you need to In the command prompt 1 Set the third last parameter of your call to 1 to indicate that distribution smoothing will be done 2 Give a factor for multiplying the species specific a values as the second last parameter in your call a values species specific widths of kernel are in the second column of your species list file The factor is useful if you are interested in running multiple solutions with e g assuming several levels of dispersal capabilities because the factor allows you to multiply all dispersal capabilities simultane
65. 1 specifies that if nothing has been lost from the neighborhood of a cell then the nominal predictions should be used as there is no uncertainty caused by nearby habitat loss In the Zonation program a simple relationship f aLc exp aLc is used but any other decreasing function would give similar results With this relationship if a 0 no uncertainty or if there is no habitat loss Lc 0 then from 1 it follows that p se P sc Increasing the uncertainty parameter a or the fraction of habitat lost will result in a potential loss of conservation value in the target cell Note that it is not claimed that any correct value of a is known Rather the uncertainty analysis will proceed to analyse how robust different solutions reserve network candidates are to increasing uncertainty The second component needed for the uncertainty analysis is a measure of the performance of a candidate reserve network as uncertainty is increased The performance of species s in candidate reserve network X Vs X p was defined as the proportion of the original full distribution of the species remaining in the given reserve structure X gt xp SE VAX p 2 where p sc is the original value for species s in the cell c and psc is the final value after the loss of neighbourhood habitat has been taken into account xc receives values depending whether the cell in question is included to the candidate reserve network xc 1 or not xc 0 Just like in distributio
66. 2 sp1 and outS3_sp1 Then the program repeats the procedure with the other two species list files species2 spp and species3 spp Thus running the batch O 2004 2008 Atte Moilanen ZIG The Zonation software 48 produces nine solutions with different settings and species weights composition Using nested batch files is extremely useful when running many solutions using combinations of settings For example if we would use only one batch file to run the solutions described above we would need to write a separate call for each of the nine solutions and with many calls it is relatively to introduce errors into some of them E do_zig2_all bat txt Notepad File Edit Format View Help zig2 r settingsl dat speciesi outsl_spl Zig2 r settingsl dat species 2 outsl_sp2 Zig2 r settingsl dat species outsi sp3 settings2 dat speciesl s outs2_spl 7 Settings2 dat species2 outs _sp2 SeTtings2 dat specles3 QUtS2_5p3 settings3 dat speciesl spp outs3_spl settings3 dat speclesz outs3_sp2 settings3 dat species3 s outs3i_sp3 bi ai a Re aa ocococooo 000000000 A ok oe oeooccoeceocse ppp pa A lp ps 3 2 4 Loading previously calculated Zonation solutions It is also possible to load previously calculated Zonation solutions This is a useful utility if you want to make some further analysis with your old solution but also if you need to test the performance of your old solution in different circumstances s
67. 7 planning units Spot count 798 CE Spots including best areas count 798 Found networks count 31 Cells in classified landsca 1 000 0 261 species asc 1 000 0 281 species asc 000 0 250 species asc ee ee n nets DE A DO PRESA as 1157 LC ris 3 l AU 1 334 Species asc y i Ani Aaaa E ew 3 3 LVV Vathi PECES dat Aan AS U UE Litt ti G EI LI 41 species asc In addition the landscape identification produces a basic raster output file ras asc file Here the matrix indicates which cells belong to which management landscapes Each landscape has an integer starting from number one If a cell has a value of 2 it means that the respective cells has not been included in the given top fraction see Percentage of landscape above Remember that this file as any of the ASCII files produced with Zonation can be imported to GIS programs However when importing this file select integers as the format of you cell values Statistics for management landscapes The landscape identification procedure also produces a text output file network species data containing statistical information of the management landscapes With the file you can receive a set of information about the different management landscapes in your data The first part of the file contains statistics about species occurrences in each landscape This part is divided to eight columns 1 Number of the management landscape 2 Area of the management
68. Generalized benefit function M Use interactions file moo z 0 250 Warp factor Resampled species count Info gap settings Use info gap distribution discounting 0 000 Info gap alpha Uncertainty model Uniform error default C Proportional error List of error weight map layers load from ju Cweights spp UCweights spp Run settings Species info Memo Landscape identification Solution comparison About MISE my_splist spp transform species occupancy probabilities from logit values output_BQP dat Annotate outputfile name Settings for generating spatial aggregation into the lution N Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000 VW Use boundary quality penalty BQP definitions file name BQPcurves txt BQP mode BQP aligns missing data C Potential habitat is species specific Use directed NQP PLU connectivity load from file 0 000 Boundary Length Penalty Additional edge points O 2004 2008 Atte Moilanen 83 Zonation User manual 3 5 4 1 Select the BQP option from the Run settings window 2 Define the name of your BQP definition file 3 Select the correct BQP mode based on whether the data no data matrix of your species distribution layers is uniform in all layers or if there are differences between the layers Output Including BQP into your analyses
69. IG The Zonation software 44 3 1 3 2 3 2 1 ZIG The Zonation software Introduction In the following sections we describe how the Zonation software is used what kind input files it requires what kind of output it produces and which are the analyses that can be done with the program If a phrase is underlined in the text it means that it is just mentioned here and explained in more detail in another section Just click the word if reading the manual pdf or use the index to find these sections For concepts and analyses implemented the software see section 2 Running Zonation When running the Zonation program you have three options You can operate the program by using either the Command prompt the Windows interface or batch files The Windows interface can be more easily approachable when familiarizing oneself with the program but when running several more advanced analyses it is highly recommended to operate the program from command prompt together with batch files These options are explained in the following sections Note that with Zonation you can also either i make new analyses or ii load old solutions The solution loading option can be used either to review old solutions or to investigate how old solutions would perform under new different assumptions cross compare After running Zonation please check the Memo or the respective run_info txt file for verification about data characteristics and op
70. IG Sum Utility This analysis is also described in Moilanen et al 2006 Conservation Biology 20 1688 1697 Essentially the ZIG_ Sum utility does an analysis of selection frequency As concluded so far you might get varying solutions for your data by using different settings e g weights a values or additional utilities BQP UCA With ZIG Summary you can compare these several solutions and find out which areas in the landscape are most often included in the final solutions and thus have the highest conservation value regardless of the analysis settings Input files ZIG Summary uses raster files that have been produced with the Landscape identification utility in the Zonation Note that all raster files should include the same proportion of the original landscape otherwise the analysis may become difficult to interpret Thus when producing the landscape identification rasters the value in the Percentage of landscape field should be equal in all runs Also the Inclusion minimum percentage needs to equal with the Percentage of landscape value This is to ensure that all areas in the given fraction are identified in the solution LI_out ras asc Notepad File Edit Format View Help ncols 649 nrows 555 xllcorner 294205 0000 6283604 5000 O 20000000 A lalala LL ell dl Le A 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i In addition to the rasters you need a txt fi
71. The fourth column represents the average proportion over all species The fifth column shows the average extinction risk of species as landscape is iteratively removed as calculated from the species area relation using parameter z and the sixth column is the weighted extinction risk where species area extinction risk has been weighted by the species weights The following columns show the proportion of distribution remaining for each species in the same order as the species are listed in the beginning of the file Note that for the output file to be readable the program does not print every step of cell removal this file only includes a maximum of 10 000 rows rank asc file A raster file representing the ranking of the landscape or in other words the order of cell removal The file includes all basic raster information as explained in species distribution map files and a 2004 2008 Atte Moilanen 75 Zonation User manual matrix of cell removal order Here the cells receive a value between 0 and 1 Low values close to zero indicate that the cell has been removed in a early state of the process whereas cells with high value are removed last output rank asc Notepad Em lx File Edit Format View Help 649 294205 0000 6283604 5000 O 200 NODATA_ value 999 999 999 999 999 999 999 999 999 999 999 999 999 999 999 999 999 999 999 9 999 999 999 0 8105856 0 7911264 0 7811066 0
72. Use boundary quality penalty BQPcurves txt BQP definitions file name BQP mode BQP aligns missing data C Potential habitat is species specific Use directed NQP PLU connectivity load from file 0 000 a Boundary Length Penalty Additional edge points solution with a mask file Also the distribution curves in Species info window may show some changes depending on the areas that have been included excluded Note that in many cases the use of mask files results a suboptimal solution In other words the program cannot select the best possible solution because it is forced to either exclude biologically valuable areas from or include poor quality areas into the 2004 2008 Atte Moilanen ZIG The Zonation software 94 top fraction By comparing the curves txt files that are produced after each run it is possible to evaluate the costs induced by the usage of inclusion exclusion masks see also section 2 6 prop distributions remaining 0 01 02 03 04 05 06 0 7 08 09 1 proportion of landscape lost Picture of distribution curves when the program has been forced to include low quality areas to the solution above Note the changes at the end of the curves clearly demonstrating that the forcibly included areas were not what one would have ideally chosen O 2004 2008 Atte Moilanen 95 Zonation User manual 3 6 3 6 1 Post processing analyses amp options T
73. Using the BQP decreases the biological quality of a land unit grid cell that is located close to the edge of the reserve which results in a more highly aggregated optimal reserve structure The theory and algorithm behind boundary quality penalty is explained in section 2 4 2 This method has also been described in Moilanen and Wintle 2007 Running BQP To include BQP in to your analysis you need to 1 Create a BQP definition file which contains all penalty curves This file determines different responses of species to habitat fragmentation 2 Link all species to the correct penalty curve by entering the correct row number of the respective curve into the third column of your species list file Multiple species can and commonly will link to the same response curve 3 Give a suitable buffer size in cells for each species in your species list file The buffer size indicates the area in which any habitat loss and fragmentation will influence the biological value of the focal cell for that particular species In addition you need to adjust your settings In the command prompt 1 Set use boundary quality penalty to 1 in your run settings file to indicate that BQP will be used 2 Also define the name of your BQP definition file in the run settings file 3 Select the correct BQP mode based on whether the base habitat requirement is the same for all features or whether there are different species groups with broadly different requir
74. ZIG software has been done with the best of intentions it is quite beyond one researcher to ensure its correct operation under all operating systems and environments Also anticipating and checking for all combinations of erroneous input has not been possible Therefore use the software with care and make an effort to understand the output and make a reality check as to whether the results make logical sense Publisher Atte Moilanen Metapopulation Research Group Managing Editor Atte Moilanen Technical Editors Heini Kujala Anni Arponen Cover Designer Heini Kujala Atte Moilanen Cover Photo Heini Kujala Evgeniy Meyke Special thanks to Anni Arponen Alison Cameron Aldina Franco Ascelin Gordon John Leathwick Grzegorz Mikusinski Chris Thomas and Brendan Wintle are thanked for comments on the software and its functions and usability Special thanks to Anni Arponen who helped with an early version of Zonation documentation and to Mar Cabeza who commented the first version of the manual Brendan Wintle generously provided sample data files from the Hunter Valley region to be included in the tutorial and Evgenyi Meyke kindly provided beautiful background photographs for the Zonation documentation Thanks to all who have collaborated in the development of Zonation methods and applications This work has been supported by the Academy of Finland project 1206883 to the author and two Academy of Finland Centre of Excellenc
75. Zonation Use cost file Additive benefit function Use incl exc mask Use SSI fil M Use ph C Target based planning C Generalized benefit function ning unit layer VF Use interactions file Warp factor moo 7 2 0250 Resampled species count Info gap settings Use info gap distribution discounting Info gap alpha 1 000 Uncertainty model Uniform error default C Proportional error List of error weight map layers load from UCweights spp MISES About my_splist spp Z transform species occupancy probabilities from logit values output_ Sl dat Annotate outputfile name cost asc mask ras asc SSI ist txt PLU asc interact spp Settings for generating spatial aggregation into the solution M Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000 Use boundary quality penalty BQPcurves txt BQP definitions file name BQP mode BQP aligns missing data C Potential habitat is species specific Use directed NQP PLU connectivity load from file tree txt 0 000 ma Boundary Length Penalty Additional edge points 1 Select the species interactions option from the Run settings window 2 Define the name of your species interaction definition file Output As usual Zonation produces a map of cell ranking and all the visual and file output as with other analyses In addition the program prints
76. Zonation V2 4 the generalized benefit function which is a two piece power function that can assume very versatile forms allowing flexibility in the specification of conservation value Note that core area Zonation has the property that it can identify important otherwise species poor locations where a single species has an important occurrence The additive benefit function analysis gives more weight to locations with high species richness Therefore it may be useful to run both analyses and compare results If the top fractions do not agree then there are some species rich areas but also some species poor areas with occurrences of otherwise rare species Thus running both core area Zonation and the additive benefit function analysis may reveal information that is interesting for conservation planning Basic core area Zonation This section is mainly based on Moilanen et al 2005 and Moilanen 2007 In basic core area Zonation cell removal is done in a manner that minimizes biological loss by picking cell that has the smallest value for the most valuable occurrence over all species in the cell In other words the cell gets high value if even one species has a relatively important occurrence there The removal is done by calculating a removal index 0 minimum marginal loss of biological value for each of the cells where en S M i Ja where w is the weight or priority of species j and c is the cost of adding cell to
77. a and 2 to the species and cell specific relative error measure Wsc Species cell specific relative errors are the ones given in the uncertainty map layers Using species cell specific errors i e the uncertainty map layers is optional but you always have to give a value to the uncertainty parameter Note that if you are using a as the only measure of error thus not using the distributional uncertainty map layers it is important that the value of a is determined in relation to your data For example if your species data are probabilities of occurrence 0 1 the uncertainty parameter should be set to a reasonably small scale e g a lt 0 4 to avoid complications which may arise if all cells receive an effective discounted value of zero leading to full loss of information from the distribution of the species 1 Distribution uncertainty map layers for each species These layers show the relative magnitude of error uncertainty of species occurrence in each cell Remember that the meaning of a must be interpreted with respect to the error measure you use For example if your error measure is the standard error of statistical prediction then a 1 essentially means subtracting one SD from the value of each cell 2 Uncertainty analysis weights file containing a list of each distribution uncertainty layers and species specific error weights With error weights you can stress the data accuracy for certain e g very rare species If no speci
78. ade offs between biological quality and the certainty of that information Ideally one would identify a reserve network that guarantees high biological quality despite some uncertainty in input data Uncertainty analysis could also be used for evaluating the opportunities arising from uncertainty that is potential for positive surprises Here we introduce two methods of uncertainty analysis that can be used in the Zonation context The first method is called distribution discounting which enables the ranking of the landscape using species distribution data that includes uncertainties Our second method is for testing reserve network structures to see how robust they are against uncertain negative influences of habitat loss and fragmentation For more information about the aims and methods of uncertainty analyses in reserve selection see Moilanen A Runge M C Elith J Tyre A Carmel Y Fegraus E Wintle B Burgman M and Y Ben Haim 2006 Planning for robust reserve networks using uncertainty analysis Ecological Modelling 199 115 124 Moilanen A and Wintle B A 2006 Uncertainty analysis favours selection of spatially aggregated reserve structures Biological Conservation 129 427 434 Moilanen A Wintle B A Elith J and Burgman M 2006 Uncertainty analysis for regional scale reserve selection Conservation Biology 20 1688 1697 Uncertainty in species distributions distribution discounting This section is
79. al estimates thus emphasizing locations with relatively certain predictions O 2004 2008 Atte Moilanen 33 Zonation User manual Distribution model Error surface Discounted distribution Picture demonstrating the concept of distribution discounting Here the first picture shows a modeled map of species distribution white areas representing a high probability of occurrence The second picture displays an error surface eg standard deviation of the modeled values again white color indicating large deviation and therefore high uncertainty The a value horizon of uncertainty can be used to either enhance or diminish the strength of the error surface e g a 2 would double all error values in the map The third picture is species discounted distribution where the error surface has been subtracted from the modeled distribution map This is the map that Zonation finally uses to run the analysis An expanded explanation for distribution discounting In more detail The occurrence of species s in a cell c here indicated as p sc is by no means certain but merely the nominal best guess probability Thus the true probability psc 0 1 could be within an interval given by P E aws 4 pse 4 P ct cows 1 where a is the horizon of uncertainty and ws is any error measure related to the accuracy of p sc for species s in cell c Thus the true probability psc could be either higher or lower than the estimate p s
80. ameter indicates the row number in BQP definitions file linking the species to a correct penalty curve When using NQP this parameter serves the same purpose but only for upstream connectivity Thus in NQP this number links the species to a penalty curve that describes how the value of focal planning unit changes when other planning units are lost upstream from the focal planning unit O 2004 2008 Atte Moilanen ZIG The Zonation software 52 4 When using BQP this parameter gives the species specific buffer size number of cells The buffer size indicates the area around the focal cell in which any fragmentation removal of cells influences the quality of the focal cell For species with large home ranges the buffer size should be larger and for species with small home ranges a smaller buffer size is adequate buffer size 3 buffer size 5 When using NQP this parameter indicates the row number in BQP definitions file linking the species to a correct penalty curve this time for downstream connectivity Thus when running NQP every species have two penalty curves one for upstream and another for downstream As NQP option uses planning units instead of singular cells no buffers are needed The connectivity of separate planning units i e which ones are upstream or downstream is defined in the directed connectivity description file Column 5 is used when using additive benefit function target based planning or
81. ampled species count Info gap settings Use info gap distribution discounting Info gap alpha 0 000 Uncertainty model f Uniform error default Proportional error List of error weight map layers load from u Cwerghts spp Using the command prompt Seles Imy_splist spp transform species occupancy probabilities from logit values output dat Annotate outputfile name Interact spp Settings for generating spatial aggregation into the solution i Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000 Use boundary quality penalty BQPcurves txt BOP mode BOP aligns missing data BAP definitions file name C Potential habitat is species specific Use directed NOP PLU connectivity load from file tree Ext 0 000 m Boundary Length Penalty Additional edge points 1 Torun the program from command prompt in addition to species distribution map files and species list file you need a third input file called run settings file This will define the setting of your analysis 2 Select the suitable cell removal rule in your run settings file 3 Open the command prompt from your Windows Start menu 4 Use cd directory name command to change to the correct working directory which contains the zig2 exe and all the input and settings files See section 3 9 for working with the command prompt 5 Call the program with the follo
82. asy visualization We will elaborate the item iv If species performances are declining rapidly at the chosen landscape fraction it means that the solution is not stable with respect to uncertainty in input data and that smallish changes in the selected fraction and or spatial pattern might have large consequences for species lf the species performances are stable at the chosen fraction then small changes in the fraction spatial pattern are unlikely to have any significant effect on the solution quality Additionally the core area Zonation method has a specific feature in that it emphasizes best areas for all species instead of treating low to medium quality locations as additive Optimality The optimality characteristics of Zonation have not been conclusively examined but this is our present evaluation of this issue 1 Zonation using additive benefit functions or the targeting benefit function above the target is very close to globally optimal This is because with these cell removal rules the optimization problem is convex and can thus be solved using a gradient like iterative heuristic van Teefelen and Moilanen 2008 Also with the additive cell removal rules the degree of suboptimality goes down when the landscape size number of cells increases Thus optimality is not a problem with the additive cell removal rules Except use of the Boundary Quality Penalty BQP renders the problem non convex especially if some species benefit
83. at the notion of complementarity is inherent in the way the cell removal rule is defined There is one feature which according to Moilanen et al 2005 is a part of the Zonation algorithm but which is more appropriately seen as a relevant detail for which there are alternatives This is O 2004 2008 Atte Moilanen 17 Zonation User manual 2 3 2 3 1 edge removal by which it is meant that cells can only be removed from the edge of the remaining landscape Edge removal may promote maintenance of structural habitat continuity in the removal process lt also makes the cell removal process much faster with large landscapes which is the primary reason for using it The cell removal rule This section is mainly based on Moilanen 2007 The Zonation meta algorithm is the same for all analyses described in this manual The rule that determines the loss of which cell leads to smallest marginal loss and is therefore removed next differs depending on the cell removal rule that is chosen There are three conceptually different cell removal rules 1 Core area Zonation 2 Additive benefit function 3 Target based planning In the following sections we first describe these three main cell removal rules and the theory behind them and then highlight the differences between the rules and give some guidelines how to choose the most suitable one for your analysis Finally we describe the fourth cell removal rule added in
84. atio Cost This will tell you how much the selected area costs if you are using a cost layer If no cost layer is used all cells are given a equal value of 1 thus this will show the number of cells required for the area min prop_rem The proportion of distribution remaining in the selected area for the worst off species note that other species may have larger proportions remaining mean prop_rem The average proportion of distribution remaining in the selected area across all species weighted_mean_prop_rem The weighted average proportion of distribution remaining in the selected area across all species e the values for each species have been multiplied with a possible weight If no weights are used this equals to the basic average proportion average_90 _cell_count and 50 _cell_count These numbers will tell you how many cells are required to represent on average 90 50 of the species distribution that are located in the selected area We illustrate the 90 50 cells statistic with the following example Assume you select a Zonation solution top fraction that includes at least 30 of the distributions of all species that is Remaining 0 3 However for species A the selected area covers a bit more namely 40 of its original occurrences Now the 90 cells statistic is asking for 90 of the 40 36 of original remaining To get the number of cells the occurrences of species A are first ranked into decreasing order of
85. ature species interactions such as interactions between prey and predator or host and parasite can now be implemented to the landscape ranking process Thus sites can be valued not only by their importance to the target species but also how well they are connected to resources such as food wintering areas etc e Additional info in the Memo window Because of all the cool features listed above there is a lot of new information printed into the Memo e Facility for selecting top fraction of given cost Cost can now be used to calculate the corresponding top fraction of the landscape This new feature can be found in the Landscape identification window e New outlook for run settings Because of all the new features the outlook of run settings file and window have changed quite a bit Have a look e Special maps New special maps are accessible via the Map window These include the old richness and rarity maps but also a planning unit map and river basin map which indicates linked planning units with the same color e Addition to Curves output file If you have given conservation targets to your species ie using target based planning as your cell removal rule the curves output file now tells you at which level of cell removal process those targets for each species have been violated e Implementing corridors Corridors can be designed by utilizing a combination of mask file use and the BQP Essentially some good quality areas are masked in
86. be defined in terms of the extra funding required to maintain conservation value that is equal to the value of the optimal solution Thus one can define four variants of replacement cost 1 Biological exclusion cost Decrease in conservation value following forcible exclusion of a given group of site s 2 Biological inclusion cost Decrease in conservation value following forcible suboptimal inclusion of a given group of site s 3 Economic exclusion cost Increase in solution cost required to keep the same total conservation value following the forcible exclusion of a given group of site s 4 Economic inclusion cost Increase in solution cost required to maintain conservation value following the forcible inclusion of a given group of site s For practical purposes the replacement cost is calculated in the following manner First find optimal reserve selection X which has highest possible value F X obtainable with the available resource Cmax Then rerun the analysis with some areas forced in or out Replacement cost is the difference between the value of the optimal solution and the value of the new solution We emphasize that this does not mean keeping the optimal set of sites plus minus a particular site but finding a completely new solution given that the particular site s are forcibly included excluded A replacement cost of zero tells us that there exists an alternative solution with the same value as the current
87. best solution has i e same cost and same conservation value although obtained via a different selection of areas compared to the original optimal selection A replacement cost larger than zero means that any alternative solution including excluding the focal site s will have either a lower conservation value or a higher economic cost than the optimal one 2004 2008 Atte Moilanen Methods amp algorithms 38 Economical exclusion cost kr Biodiversity Value E Cre AC Resource b Biological inclusion cost CA Optimal Solution Replacement of site x Biodiversity Value CCX Cra Resource A conceptual illustration of the replacement cost of a hypothetical site a in terms of increase in resources required to maintain value AC and b in terms of loss of biodiversity value AF from Cabeza and Moilanen 2006 a Exclusion cost the dashed line indicates the value of the best solution when site x is forcibly excluded Up to a certain resource level CE site x does not belong to the optimal solution and thus exclusion cost is zero Even with C gt CE k exclusion cost can be zero if the site is fully exchangeable with another site or a combination of other sites b nclusion cost the dashed line indicates the value of the best solution when a site is forced to be included in the solution Inclusion cost is likely to be highest with low resource when the forced inclusion of the unwanted site
88. bution map file E g if the cell size of previous example would be given in meters instead of kilometers thus the cell size would be 1 000 m in the asc files instead of 1 km the a value would be as follows 21 a 0 00067 341000 Another common unit used in raster files is degrees Also these need to be converted to get the correct a value Let us assume that the cell size in our example was 0 0083 degree equaling approximately 0 860 kilometers Thus the a value in this case is 2 0 860 a _ _ __ 67 3x0 0083 It is important to understand that this parameter is NOT the same a value as is used in uncertainty analysis The two parameters only happen to have been denoted with the same symbol in literature Note also that if distribution smoothing is not used you should nevertheless enter a value in this column This can be any positive number e g a dummy value of 1 Do not leave the column empty Columns 3 and 4 together define either BQP Boundary Quality Penalty or NQP Neighborhood Quality Penalty in directed connectivity settings for the species depending on which one of the two options is used The information of these two columns is only used if BQP or NQP is included to the analysis Note that even if BQP or NQP is not used you should nevertheless enter a value into these columns They can be any positive numbers e g a dummy values of 1 Do not leave the columns empty 3 When using BQP this par
89. c with bounds for psc determined by a and the relative error measure Wsc which could be for example related to the accuracy of statistical prediction The model of Eq 1 is called an uniform bound model in info gap terminology When using predictions based on logistic regression habitat models a plausible model for uncertainty is to define the uncertainty interval in logit space where Wsc is the standard error for the linear predictor of a logistic regression logit po logit p x lt om 2 According to info gap theory one should favor reserve structures that achieve given conservation targets even with the most adverse choice of probabilities in other words in the worst case scenario Given the present definitions the most adverse choice of probabilities occurs when all probabilities are at their lower bounds this is when the lowest expected number of populations is obtained Assuming the analysis in logit space logit px logit p E e CM se 3 Thus the program calculates the discounted biological value of a cell by reducing discounting the value of the logit of probability p sc by a multiple of the error aWsc In the distribution discounting technique the original estimated occurrence data is simply replaced by the discounted data before proceeding to do the Zonation run Thus one Zonation run with discounted data is needed for each value of the horizon of uncertainty a Note that Zonation does n
90. ccurrences of each specie is remaining when landscape is iteratively removed e A rank asc raster file representing the order of cell removal ranking This file can be used to produce map images in GIS softwares e A prop asc raster file representing proportions of species distribution across species remaining at the removal of that cell This file can be used to produce map images in GIS softwares e A wrscr asc raster file This file file contains a weighted range size normalized measure of conservation value for each cell which can be used as a scoring measure of value for cells e A run_info txt text file copy of the Memo This file will be created after you have closed the program For saving other results pictures of specific maps or curves double click on the map image See also examples on visual output 2004 2008 Atte Moilanen 11 Zonation User manual 1 6 New features This section shortly lists the new features and small additions that are included in Zonation v 2 0 in comparison to earlier versions It also present some useful tricks you can use with Zonation Added for v 2 0 e Planning units As a new feature Zonation can now be run with planning units larger than one cell With a specific planning unit layer you can group the cells in your grid into larger entities which will be removed as a whole instead of singular cells during the landscape ranking process These planning un
91. certainty analysis BQP etc several settings files or with alternative species weighting schemes several species list files You can also use batch files to run the most complicated analyses requiring long computation times overnight or over the weekend When you wish to run multiple analyses the simplest batch file consists of several command lines each calling the program with different parameters e g do zig2 bat t t Notepad M ox File Edit Format View Help call zig2 r settingsi dat species spp outi txt call zig2 r settings2 dat species spp out2 txt call zig2 r settings3 dat species spp out3 txt Here the same species list file is run with three different settings Remember to identify the output files separately give different names or the program will overwrite old results after each run Notice however that when performing multiple runs you will have only the results that are automatically saved see file output and you will not be able to explore the results in any more detail with the windows interface Further you will not be able to perform the landscape identification procedure for each run but will have to do that separately after using solution loading This is because after each run the program will have to close in order to proceed to the next analysis run Therefore there also has to be a 1 in the end of each row to close the program If this parameter is 0 the program will not close itself and can n
92. connectivity which influences resource use Notably in some cases it might be appropriate to use log of connectivity instead of connectivity direct Such complications are likely to be implemented in future versions of Zonation 2 Interaction type 2 negative interactions competition avoidance of invading species or pollution etc The general idea is that one wishes to de emphasize those parts of the distribution of feature species A that are close well connected to the distribution of B Feature B could be for example a competitor a potential source of an invading species or a source of pollution that may cause future degradation of habitat quality and consequent reductions in the population sizes of species biodiversity features of conservation interest Using the notation above we now specify that the discounted value of feature j at cell i is R ju Z EP idad R r 41 0 min 1 0 gt 41 0 min 1 0 W Pe Si max Fe Bd ju 3 which is the local density of the focal feature r discounted by connectivity to the undesirable feature to be avoided u using parameter B to model the distances to which the undesirable influence spreads Effectively R is the distribution of species j which is not connected to the distribution of u 2004 2008 Atte Moilanen 41 Zonation User manual In equation 3 the nominator inside the brackets is the connectivity of the focal cell to the
93. ctions for conservation such as areas ear marked for residential building More commonly reserve networks are planned so that due to logistic or social constraints certain sites need to be included to or excluded from the final solution In most cases this leads to a suboptimal network either in terms of conservation value or in terms of the cost of achieving a given conservation goal Instructions how to include or exclude areas to a Zonation solution can be found in section 3 5 7 It is useful to be able to asses the degree of suboptimality of solutions compared to the optimal ones Here we introduce a method called replacement cost analysis which can be used to evaluate the effects of forced site inclusion exclusion Replacement cost refers to the loss in solution value given that the optimal cost efficient solution cannot be had and that alternative solutions with particular sites forcibly included or excluded must be accepted It tells us at what cost biological or economic can we exclude or include a site from the reserve network Assuming a constant budget the exclusion cost of that site is the loss in the network s conservation value that follows when a site that belongs to the optimal solution cannot be taken The inclusion cost of a site is the loss in conservation value that must be accepted if a suboptimal site is forced into the reserve network On the other hand when the conservation budget is not fixed replacement cost can also
94. e should not be any empty rows at the end of the species list file If necessary you can enter comments in your species list file on separate rows starting with the symbol Remember also to use decimal points not commas in all the input files Tips for using the command FOR to automatically produce species list files Note that the FOR command can be used in creative ways to automatically create species list files For example the following single command row typed and run from the command prompt FOR FE oa IN 1 1 900 DO GECHO L0 1 1 1 0 25 per asc gt gt my spp 11St Spy generates a file my _spp _list spp which has rows and relevant parameters for files p1 asc p2 asc P900 asc The gt gt at the end of the command indicates redirection of output into the following file Without the gt gt my_spp_list spp output is shown on the screen command prompt Another variant of the for command allows one to loop through a set of files using the normal wildcard file name specification FOR SL IN species e aser DO CECHO 1 0 L L 4 Uso XI See the help for the FOR command for further information Run settings file A dat file containing all basic Zonation settings This input file is needed only when running the program from the command prompt When running from the windows interface these same settings can be found on the Run settings window In the run settings file the following O 2004 2008 Atte Moilanen
95. e Sections 2 3 1 and 2 3 2 Arponen et al 2005 Moilanen 2007 2004 2008 Atte Moilanen 31 Zonation User manual 2 9 2 5 1 Uncertainty analysis common problem with conservation planning is the uncertainty of planning inputs Mostly these uncertainties are due to lack of data we simply do not have a comprehensive database with accurate information of the distribution of every species Uncertainty can also arise for example from outdated or false observations the use of predicted data e g distribution models or from any future factors such as the potential for anthropogenic land use changes or climate change Taking into account both biological value and uncertainty creates a prospect of four scenarios 1 Areas with high conservation value and high certainty of that information would be important for conservation 2 Areas with low conservation value and high certainty car parks etc would ordinarily rank low among conservation priorities 3 Areas with high estimated conservation value but low certainty have potential for producing negative surprises for conservation 4 Finally areas with low conservation value and also low certainty have potential for producing positive surprises estimated probability of occurrence high certainty of information positive low robustness a requirement Sn ry The goal of uncertainty analysis in reserve selection is to implement and evaluate tr
96. e grants 2000 2005 and 2006 2011 to the MRG lead by Academy professor Ilkka Hanski Helsinki February 26 2008 Atte Moilanen Academy Research Fellow Metapopulation Research Group Dept Biological and Environmental Sciences P O Box 65 FI 00014 University of Helsinki Finland Zonation User manual Table of Contents Part I Introduction 2 TAIMA PUIDOSO nintendo cn 2 2 THE Zonation frameworks A net inde inst 3 3 Zonation compared to other reserve selection approaches ss 4 LONA A en ee tree P E Ras ceases uc Rens Se Et E E A E E 4 Int ger programming nissan anne nent sine nus 4 Stochastic global search 5 4 Atypical Zonation Work FLOW sean cd tr nant ue 6 5 Software installation and quick Start A in ads ne aies 8 6 New features ina a a in ss sen ctn wnlatad want tbs on tele n item 11 Part Il Methods amp algorithms 15 1 References Mni atid ia 15 2 The Zonationmeta algo tim 16 S The cellremovalrUlO inca ita cos 17 Basic core ar a Zona lO ri en teen EE E E aae aena 17 Additive benefit TUMGTION cesse desert al ed eue ne diras dead en tin state E E a ent a eu Re en een s 18 Targ t baSed planning sesisinana abia 20 General differences between cell removal rules oooccooccnnonancnancnonancnonanononnnonaronnnnrrnnnrrnnnnrenanrrrnnn nena nnmnnn nennen 20 Generalized benefit func hoN sismo sas ware nds ii 22 4 Inducing reserve network aggregation coomocccnocccccnnancnnnnnnnnnanonnnonancnnnnn
97. e network in order to increase connectivity In reality this apparent loss is more than offset by benefits of having a well connected area Thus it is recommended to use aggregation methods in reserve planning as the cost of loosing a minor amount biologically valuable areas is usually low compared to the benefits of high connectivity For more information on true and apparent costs related to aggregation see Moilanen and Wintle 2006 and 2007 There are some distinct differences between the aggregation methods in Zonation and choosing the right one depends on conservation targets and computational issues e Boundary Length Penalty BPL has been the most commonly used way to introduce aggregation to reserve planning However it is important to understand that BLP is a general non species specific aggregation method which does not asses the actual effects of fragmentation on species Rather the method only uses a penalty on a structural characteristic of the reserve network boundary length to produce more compact reserve network solution The method is computationally quick and effective but might not be biologically most realistic Distribution Smoothing is a species specific aggregation method which retains areas that are well connected to others thus resulting a more compact solution The connectivity of sites is determined with a smoothing kernel which means that the value of a cell is smoothed to the surrounding area Another way of
98. e new solutions to evaluate If search starts far from the good regions of the search space it actually is not guaranteed that the good regions are found at all Good convergence with large problems absolutely is not guaranteed Multiple runs from different starting points are required to test for indications of convergence and if multiple runs reliably converge to a very similar result then this indeed is an indication that the solution probably is quite acceptable in terms of optimality Probably are ok with smallish data sets with thousands or tens of thousands of sites but at the million element scale the performance of these methods is poorly known Relative performance probably degrades when problem size increases which is opposite from what is actually expected for Zonation at least with the additive cell removal rules There are piles of literature on optimization which is an enormous field of science in itself See the references below for examples of the use of stochastic optimization on nonlinear reserve selection problems Also check MARXAN reserve selection software user manual and references therein Moilanen A 2005 Reserve selection using nonlinear species distribution models American Naturalist 165 695 706 AND in particular its electronic appendixes A C Moilanen A and M Cabeza 2002 Single species dynamic site selection Ecological Applications 12 913 926 For a more philosophical intro to these optimization method
99. e of how much biodiversity value a fragmented landscape would retain 1 gt L au ne E 1 a 1 Ei a ET oit w JU Picture of typical output map BQP has been included in the analysis Directed connectivity Directed connectivity is a specially modified version of BQP called Neighborhood Quality Penalty NQP where connectivity has a clear direction such as in river systems Use of the NQP technique requires the use of planning units which can consist of one or many cells The theory and algorithm behind neighborhood quality penalty is explained in section 2 4 4 This method has also been described in Moilanen Leathwick and Elith 2008 Running the NQP To include NQP to your analysis you need to 1 Give a planning unit layer to identify which cell belongs to which planning unit 2 Create a connectivity description file describing the linkage between planning units 2004 2008 Atte Moilanen ZIG The Zonation software 84 3 Give both upstream and downstream connectivity responses for all species These responses are defined in a the BQP definition file with the distinction that instead of one every species has two penalty curves 4 Link all species to the correct penalty curves by entering the correct row number of the respective curve into the third upstream and fourth downstream column of your species list file In addition you need to adjust your settings In command prompt 1 Set
100. e yellow the best 10 25 e light blue the best 25 50 e dark blue the best 50 80 e black the best 80 100 or the least valuable 20 The information of this map is equal to the rank asc file that the program produces as part of file output This map will also automatically be saved as a picture output jpg and output emf files but you can save it again e g with a different name or to a different directory by double clicking the picture The background i e the cells for which no data exists are shown in white In the beginning of analysis before overwritten by the ranking locations with SSI species are O 2004 2008 Atte Moilanen ZIG The Zonation software 68 shown as red dots In the options it is also possible to select a certain proportion of the best landscape based on the ranking e g the best 10 Type the wanted proportion from 0 to 1 into the Top rank field and press the Top rank button The area removed after this procedure is shown in and the best remaining area is shown in the same gradual colors Note that you can also use this option to show you the worst proportion of the landscape For example if you select the best 90 0 9 of the landscape the area shown in black equals to the worst 10 etc 2 Remaining In some cases it can be useful to find out those areas that are required for representing a certain proportion of species occurrences For example if you have a conservati
101. ed for the investigation of the robustness of a network that has been designed without any spatial aggregation method The description below is a brief description of the analysis please see the original publication for details We start from the assumption that cells ending near the edge of a reserve network will end up with decreased conservation value due to negative edge effects disturbance and disrupted spatial meta population dynamics Just like distribution discounting fragmentation uncertainty analysis is based on the info gap theory Since the main interest of this analysis is uncertainty in conservation value due to habitat loss and fragmentation error rates see section 2 5 1 are explicitly related to the amount of habitat loss occurring near the focal cell assuming the habitat close to the reserve network border is degraded The error model chosen for this study specifies that the true value of a cell is inside an uncertainty interval f ad p t py NE PE 0 1 1 where the relative error measure wsc has been replaced by the proportion of cells lost from the neighbourhood of focal cell c Le This is the proportion of cells that were available but were not chosen into the reserve network The uncertainty model above is a proportional error model in the info gap terminology Ben Haim 2006 In the equation f aLc could be any decreasing function of Le with f 0 1 and f x e 0 1 for all x gt 0 The condition f 0
102. ee section 3 6 5 solution cross comparison using solution loading You can load solutions either from the command prompt or the windows interface When operating the program from batch files or the command prompt type Ifilename as the second parameter of your call I points out that an old solution is loaded For file name enter the name of the ranking file from your old solution rank asc file one of the output files produced during each run Thus a typical call when loading an older solution would look like this E Command Prompt Cisicd Zonation CisZonation gt call 2192 lold_sol rank asc settings dat sp_list spp output txt 4 4 6 1 8 amp Remember to give a new output name in the call if you do not wish to overwrite your old solution If you are using the windows interface select the Load old solution rank file option from the Run settings window and type the name of your rank file on the field below Give the name of your species list file and adjust the settings if needed O 2004 2008 Atte Moilanen Zonation User manual 49 3 3 Input files amp settings 3 3 1 Introduction To use the Zonation software you need a set of input files some of which are compulsory some optional You can use some of the tutorial files as templates when creating your own input files All of these are text files technically ASCII files but to separate different types of files from each other it is useful to
103. egions marked as residential area mask_towns asc In this exercise we use the set_maske dat settings file all other input files are identical to the Exercise 5 2004 2008 Atte Moilanen Tutorial amp Examples 126 Batch file do towns bat Rank 1 0 Rank 0 15 Remaining 30 Top 15 Area 32 360 Area 16 543 BL A 0 251 BL A 0 281 Cost 32 360 cells Cost 16 543 cells av prop 36 5 av prop 19 7 min prop 29 9 min prop 16 6 This time you can see the changes in the beginning of species distribution curves Since the two excluded areas have high biological value the proportions of species distributions decrease more steeply in the beginning when these areas are removed However since other valuable areas can be included in the solution this masking has less effect on the final 15 top fraction compared to the old reserves as you can see from the average and minimum proportions of species distributions O IN N o 0 0 1 02 03 04 05 06 07 08 0 9 1 proportion oflandscape lost proportion gf distributions remaining Average and minimum performances when residential areas are excluded from the top fraction 4 8 Exercise 8 Working with directed connectivity As a final exercise we familiarize ourselves with the directed connectivity feature NQP Neighborhood Quality Penalty This is a variant of BQP and thus the basic principles for landscape rank
104. elements 4 000 x 4 000 Boundary Quality Penalty BQP e Maximum number of penalty curve profiles 50 e Maximum number of points on a penalty curve 20 e Maximum number of different species specific buffer sizes 100 Landscape identification e Maximum number of management landscapes 30 000 Memory requirements Memory requirements depend on the data you are using Naturally the larger the data many species and or high resolution and or large landscape the more memory you will need to run the computations One raster file of 1 million elements cells that have real data not missing values for one species requires 4 MB of memory Thus you can roughly calculate the maximum number of species that you can use with the help of this formula Max species 0 7 memory in MB 4 grid size in millions of elements The 0 7 in the formula accounts for the memory needs of the operating system and the memory needs of Zonation in addition to the species data matrixes Thus with 4 GB 4 000 MB of memory you can have approximately 0 7 4 000 4 5 140 species with 5 mil informative grid elements Using BQP mode 2 species specific missing data areas approximately triples the memory consumption compared to an analysis with no BQP used leading to a respective loss in the number of species that can be sued in the analysis Interaction layers count as independent features for the purpose of memory computations Note that this estimate of
105. ements For BQP mode 2 to have any influence missing data should be different for different species Note that you do not need to make any changes to the actual call in command prompt for using BQP 2004 2008 Atte Moilanen ZIG The Zonation software Run settings dat Notepad File Edit Format View Help 82 Settings removal rule warp factor edge removal add edge points 0 use SSI 0 SSI file name 551_list txt bge planning unit layer 0 panning unit layer Tile 1 100 1 plu asc SOST O cost Tle use mask O mask fi mask ras asc use boundary quality penalty 1 BOP profiles file BOPcurves Txt BOP mode 1 BLP O us tree connectivit 0 res connectivityeTile tree txt use interactions 0 interaction Ale interact txt use cost asc annotate lame O logit face 0 treaj zero areas as missing data z 0 25 r sample species 0 Info gap settings Info gap proportional 0 use info gap weights 0 Info gap weights file In the windows interface 4 11G2 Iterative cell removal done Showing removal rank Run mode a aaa Calculate new solution pecies file list ts in file C Load old solution rank file Removal rule Original core area Zonation Additive benefit function Start of output file name Use cost file l Use incl excl mask Use SSI file M Use planning unit layer C Target based planning C
106. es count Info gap settings Use info gap distribution discounting Info gap alpha 0 000 Uncertainty model f Uniform error default Proportional error List of error weight map layers load from u Cwerghts spp Seles Imy_splist spp transform species occupancy probabilities from logit values output dat Annotate outputfile name Interact spp Settings for generating spatial aggregation into the solution i Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000 Use boundary quality penalty BQPcurves txt BAP mode BOP aligns missing data BAP definitions file name C Potential habitat is species specific Use directed NOP PLU connectivity load from file tree Ext Boundary Length Penalty 0 000 m Additional edge points Picture of Run settings window when running basic Zonation without any additional analyses Weighting of species is a critical component of the algorithm Problems associated with different initial sizes of species distributions are circumvented in Zonation by assigning a value for the full distribution of each species By default these values are equal but species can be assigned differential weights in the species list file based on for example their taxonomic status global rarity economical value or population trend Weighting of species affects the order in which cells are removed from the landsca
107. es uncertainty analysis weights are used these parameters should be set to 1 0 Note that you can not use species specific error weights O 2004 2008 Atte Moilanen Zonation User manual without the distribution uncertainty map layers In addition you need to adjust your settings and the program call in the command prompt In the command prompt fe Command Prompt C cd Zonat ion C onationcall zig2 r settings dat sp_list spp output txt 1 6 6 1 8 A PF Run settings dat Notepad File Edit Format View Help settings removal rule 1 warp factor 100 edge removal 1 add edge points 0 use SSI 0 SSI file name 55I_list txt use planning unit layer 0 planning unit layer file plu asc use cost 0 cost file cost asc use mask O mask file mask ras asc use boundar y MALE penalty 1 BOP profiles file BOPcurves TXT BOP mode 1 BLP use tree connectivity 0 tree connectivity file tree txt use interactions 0 interaction file interact txt annotate name 0 logit space 0 treat zero areas as missing data 0 25 asample species Mint Agap settings ar o qap proportional O use info gap weights 1 Info gap weights file UCweights spp 1 Give a value to the uncertainty parameter a in your program call the fourth last parameter a value determines the horizon of uncertainty in the data and is usually unknown Thus you need to test generate soluti
108. esult in a more coarse solution If the warp factor is more than 1 of the remaining cells then only 1 is removed For example if there are only 100 cells remaining in the landscape then only one cell can be removed regardless of what the warp factor is Note that if you are using planning units PLU the warp factor is automatically set to 1 If planning units are not used warp factor can be defined freely In our tests having a warp factor of 100 has had little influence on the solution compared to lower warp factor values but the run times have been considerably shorter We recommend to use a warp factor of 1 mainly for the final runs if run times allow it You can compare the effects of different warp factors with Landscape comparison 2004 2008 Atte Moilanen 95 Zonation User manual edge removal Determines whether the program removes cells from the edges of areas value 1 or anywhere from the areas value 0 Note that setting this parameter to O might increase the running time substantially with large landscapes add edge points Randomly selects additional cells inside the landscape that will be initially classified as edge cells from which removal can proceed The value of this parameter determines the number of cells that are selected When adjusting the settings it is good to understand the function of Edge removal and Add edge points options The main profits of using edge removal is that it
109. eting the results 4a Distribution smoothing Using this method results more aggregated solutions based on the connectivity of sites To add the distribution smoothing to the analysis we use the same input files as in Exercise 2 but we call the program with a new batch file do_ds bat Note that the species specific smoothing is defined in the species list file where the width of the smoothing kernel for each species is given in the 2nd column and in the call itself where the third last parameter value 1 indicates that distribution smoothing will be done Batch file do ds bat Rank 1 0 Rank 0 15 Remaining 30 Top 15 Area 27 493 Area 16 543 BL A 0 205 BL A 0 285 Cost 27 493 cells Cost 16 543 cells av prop 33 2 av prop 20 8 min prop 29 9 min prop 18 6 4b Boundary quality penalty The boundary quality penalty BQP calculates the most valuable sites based on both the value of the cell and the effects of habitat loss in the surrounding cells The effects of fragmentation loss of neighbourhood cells are species specific and thus the BQP also takes in to account how different species are influenced by fragmentation and habitat loss BQP analysis demands prolonged computation times compared to our previous method because the loss of a cell now has an effect on occurrence levels in nearby cells which has to be taken into account in calculations To include the BQP to our analysis we now use
110. ettings window 1 First define the percentage of landscape This will determine how large part of the entire landscape will be included in the classification Note that the percentage here means the top fraction of the landscape E g value of 20 includes the best 20 area from the landscape to the solution 2 Give the nearest neighbour maximum distance in cells which is allowed between spatially discrete patches that are included in the same management landscape E g a maximum distance of 0 would mean that all separate groups of cells are identified as unique management landscapes 3 Give the maximum difference in species composition This determines how much the species compositions between two cells are allowed to differ in terms of relative densities for them to be joined to the same landscape value of 0 indicates that the species composition in two patches is identical Value of 1 indicates that the difference in relative density between two patches is on average log10 across species E g a maximum difference of 0 2 means that on average two species out of ten have a 1 log difference in their density or that one tenth of all the species have a 2 log difference For more details see Moilanen et al 2005 4 Define inclusion minimum which in turn determines how highly ranked cells must be included in each management landscape E g value of 10 means that each management landscape has to contain at least one cell which belong
111. evel of smoothing for a given species would be determined based on a conception of the typical dispersal distances for that species or from information concerning home range sizes for the species When using smoothing the value for species j in a focal cell is Oj gt gt exp d x u y r Or where O is the original occurrence level of species j at cell Cell is located in u r and d x u y r is the distance between locations x y and u r The summation is over the landscape grid and o is the parameter of the dispersal kernel for species This is a two dimension kernel smoothing using a radially symmetric negative exponential dispersal kernel Boundary Quality Penalty BQP This section is mainly based on Moilanen and Wintle 2007 Instructions to how to use BQP in Zonation can be found in section 3 5 3 The boundary quality penalty is a quantitative species specific way of inducing aggregation into Zonation solutions It can be seen as a way of approximating nonlinear effects of connectivity that O 2004 2008 Atte Moilanen Methods amp algorithms 26 may be present in habitat models The rationale behind the BQP goes as follows There are very many different statistical species distribution modeling techniques a k a habitat models resource selection functions Typically in such models the abundance of a species at a location is influenced not only by local habitat quality but also b
112. f all if you use any additional analyses such as aggregation methods or the uncertainty analysis the value of the cell will be calculated based on not only the species data but other features as well e g connectivity of the cell Thus differences between areas where the species is present do emerge Secondly even though in presence absence data there are no core areas in terms of relatively higher occupancy densities there still is the significant difference in the cell removal process between core area Zonation and additive benefit functions We highlight this with an example Let us assume we have a landscape where 7 different species occur Six of these species have overlapping distributions and one denoted here as species A has a distribution isolated from the other species Because benefit functions take species occurrences as additive the cells in sites where distributions of several species overlap receive a higher value than the cells where only one species occurs as is the case with species A Thus in the cell removal process the additive benefit function would always favour cells with multiple species over the cells of species A which would lead to unequal preservation of species in other words species A would loose its distribution much more quickly than the other species In contrast Core area Zonation would retain all species distributions equally meaning that species A would loose its distribution at the same pace as do the othe
113. f the network is relatively insensitive to an increasing If the proposed network has a spatial pattern that is sensitive to potential negative effects of habitat loss then the performance of the network degrades rapidly as a increases This would be the case when locally high quality locations are right at the edge of the proposed network which suggests that habitat loss outside the reserve could well influence what happens just inside the reserve at the locally valuable places Solution cross comparison using solution loading This analysis is used for example in figure 4 of Moilanen amp Wintle 2007 Conservation Biology 21 355 364 where effects of BQP on solutions originally calculated with the BLP are evaluated This is a major analysis which can produce very important information for example e How well a solution produced without connectivity criteria works if connectivity is actually needed e How much apparent conservation value is lost if a solution is developed requiring connectivity which actually is not needed e Likewise for the inclusion exclusion of interactions between species distributions e Surrogacy analysis a solution developed for one set of species can be evaluated for performance across a completely different set of species Overall the main point is that a solution can be developed using one set of criteria but post hoc evaluated using another set of criteria When loading an old solution the program does
114. fifth parameter w4 is the ordinary weight given for the species as the first column in the species list file Note that even though benefit functions or target based planning is not used you should nevertheless enter a value into the fifth column however no dummy values are needed to the three extra columns used with generalized benefit function This can be any positive number e g a dummy value of 1 Do not leave the column empty 6 Name ofthe species distribution map file asc raster file If your species distribution maps are in a different directory than your species list file remember also to type the correct path in front of the names Note that if you are using generalized benefit function as your cell removal rule the context of this column is shifted to column number nine E Species list file spp Notepad Alle Fie Edit Format View 0 O 10 0 0 4 3 25 4 5 F 10 speciel speciez specie3 specie species Lid LA la LAJ Li Picture of species list file wnen using generalized benefit function as cell removal rule When using the windows version type the correct path of your species list file on the Species file list in file box in Run settings window if all the files are in the same directory with Zonation program only the name of the file is required If you are running the program from command prompt type the name of your species list file as the third parameter on the command row Ther
115. g you need to decide the most reasonable options for your analysis These options would depend on the availability of data and on your planning needs Things that need to be decided include 11 1 Decide species weights Equal weights is the basic option but there may well be good reason to favour particular species by giving them more weight 11 2 Decide about how to induce aggregation into the final solution Options include distribution smoothing boundary quality penalty directed freshwater connectivity and boundary length penalty In general you want aggregation at least if your planning units are small like hectares or so because with small selection units population dynamics of nearby cells are strongly linked If planning units are very large like 10x10km cells then aggregation could plausibly be omitted 11 3 Decide if some amount of distribution discounting uncertainty analysis would be appropriate 111 Base analysis and sensitivity analysis At this point you have identified the analysis options which you believe to be most appropriate Next 111 1 Run your base analysis preferably using a relatively low warp factor 111 2 Run variants around your base analysis varying a single analysis feature at a time you probably cannot run all combinations of everything This is essentially a sensitivity analysis which is done by varying weights aggregation and uncertainty analysis settings within reasonable bounds Investigate usin
116. g solution comparison how big a difference various options make 111 3 An analysis of selection frequency with ZIG_Sum utility may provide useful summary information over analyses At this point you have a good idea of how different planning options influence your analysis and solutions iv Identification of reserve areas Identify management landscapes and check their statistics to find out why different areas are important what are the biodiversity features that occur there O 2004 2008 Atte Moilanen Zonation User manual v Evaluation of proposed reserve areas using replacement cost analysis If you need to evaluate proposed or existing reserve areas you can do that using mask files and replacement cost analysis This involves repeating your base analysis both with and without existing proposed areas included excluded O 2004 2008 Atte Moilanen Introduction 8 1 5 Software installation and quick start Installation The installation package includes the Zonation program zig2 exe the ZIG Sum utility zig_sum exe a user manual pdf and tutorial files You can find the installation package via the Metapopulation Research Group website www helsinki fi science metapop or directly from the Zonation pages www helsinki fi BioScience ConsPlan For practicality reasons it is recommended to keep data files including the tutorial in the same directory with the program One option is
117. having any information from that same place missing data Thus use this option with care The use of this option does not change your input files anyway thus cells with value O will remain as they are it will only change the way the program is reading the files This value is used to calculate the extinction risks of species as their distribution sizes are decreasing Z is the exponent of the species area curve S cAz which has been widely used in ecological studies In theory you can give Z any positive value but as a default it has been set to a commonly found empirical value Z 0 25 If using an exponent equal to zin an ABF analysis then Zonation is essentially minimizing the SA_curve predicted extinction risk across species This option allows you to for example test analyses using only a subset of species The program selects an random set of species from your species list file and uses them to run the analysis Thus you can run several analyses and check how the selection of species influences the outcome Note that the random sets do not include multiple selections of one species all species in the set are different ones To use this option enter the number species you wish to include in one set If this value is zero any negative value or equal to the total number of species no sampling is done O 2004 2008 Atte Moilanen ZIG The Zonation software 58 3 3 3 3 3 3 1 Info gap settings The
118. he respective confidence interval Or the probabilities of anthropogenic threat e g the uncertainty of occurrence of species in a given cell due to human activities in the near by future can be used as uncertainty value Or both The higher the uncertainty value the greater the risk that the species does not actually occur there although the species distribution data might suggest so Thus an uncertainty value of 0 indicates that the observed occurrence of a species Ain a given cell is trusted to be completely accurate Note that the uncertainty values should not be negative values BP Specie1 asc Notepad File Edit Format View Help ncols 649 nr ows 55 xllcorner 294205 0283004 6 00 NODATA_ value 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9 9999 9999 9999 1171011000 999977171111110010000000001 999 1 1 1 O0 0 0 0 002 2272 0072 000 0000003 2321 MIE xllcorner yllcorner 6282604 6 cellsize 200 INODATA_value 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 a 9999 9999 9999 0 0382769 0 095827 0 099487 0 083720 0 048773 0 044738 0 10038 0 100348 1999 0 023987 0 0338479 0 0278456 0 0298762 0 075628 0 044298 0 046287 0 086539 0 095287 0 0 Es gt Picture showing both the species distribution layer species1 asc and the uncertainty layer sp1_UC asc for spec
119. he beginning of each raster file No data rows at the edges of species distribution matrixes Computational efficiency requires the input data to have at least one row of no data on each edge of your species distribution grids Otherwise the program will automatically transform all values on the edge rows to missing data Differences in grid matrixes between species distribution files This might be a problem when you are using boundary quality penalty Remember to select the correct BQP mode based on the alignment of your species distribution data The correct O 2004 2008 Atte Moilanen 109 Zonation User manual BQP mode for non uniform data may be mode 2 which is slower and uses more memory e Differences in grid matrixes between species distribution and other data files Check that all species distribution rasters are congruent with any other raster files used in analysis e g cost layer uncertainty layers etc This means that all those cells in a grid which have data for any of the species used in the analysis that is to say the cells that are NOT marked as no data in all species distribution files also have to have a value in the optional raster grids Equally all cells marked as no data in all species distribution rasters should have the same definition in any optional rasters O 2004 2008 Atte Moilanen ZIG The Zonation software 110 3 9 Tips for using the command prompt Here
120. he program can also be operated by using the windows interface Double clicking on the zig2 exe icon starts the windows version of the program To run the program go to the Run settings window Click here to enter run settings 4 HG2 the Zonation software for spatial conservation prioritization Maps Run settings Species info Memo Landscape identification Solution comparison About ComboBox2 M wmap Special maps Top rank 1 0 Remaining 10 3 A Stop Z Picture of the Zonation interface when opened _ Enter the name of your species list file and 2 Give a suitable name for your output files Remember to write the correct suffix after each file name e g ouputfile txt 3 If needed adjust the settings see section 3 3 2 3 Run settings and press the Run button to initiate the computation After running the program the results of your data are displayed in the Maps window Rest of the visual output is shown in the Species info window Note that although no aggregation has been included in this example it is recommended that some aggregation methods e g distribution smoothing BQP etc would be used when running the final analysis to increase the conservation values of the solution 2004 2008 Atte Moilanen ZIG The Zonation software 46 4 Z1G2 Iterative cell removal done Showing removal rank Maps Run mode g ee fie ket iain Gl x ecies file list 1s in file f
121. he proportional coverage minimum set solution for the data In this function value Vj is zero until representation Rj reaches the target Tj Then there is a step with the height of n 1 where n is the number of species When Rj increases above Tj and approaches 1 there is a convex increase in value with a difference in value V 1 V 7 1 This means that the loss in value from dropping any one species below the target is higher than any summed loss over multiple species that stay above the target value Y 0 0 0 2 0 4 0 6 0 8 1 0 proportion of distibution remaining The idea is that as cells are iteratively removed species representations will approach the species specific targets from above and that the convex formulation with increasing marginal losses will force species to approach targets in synchrony in terms of lost value Thus as one of the species approaches the target level the program starts to avoid removing cells that contain that particular species at the expense of other species in order to retain the target At some point it will not be possible to remove any more cells without violating the target for at least one species After one of the species has declined below target the remaining distribution of that species has no value for the reserve network Thus removing cells where only this species occurs does not increase the loss of biological value from network anymore Note however that the definition of how margi
122. his chapter includes descriptions for the three different types of analyses that can be conducted for solutions produced from the main Zonation runs These three groups are 1 Post processing analyses which can be done manually from the Zonation windows interface These include e Landscape identification e Statistics for management landscapes e Solution comparison e Fragmentation uncertainty analysis 2 ZIG Summary utility which is run by a separate ZIG Sum program comes together with the Zonation program 3 Solution cross comparison using loaded solutions Landscape identification This process is also explained in Moilanen et al 2005 Proc R Soc B 272 p 1886 gt and figures 2 and 3 This option allows identification of separate management landscapes based on the distance and similarity in species composition between two sites Spatially distinct areas consisting of multiple grid cells in a Zonation solution can be classified into management landscapes An area is joined to a landscape if it is close enough and similar enough in the species composition to any other distinct area in the same landscape Landscape identification is done for a given fraction of the landscape Running landscape identification The landscape identification can only be done from the windows interface Being a post processing analysis you first need to run the Zonation analysis with your data or load an old solution rank asc file in the Run s
123. hree new edges Thus as a result the total change in boundary length is 2 and so on To get from AL to A BL A one needs to account for both the change in boundary length and the decrease of the reserve area by one Note that BLP is different from both distribution smoothing and the BQP First the BLP is not a species specific way of handling reserve connectivity lt simply uses a penalty term that devalues reserve structures with lots of edge This is completely qualitative there is no species specific parameter or response Like distribution smoothing the BLP may be expected to perform poorly for species that happily occur in fragmented habitats This is because the BLP qualitatively favours structurally connected areas and it will therefore have a tendency to remove small habitat fragments from the solution irrespective of whether some species can actually persist in them or not Directed connectivity NQP This section is mainly based on Moilanen Leathwick and Elith 2008 Instructions for how to use directed connectivity in Zonation can be found in section 3 5 4 The directed freshwater connectivity measure is a generalization of the BQP technique in which the concept of neighborhood is generalized hence the name Neighborhood Quality Penalty NQP Instead of using a circular neighborhood the NQP is defined using a tree hierarchy of linked planning units A focal area planning unit is influenced by negative action habitat loss
124. i Ta Mhl A 1 yl a En A F asl 0 A picture of landscape identification output map 7 Running the landscape identification produces a map which shows you the separate landscapes in different colors Areas that were not included in the selected top fraction see Percentage of landscape above are shown in blue The colors in landscape identification output have no special 2004 2008 Atte Moilanen 97 Zonation User manual 3 6 2 interpretation they are random colors for distinguishing the separate landscapes from each other Repeat analysis if colors appear unsatisfactory F1G2 Iterative cell removal done Showing removal rank Maps Run settings Species info Memo Landscape identification Solution comparison About 110260 E Miscellaneous information about the progress of analyses species count removed 090000 smallest remaining proportion over all species 0 2235 a removed 100000 smallest remaining proportion over all species 0 1154 removed 110000 smallest remaining proportion over all species 0 003 SSI spp count Total count of cells removed 110283 anes 1O pann R Done in 29 seconds matrix x dimension Species performance levels check Pa it E arian P E P matrix y dimension IAJ Species 2 i Species 3 HEEL Species 4 0 0000 Species 3 Species Species cells with species data missing data cells DONE 494913 Potential cells count 2205
125. ies Essentially the rarity map is scaled by the highest fraction of the distribution of any species occurring in the cell Note the difference between rarity and richness maps areas can have a high conservation value by containing cells that are either good for many species combined richness or that are most important for a single species rarity Both species richness and rarity maps are in grey scale An example of a rarity map where black areas show the core areas of species distributions Areas with no data are shown in blue Planning units If you are using planning units in your analysis this option displays a map of them using randomized colors River basins This map of river basins shows linked planning units by the same color it is useful for visually checking that the linking of planning units is correct Make sure to repeat the display a few times especially when using grey scale color as units could accidentally end up having a similar shade making them hard to separate from each other 4 Species specific maps All maps explained above are summaries over all species From the dialog box in top left corner of the map window you can select species specific maps showing the input distributions of species Note that if the species distribution layers have been modified anyway during the analysis e g smoothed with distribution smoothing or discounted with uncertainty analysis these maps show the distributions after
126. ies 1 During the distribution discounting process the value in each cell of the distribution layer will be discounted by a multiple of the corresponding value in the uncertainty layer It is important that the uncertainty layer raster has the same grid size as the species distribution map rasters This means that in all files the number of columns and rows as well as the size of cells should be equal It is equally important that all those cells which have data of the species occurrence that is to say the cells that are NOT marked as no data in the respective species 2004 2008 Atte Moilanen 63 Zonation User manual 3 3 3 6 3 3 3 7 distribution file also have to have an uncertainty value Otherwise undefined program behaviour may occur Remember also to use decimal points not commas in all the input files Uncertainty analysis weights file A file containing a list of all distributional uncertainty map layers asc raster files with each species file on their own row This file is needed if you are including the uncertainty in species distributions into your analysis This file always contains two columns E UCweights spp Notepad AE Fie Edit Format View Help 0 UClayer_speciel uc layer_specie2 uc layer_specie3 uclayer_specie4 uc layer_specie5 Picture of uncertainty analysis weights file 1 Species specific weights in the uncertainty analysis With these you can stress the accurac
127. ignificant fraction of the entire range of an endemic species or two with relatively high weight This map can be used as a scoring value for the cell which can be useful for example when comparing two cells with a replacement cost value of zero the cell with the higher wrscr value would be more important Wrscr values could be used for example to inform agro urban landuse planning of the potential intrinsic conservation value of small land parcels It is emphasized that the wrscr measure does not take into account any complementarity or connectivity considerations and use of this measure does not replace a full Zonation analysis Two areas could have equally high wrscr values but due to the occurrence of a completely different set of species which is accounted for in a Zonation analysis but not by the wrscr measure Note also that distribution smoothing and interactions influence wrscr values as it is calculated from the data that is used in Zonation computations and this data includes effects of all transforms done to input maps run_info txt file A text file copy of the Memo This file will be created only after you have closed the program You can use it to go back to see what happened in your analyses Note that some error messages or warnings may appear here The content of the memo should be checked after a serious analysis run to verify that correct options appear to have been used and that there are no worrisome error
128. ilanen Tutorial amp Examples 122 4 5 Exercise 5 Possible gaps in species information Due to lack of data uncertainties in species distributions are a common problem encountered in ecological studies However these uncertainties can be accounted for in reserve selection when using the Zonation program Conceptually the program uses uncertainty analysis to focus on sites where the prediction uncertainties are low compared to the predicted representation levels Thus the program prioritizes sites that have both high abundance and low uncertainty To continue with our exercise let us think that our species distribution data has been provided by statistical species distribution models a k a habitat models or resource selection functions Models contain many uncertainties in their predictions and we want to take this into account when we are selecting the best sites to be protected We have uncertainty layers for each of the species sp1_UC asc sp2_UC asc etc which show the spatial distribution of uncertainty in our data The pair of figures below shows as an example the distribution and its error surface for one species Distribution layer Uncertainty layer Figure showing the distribution of species 2 and its corresponding uncertainty layer Black indicates high occupancy levels or high error and white low occupancy or low error You can display these maps by selecting the species from the pull down list in the
129. indows interface 8 45 www 8 Z ZIG sum utility 103 Zonation compared to other reserve selection approaches 4 Zonation features 3 Zonation meta algorithm 16 Zonation V2 11 15 22 28 39 58 59 83 89 Zonation workflow 6 z value 53 2004 2008 Atte Moilanen For the newest version of Zonation software visit our web page www helsinki fi bioscience ConsPlan 2004 2008 Atte Moilanen Contact information Metapopulation Research Group Department of Biological and Environmental Sciences Phone 358 9 1911 Exchange P O Box 65 Viikinkaari 1 Fax 358 9 191 57694 FI 00014 University of Helsinki FINLAND
130. ing units which are then removed as a whole during the landscape ranking process planning unit layer file Indicates which planning unit layer will be used use cost Determines whether land costs are included in the analysis value 1 If no land costs are used this parameter should be set to 0 cost file Indicates which land cost file will be used use mask Determines whether a removal mask layer is used value 1 or not value 0 mask file Indicates which mask layer file will be used use boundary quality penalty Determines whether BQP is used value 1 or not value 0 Use of BQP leads to solutions that include aggregation at scales relevant for individual species features BQP profiles file Indicates which BQP profiles file will be used BOP mode Determines how the program will calculate the effects of fragmentation from species distribution data Essentially this parameter tells the program what type of species distribution layers you are using Mode 1 indicates that the data no data matrix in all species distribution map files should be uniform and aligned and that there are no differences between species in terms of which cells are considered O 2004 2008 Atte Moilanen ZIG The Zonation software 56 potential habitat and which are then used in BQP buffer calculations In other words all species would be dependent on the same general habitat type such as forest With mode 1 Zonation a
131. ing are the same as in BQP The essential difference is that here the connectivity of sites is clearly directed and defined as a set of linked planning units instead of a buffer area O 2004 2008 Atte Moilanen 127 Zonation User manual around the focal cell as in basic BQP To illustrate this we now forget all the previous analyses we done so far and assume that the seven species we want to protect are in fact freshwater species that live in the numerous rivers in our study area Thus to account for connectivity while selecting sites for conservation actions we need to consider both upstream and downstream connections of a site Picture of the study area divided into planning units color and river basins grayscaled Because NQP works with planning units instead of grid cells we have created a planning unit layer of the area called plu_file asc Planning units can be defined by any criteria so lets assume that in our case they describe the smaller water catchment areas of riverine systems The direction of water flow ie linkage between planning units has been defined in a directed connectivity description file rivers txt where every single planning unit has been linked to the following downstream unit The penalty curves are given in the NQPcurves txt file where we have a general curve for upstream and downstream connections respectively Settings for running the analysis are in set_nqp dat and in this case
132. ion file on the field If you are running the program from command prompt type into your Run settings file use interactions 1 option selected and interaction file myfile txt name of your interaction definition file O 2004 2008 Atte Moilanen 67 Zonation User manual 3 4 Standard Zonation output Next we will describe the basic output produced by the program Running Zonation automatically produces two sets of outputs 1 Visual output in the windows interface 2 File output These files will be saved in the same directory as the program unless you have specified another path for your output 3 4 1 Visual output Maps window Species info and Memo window 9 OQ Q 3 FIGZ2 iterative cell removal ome Showing removal rank KER Aa PE E a gt a E F 3 Maps 4 Run settings Species info Memo Landscape identification Solution gomparison About 5 T Wmap Special maps Top rank 1 0 Remaining 0 3 A stop Z Picture of the Maps window after the program has finished running 1 Top rank In the Maps window the first map that appears and the creation of which you can follow on screen during the iterative cell removal is the ranking of the landscape In the picture sites are ranked by using different colors to indicate the biological value of the site e red the best 2 of the landscape e dark red the best 2 5 e magenta the best 5 10
133. ions are actually remaining in your reserve network when you take the uncertainty effect into consideration Vs but when there is yet no fragmentation Note that here again the value indicates the proportion of distribution for the worst off species Thus other species may have a larger proportions remaining For example from the picture above we can see that when a has a value of 0 000 no uncertainty each species has at least 57 56 of their distribution remaining in the landscape But as the a value increases up to e g 0 577 the minimum proportion of species distribution in the very same landscape decreases to 47 78 In addition the level of fragmentation amplifies the loss of biological value in single cells When 87 5 of the neighbourhood habitats are remaining the biological value of a focal cell has decreased by 7 When 50 of habitats are remaining the biological value has decreased by 25 etc But how to determine the value of uncertainty The answer to this is that a has no correct value the range of a values that are of interest depends on how much certainty is required The main question here is what happens when you increase the value of a As mentioned earlier the primarily goal of this method is to compare the performance of different network candidates in the light of increasing uncertainty lf a network has a spatial structure that is resistant to negative effects of neighboring habitat loss then the performance o
134. is the fraction of the original distribution of species in the remaining set of cells S The min max structure of the equation also indicates a strong preference to retaining the best locations with highest occurrence levels Thus the program can spare otherwise species poor cells if they have a very high occurrence level for one rare species It is important to understand that core area Zonation does not treat probabilities of occurrence as additive ten locations with p 0 099 is not the same as one location with p 0 99 However this is strictly true only when analysis is based on biological value only and when a landscape cost layer is not used in the analysis When cost is used cell removal is based on local conservation value divided by cell cost efficiency and now a high value for a cell can be explained with either i a very high occurrence level for some species or ii low cost for the cell Thus when cost information is used the interpretation of a core area becomes vague and this should be recognized in planning Therefore it is not recommended to use cost layers when trying to find out biologically most important areas with core area Zonation A B gt w 1 w 10 This figure illustrates principles that core area Zonation implements in numerical form Essentially the question is if you have two multiple species and you are going to lose a fraction here one cell marked as yellow of one distribution then where w
135. istribution files also have to have a cost value Otherwise undefined program behaviour may occur Remember also to use decimal points not commas in all the input files When using the cost layer remember to type in to your Run settings file use cost 1 cost option selected and cost file yourcostfile asc name of your cost layer file If you are running the program from windows version go to Run settings window select the Use cost information option and type the correct path to your cost layer file if the file is in the same directory with Zonation program only the name of the file is required Removal mask layer A standard GIS raster file asc file which determines the removal hierarchy of the edge cells The main use of the mask layer is replacement cost analysis see section 2 6 This file includes all basic raster information as explained in species distribution map files followed by a matrix where cells are categorized as follows 0 Cells with a value of 0 are considered as normal cells with no preference to removal hierarchy These cells will be removed after there are no more cells with a value of 2 left 1 Cells with a value of 1 are removed last These cells may for example have a special conservation value or they may already be ear marked for conservation These cells will be removed only after there are no more cells with values of O or 2 left and are therefore forced into the top fraction of the sol
136. istributions for the worst off species compared to core area Zonation see figure of the first three removal rules in section 2 3 4 To find more information about the use of benefit functions see Arponen A Heikkinen R Thomas C D and A Moilanen 2005 The value of biodiversity in reserve selection representation species weighting and benefit functions Conservation Biology 19 2009 2014 Arponen A Kondelin H and Moilanen A 2006 Area based refinement for selection of reserve sites with the benefit function approach Conservation Biology 21 527 533 Cabeza M and A Moilanen 2006 Replacement cost a useful measure of site value for conservation planning Biological Conservation 132 336 342 Moilanen A and M Cabeza 2007 Accounting for habitat loss rates in sequential reserve selection simple methods for large problems Biological Conservation 136 470 482 van Teeffelen A and A Moilanen 2008 Where and how to manage Optimal allocation of alternative conservation management actions Biodiversity Informatics 5 1 13 2004 2008 Atte Moilanen Methods amp algorithms 20 2 3 3 Target based planning 2 3 4 This section is mainly based on Moilanen 2007 Target based planning is implemented in Zonation by using a very particular type of a benefit function the purpose of this special functional form is to enable the Zonation process to converge to a solution that is close to t
137. it layer is in use the entire planning unit is removed simultaneously Note that cells within a planning unit are removed simply from top to bottom thus here order of individual cell removal has no biological function The cell removal rules operate as before but they operate on value aggregated across the planning unit Also the cost of the planning unit is taken as either the summed cost of cells if the cost layer is used or as the area of number of cells in the planning unit if costs are not used Note that each planning unit does not need to be spatially continuous a planning unit may consist of a scattered collection of cells plu asc Notepad wmat View Help 649 555 294205 6283604 6 0 2 NODATA_value 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9 9999 9999 E888888888BEBBEREBEBE88BE8BBEBEBEBEEBEBEBEBEBEBEB 2 42 42 42 42 42 42 42 42 42 42 42 42 42 42 42 42 42 9999 42 42 42 42 42 42 42 43 43 43 43 gt E A planning unit number should be defined for every cell that has species data if not some kind of error condition is likely to occur It is not harmful to have planning units extending outside the area with species data the critical bit is that all locations with species data are covered with planning units Use of large planning units will automatically cause a decrease in the quality of results The reason for this is that large planning units will probably contain bo
138. its can be of any shape and size This option can be useful in situation when e g land ownership dictates that certain groups of grid cells should be treated as distinct units e SSI species A k a species of special interest These are the second kind of species occurrence information that can be entered into Zonation in addition to traditional species distribution maps The input for a SSI species is a probably relatively short list of observation locations instead of a map This option is particularly useful with species that have very few occurrences and can not be modeled to produce a comprehensive distribution map e New output Zonation v 2 0 has new outputs The species specific habitat quality distribution which is displayed in the Species info window and the scoring measure for grid cell value which is produced together with the other output files e New cell removal rule Cell removal rule number four is the generalized benefit function This is a two piece power function which allows high flexibility in defining the shape and operational features of the function e Directed freshwater connectivity This is a development of the aggregation method boundary quality penalty with the distinction that the connectivity between sites is directed thus it is not measured all around the cell as in BQP This feature is particularly useful when working with clearly directed systems such as rivers e Species interactions With this new fe
139. l habitat models The difference between the methods would be most strikingly visible in fragmented areas Distribution smoothing perceives the value of fragmented areas as relatively low In comparison the BQP could recognize a species that happily lives as a metapopulation in a fragmented environment the response for that species would be such that it is recognized that the species can have high value habitats in fragmented areas In the implementation of the BQP into Zonation the value of a cell that is removed is now divided into two components i local value which is as before and ii loss of conservation value in the neighborhood of the focal cell as modeled via the BQP specification Thus with BQP the effect of cell removal is not only the loss of the value in the cell itself but also a potentially species specific reduction in quality in the neighborhood cells Fi m l 7 H j Mg Ay where Nj i indicates the cells containing data for that species within the species specific radius of cell for species j Denoting by h the fraction of original neighbors that have been lost from within the species specific buffer around the site Hi h is the proportion of the original value of cell k remaining for species j when the focal cell has fraction h of its neighbors remaining The fraction of cells remaining is simply h nx nx where nx is the number of neighbors remaining for cell k within the buffer radius of species j and nx
140. le tells the program if some areas need to be included to the top fraction e g old reserves or excluded from it e g areas ear marked for residential building In addition you need to adjust your settings In the command prompt 1 Set use mask to 1 in your settings file to indicate that a removal mask is used 2 Give the name of your removal mask file PB Run settings dat Notepad File Edit Format View Help settings removal rule 1 warp factor 100 edge removal 1 add edge points O use SSI 0 SSI file name 5sI_list txt use planning unit layer 0 planning unit layer Tile plu asc us amp COST 0 costfile cost asc use mask 1 mask Tile my_mask ras asc use boundaf y Me penalty 0 ROP proaf les file ROQPcurves txt BOP mod 1 BLP use free SEO ey 0 tre connectivity file tree txt us interactions O interaction file interact txt annotate name 0 logit space 0 treat zero areas as missing data 0 zZz 0 25 resample species 0 Info gap settings Info gap proportional use info gap weights 0 Info gap weights file UCweights spp Note that you do not need to make any changes to the actual call in the command prompt for using a mask file In the windows interface 1 Select the Use incl excl mask option from the Run settings window 2 Define the name of your removal mask file 2004 2008 Atte Moilanen 93 Zonation User manual
141. le which contains a list of all the landscape identification files All input files should either be in the same directory with the ZIG Summary program or you should type the path to the correct directory in front of the file names E solutionlist txt Notepad File Edit Format View Help loutputi ras asc Output ras asec oOUtputs ras asc Pas asc Pas asec Fas asc O 2004 2008 Atte Moilanen ZIG The Zonation software 104 Note that technically there is no reason why the proportions of original landscape should be equal in all raster files The program will run regardless of the settings but the interpretation of results changes if different fractions are used Running ZIG Summary ZIG Summary can only be run from the command prompt 1 Open a new txt file for example with notepad and type the following command in it zig sum solutionlist txt output asc O Maps Jpg First parameter is the name of the program Second parameter is the name of the txt file where you have a list of your landscape identification solutions For third parameter enter a suitable name for your output raster data asc file Fourth parameter determines the layout of your output map value O resulting a grey scale picture and value 1 resulting in color picture For fifth parameter enter a proper name for your output map jpg file 2 Save the file as bat file e g dozigsum bat in the same directory where you have the ZIG
142. lel txt copies the file file1 to directory my_docs Command copy filel my docs copies all files with the name file1 into the directory my_docs with their original names and formats You can also copy your files under a new name i e with command copy filel my docs file2 all files are copied to the directory with a new name file2 Renames a file After the command type the name of the file you whish to rename and then the new name for that file l e rename old name txt new name txt Prints the contents of a text file in the command prompt Edits command lines calls Windows commands and creates macros A useful tool which allows you for example to move back and forward in your commands with the cursor moving keys Thus you do not need to repeat same commands by writing them time after time pressing the arrow up key displays all recently used commands To activate the tool write doskey and press enter Calls a program inside a batch file E g call zig2 With this command you can enter comments into your batch files Closes the command prompt 2004 2008 Atte Moilanen 111 Zonation User manual 2004 2008 Atte Moilanen 113 Zonation User manual Tutorial amp Examples The purpose of this tutorial is to illustrate the use and function of different Zonation analyses They also help you to familiarize yourself with the program and its settings Later on when
143. lgorithm The Zonation algorithm Moilanen et al 2005 produces a hierarchical prioritization of the conservation value of a landscape hierarchical meaning that the most valuable 5 is within the most valuable 10 the top 2 is in the top 5 and so on At a high level Zonation simply iteratively removes cells one by one from the landscape using minimization of marginal loss as the criterion to decide which cell is removed next The order of cell removal is recorded and it can later be used to select any given top fraction like best 10 of the landscape Simultaneously information is collected about the decline of representation levels of species during cell removal Essentially the algorithm applied by Zonation is a reverse accelerated iterative heuristic Reverse comes from starting from the full landscape and removing cells this is important for the treatment of connectivity Accelerated comes from the option of removing more than one cell at a time via the warp factor The Zonation meta algorithm 1 Start from the full landscape Set rank r 1 2 Calculate marginal loss following from the removal of each remaining site i i Complementarity is accounted for in this step 3 Remove the cell with smallest di set removal rank of to be r set r r 1 and return to 2 if there are any cells remaining in the landscape Thus sites are ranked based on biological value and the least valuable cells are removed one or more at a time
144. looking at distribution smoothing is that it does a two dimensional habitat density calculation identifying areas of high habitat quality and density Consequently cells that have many occupied cells around them receive a higher value than the isolated ones The widths of the smoothing kernel are species specific implicitly expressing the species dispersal capability or scale of landscape use This aggregation method is computationally very quick However it assumes that fragmentation low connectivity is generally bad for all species and it always favors uniform areas over patchy ones Boundary Quality Penalty BQP is biologically the most realistic aggregation method included in Zonation This method describes how the local value of a site for a species is influenced by the loss of surrounding habitat The change in local value is based on species specific responses to neighborhood habitat loss thus local value may also increase if the site includes species that benefit from fragmentation The downside of this method is the required computation time which is much higher compared to the other two aggregation methods This is because each cell removal influences the habitat value in all remaining neighborhood cells which needs to be accounted for in the cell removal process e Directed connectivity Neighborhood Quality Penalty NQP is a generalization of O 2004 2008 Atte Moilanen 25 Zonation User manual 2 4 1
145. mainly based on Moilanen et al 2006 Instructions for how to use distribution discounting in Zonation can be found in section 3 5 5 Distribution discounting is a method for including uncertainty analysis into the conservation prioritization done in Zonation This method helps you find the most robust solutions those that 2004 2008 Atte Moilanen Methods amp algorithms 32 most likely achieve a conservation goal given a level of uncertainty in species distributions This analysis utilizes both the estimated biological value probability of occurrence of a species in a cell and the certainty of that information Looking for robust reserve networks In the framework of uncertainty analysis one goal for reserve selection would be to find those network candidates that would achieve the given conservation targets despite uncertainty in input data Thus cells need to be ranked so that the highest priority is given to cells that have both relatively high conservation value and high certainty of information In Zonation uncertainty analysis has been implemented according to a convenient formulation that uses information gap decision theory see Ben Haim 2006 Conceptually relevant components of the info gap theory are 1 The nominal model This is your best set of predictions for species 2 The uncertainty model This states that even though you do have a nominal estimate your true probabilities of occurrence are cer
146. memory consumption is only indicative but sufficient for getting an idea if an analysis definitely should or should not run Check amount of available RAM in Windows task manager to verify that Zonation has not run out of memory O 2004 2008 Atte Moilanen ZIG The Zonation software 108 3 8 Troubleshooting Here is a short list of things to check when encountering problems Directory paths Check that you have entered the correct paths to your files so that the program can find them If you have the program in the same directory with your input files recommended you do not need to type the paths just the file names However if some all of your input files are located some where else a directory path is needed for these files If files are in a subdirectory from the exe directory then filenames can be entered as sub_dir_namefilename File names One reason for problems can be the long directory and file names especially if you are using the command prompt Thus try to keep the directory names short e g max 8 characters Also do not use any spacings in your directory or file names Computer memory capacity If the program is running very slow during computations check Task Manager Performance If your RAM Physical Memory is close to zero you have run out of memory This does not mean that the program has jammed but it will take a couple of lifetimes for it to finish up the calculations In other words st
147. most likely increases computation times significantly especially if the species specific buffer sizes are large many cells Thus it might be wise to reduce the data resolution if computation times start to increase in undesirable ways lt is also recommended to use a moderate to high warp factor for your preliminary runs and a low warp factor only for making the final solutions low warp factors are recommended for final BQP runs Using BQP should result in a distinctively more aggregated solution compared to basic Zonation analysis at least in some part of the landscape Thus these two solutions should not have e g 99 overlap with each other unless your species are not influenced by habitat loss as defined in the BQP file If your solution with BQP shows signs of high fragmentation check the run settings for possible errors Note however that a BQP solution may well include fragmented areas if the data contains many species that are indifferent to fragmentation or even favor fragmented habitats Note that the use of BQP will produce species specific curves that show lower proportions retained as compared to the basic non spatial analysis This is because use of the BQP implies that habitat loss and fragmentation will have negative consequences also for areas not yet lost We emphasize that this does not mean that the solution developed by BQP is inferior to the basic analysis but rather the basic analysis gives an overly optimistic estimat
148. n discounting also here the evaluation of robustness means that we are interested in the most adverse choices Therefore we look at the lower bounds of probabilities This means that in Eq 2 psc is equal to f aLbc p sc Different reserve structures will be differentially resistant to negative effects of fragmentation very fragmented reserve networks will have the characteristic that if soecies occurring in the network are sensitive to fragmentation then the true value of the network can be much less than what is estimated based on an analysis that does not include explicit spatial effects In contrast a large 2004 2008 Atte Moilanen Methods amp algorithms 36 and well connected reserve will be relatively insensitive to negative effects of fragmentation because only a small fraction of the reserve area will be close to the reserve edge Essentially such a reserve would have a large core When running the fragmentation uncertainty analysis the program calculates how much biological value might be lost from a candidate reserve network as the uncertainty and or the fragmentation of surrounding habitats increases see section 3 6 4 O 2004 2008 Atte Moilanen 37 Zonation User manual 2 6 Replacement cost analysis This section is mainly based on Cabeza and Moilanen 2006 Very seldom can reserve planning be started from a fresh table with areas that have no older reserves nor any restri
149. n discounting can be significant or not In the Memo window you can find more detailed information about the analysis Note that as the program starts to run the analysis it recalculates the species values based on the amount of uncertainty Thus for each species the program first displays the absolute value in the whole landscape sum over all cells and then calculates how large fraction of this value can be expected to occur in the landscape with certainty This value depends on the horizon of uncertainty parameter and on the level of uncertainty in the data 2004 2008 Atte Moilanen 89 Zonation User manual 3 5 6 Be Picture of our example landscape when uncertainty in species occurrences was included in the analysis 11020 AUN Miscellaneous information about the progress of analyses Loading feature e g species data layers ee sol UL asc 555 rows read in total sum of non missing data elements 18239 14 6086321 125 800003 111 000000 5 000000 649 555 DISTRIBUTION DISCOUNTING loaded error weights map from filesp1_ speces asc 555 rows read in total sum of non missing data elements Used Info gap alpha 0 5 spw 1 Using Info gap weight matris species asc fraction of onginal occurences remaining after DISTRIBUTION DISCOUNTING info gap 125 300003 111 000000 5 000000 649 555 Completed load of species file species asc UC asc 555 rove read in total sum of non mizaing
150. n smoothing as our aggregation method but you can choose other aggregation methods as well The run settings for this exercise are defined in set_costds dat file Use the do_cost_ds bat batch file to run the analysis How do the most important areas change compared to the solution from Exercise 5 Compare also the species distribution curves of the two solutions what changes do you see Batch file do_cost_ds bat Rank 1 0 Rank 0 15 Remaining 30 Top 15 Area 37 682 Area 16 543 BL A 0 717 BL A 0 329 Cost 18 737 cells Cost 7 411 cells av prop 36 7 av prop 16 8 min prop 29 9 min prop 11 0 Exercise 7 What about the already existing reserves When it comes to reserve network planning managers seldom have the change to start from a fresh table In many cases the target area already includes older reserves or areas that are ear marked for other land uses such as agriculture forestry or habitation etc Thus one often has to take into consideration areas that either can not be included or need to be included into the reserve network Let us think that our example landscape already has a couple of reserves that we have to include to the final solution To do this we use the mask option which allows us to classify cells to different categories which in turn define the cell removal order In the tutorial package you can find a mask file which includes three reserves mask_rs asc We run the analyse
151. n v 1 0 and v 2 0 with the other cell removal rules has the following parameters B Species list file spp Notepad speciel asc speciel asc specie3 asc specied asc specie5 asc Picture of species list file 1 Species weight wj in the Zonation algorithm If no weights are used this should be set to 1 0 In the example above all species have equal weights 1 0 but weights could go from zero to any positive value No negative value can be used as species weight Using O as a weight means that the program calculates the performance of the species during cell removal but that the species does not in any way influence the cell removal order Thus other species are acting as surrogates for weight zero species Note that species specific weights have no influence on the analysis if you have chosen the target based benefit function as your cell removal rule This is because all species have given targets when using this cell removal rule thus weights have no influence on the outcome The weight parameter can include considerations such as degree of historical distribution loss species local or global scale priority taxonomic uniqueness etc All such considerations should be aggregated to a single weight reflecting the importance of the species biodiversity feature in the analysis Some notes about weighting i Weighting is a political decision and there is no general method for determining correct weights ii Weights sh
152. nal value is calculated does not change from that of additive benefit function Also with this cell removal rule species occurrences are considered as additive and the cell that has the lowest marginal value summed across all species will be removed next Also when using target based planning the species specific weights have no function as the goal is to retain a given proportion of distributions for all of the species However it is possible to set different targets to different species It is also recommended to avoid using very high warp factors to allow the program to find the most optimal solution near the targets General differences between cell removal rules It is important to realize that there may be significant differences between different cell removal rule solutions and that the most preferable solution method depends on the goals of planning Thus different cell removal rules may be conceptually better suited for different situations e Core area Zonation is most appropriate when there is a i definite set of species all of which are to be protected tradeoffs between species are discouraged ii the hierarchy of solutions and easy weighting of species is desired and iii importance is given to core areas locations with highest occurrence levels occurrences in cells are not additive meaning that twenty locations with p 0 05 is not the same as one location with p 1 0 e The additive benefit function formulation may be more
153. nalysis Weldhts Meat Sedan O ada 63 Boundary quality penalty definitions file ss 63 Directed connectivity description Ml a a 64 Species interactions definition file oooocccoconcconnnoconononnncnncnnononononanononcnnnnnnnnnnnnnncnnononnnonannnancnnonanins 65 A Standard ZOMAtiOn Output sente are 67 Visual OAC UU E ance dra dns ei across ete oceano 67 Automated fe output atasca ii a E 73 5 Main analys s cintia ta dices 77 Basic Zonation and species weighting iss isiiaiinsa addict diam amende te nee te tee nnmnnn nnmnnn eene 77 Distribution smoothing Zonation ss cenar aa an AAAA SE Oa aaa RaO nana na maneras 79 BOP ZONatoON cia A A IA asias 81 Directed connectivity asnicar ees Seneca cee Ann iia atacadas 83 Including distributional uncertainty Distribution diSCOUNtINY ccccccccccnnnnnnnnnnancancanaanancncnnnnnncnncnnnncnnennnnns 86 Species interaction Sarna ui ici oido id 89 Inclusion Exclusion cost analyses using Mask files ins sirsrnrrrrsssnnresssnnenessnnes 91 6 Post processing analyses amp options nn 95 Landscape identification dis eee E E EA 95 Statistics for management landscapes 2 22 nn nn nee rene ana one as sn te a mi ans aaa aana aani 97 Solution COMPARISON A cis 99 Fragmentation uncertainty anlysiS iii id 100 Solution cross comparison using solution loading ss 102 ZAG Sumi U Ci caca E 103 Obtaining species specific information about solution quality
154. ndow and give the name of your species list file 4 Select the suitable cell removal rule 5 Give the name of your species list file 6 Define the name of your output files 7 Press Run button to initiate the computation 8 For additional settings and analysis see sections 3 3 2 3 Run settings file and 3 5 Main analyses Features you might consider changing include the warp factor the uncertainty analyses and the aggregation methods 9 Note that using the windows interface is the secondary mode for running Zonation Typically runs would be started by calling batch files which are simple text files that include a Zonation call Such runs are easily repeatable and less prone to parameter input errors than the use of the interactive interface directly 2004 2008 Atte Moilanen Zonation User manual 4 Z1G2 Iterative cell removal done Showing removal rank Maps Run settings Species info Memo Landscape identification Solution comparison About Run mode ee fie ket iain Gl x ecies file list 1s in file f Calculate new solution E C Load old solution rank file Removal rule Start of output file name Original core area Zonation ES z Use cost file Additive benefit function Use incl excl mask Use SSI file Target based planning f Generalized benefit function Use planning unit layer Use interactions file 1100 0 Z 0 250 Warp factor Res
155. ne represents one solution spatial reserve structure An increasing robustness requirement a implies that a decreasing biological value can be achieved reliably The thick line is the Pareto optimal boundary representing solutions that are optimal in the sense that increased biological value can only be obtained with the cost of lowered robustness and vice versa When doing reserve selection on a large grid there is a huge number of potential reserve structures but only one or few of them would correspond to the Pareto optimal boundary at any given a level and resource fraction of landscape Any solution not at the Pareto optimal boundary is inferior in the sense that another solution exists with either higher biological value or higher robustness or both The distribution discounting technique used inside Zonation automatically identifies the robust optimal nested Zonation set of solutions for the given level of a 2004 2008 Atte Moilanen 35 Zonation User manual 2 5 2 Uncertainty in the effects of landscape fragmentation This section is mainly based on Moilanen and Wintle 2006 Instructions of how to use fragmentation uncertainty analysis in Zonation can be found in section 3 6 4 This analysis is for investigating how robust a given reserve network is to potential negative effects of fragmentation This is a minor analysis that would not be relevant for most applications of Zonation essentially it is intend
156. nectivity interactions in Zonation see Rayfield Moilanen Fortin submitted There are four distributions of interest the first two of which are not interaction layers i the distribution of the resource prey host etc and ii the distribution of the consumer predator parasitoid The two other distributions are interaction distributions First iii the resource use intensity layer is defined as resource use intensity local occurrence level of resource x con to distribution of consumer and iv turning the interaction around one gets the resource connectivity layer defined as resource connectivity local occurrence level of consumer x con to distribution of resource In general the the interaction distribution layer is defined as local quality of distribution 1 x metapopulation connectivity to distribution 2 where the computations are done for each grid cell Resource use Resource intensity connectivity 1 iv 11 iii Following this mark by r and c the local abundances of resource j and consumer k in grid cell i respectively Let 6 be the parameter modeling the spatial scale of foraging for consumer k B is the parameter of a negative exponential function We specify that the resource use intensity of resource j at cell by consumer k is R 2004 2008 Atte Moilanen Methods amp algorithms 40 gt eap Ad ku Hem l R min 41 0 _ min 41 0 E Y max
157. ning map 67 replacement cost analysis 37 61 91 reverse heuristic 16 richness 73 richness map 6 7 robustness 31 2004 2008 Atte Moilanen Index 130 run info file 73 run settings 53 running basic Zonation 77 running the program 44 45 46 S scale of landscape use 50 selection frequency analysis 103 sensitivity analysis 6 settings window 45 simulated annealing 5 solution comparison 99 solution cross comparison 48 95 102 solution cross load 95 102 solution loading 48 102 species distribution map files 49 species info window 67 species interactions 39 53 65 89 species list file 50 species area curve 18 67 species cell specific errors 62 species info window 67 SSI species 53 SSI species list file 58 statistical distribution modelling 2 step function 20 stochastic optimization 5 surrogate species 77 system requirements 107 T target based planning 1 7 20 testing old solutions with new settings 102 tree connectivity file 64 tree hierarchy 64 troubleshooting 108 tutorial 113 U uncertainty analysis 31 35 62 86 100 uncertainty analysis weights file 63 uncertainty model 31 uncertainty parameter 31 86 uniform error 53 upstream connectivity 28 using command prompt 110 _V vector data 42 version 2 3 visual output 67 W warp factor 53 weighted range size corrected richness 73 weighting 77 weights 50 77 weights in uncertainty analysis 63 w
158. nnnnnnnennnananrrnnnnnernnnarnan 24 Distribution smoothing sitos ia ri a 25 Boundary Quality Penalty BOP iio diodo dali 25 Boundary Length Penalty BLP pioetan A ean A A Aa aiaia aaa 28 Directed Connectivity NOP coincida cir 28 5 Uncertainty analysis nine ne A ce an agen de ee de a 31 Uncertainty in species distributions distribution discounting cccooccccconnnncnononnncnnnnnnnnnnnnannnnnnnnnnnanannnnnnnnnnnas 31 Uncertainty in the effects of landscape fragmentation RS 35 6 Replacementcost a alvSiS cosa 37 T Species interactions sica ci 39 8 Assumptions amp limitations iia e ad a 42 Part II ZIG The Zonation software 44 TIRTOduUcuion ES ns RS a A UN 44 2 R NHINO ZONAtION 0 isis 44 Command prompt ee NU ns ce hd nan een 44 WINCOWS interfaC sanador Et 45 BAC a capability ere a oot eet tee nn nee 46 Loading previously calculated Zonation solutions ss 48 SIN PUCTIOS amp Settings snas a nent ant nn Re tienne 49 INntrOdUCION sit A al 49 COMPDUIS ON HIS sia litio A de nd a tien 49 Species CIS MBUTON MAD MMS amener E EET dd A non riens leds 49 SPECIES tao iantada 50 O 2004 2008 Atte Moilanen Contents Il RUN settings Mens teen nt sidi todo adn 53 Optional files A a A de CR ade 58 SMS AN MN ES ae eo 58 Planning AMY a ia 59 COSUIAY Cli a area 60 Removal A SR RS PR Re a en one ee Rene 61 Distributional uncertainty map year bien dns tienne 62 Uncertainty a
159. ns of SSI species will be retained far into the cell removal process especially if there are relatively few locations with observations of these species To include SSI species into the analyses you need two kinds of input files 1 SSI species list file 2 Species specific coordinate files SSI species list file has an identical structure to the ordinary species list file However it should be understood that the columns for the dispersal alpha and BQP parameters contain dummy values whatever is entered there will not influence computations Thus the relevant columns for a SSI species are i the first one defining the weight of the SSI species and ii the last numeric column O 2004 2008 Atte Moilanen 99 Zonation User manual 3 3 3 2 which gives the parameter either for additive benefit function or for targeting analysis B SSI sp_list spp Notepad File Edit Format View gt 1 0 1 1 0 25 SSI speciesi 551_specles 2 551_species3 O 0 L 0 551_speclies4 O 1 1 y 551_species5 Picture of SSI species list file The second set of files are the species specific coordinate files one for each species which give the exact locations for each observation The given coordinates must fall within the area of the maps loaded for map species as defined in the species distribution map files for ordinary map species or otherwise an error will be reported Here the final column is the info gap
160. nteractions In most cases conservation planning is done purely based on occurrence data either of species or other biological features such as vegetation types etc However sometimes more specific information is needed for example if we want to include interactions between species and e g their food resources predators or competing species into conservation planning process With Zonation v 2 0 this type of specific planning is possible as the software includes a facility for modeling a variety of species interactions In this section we describe the method and philosophy behind this facility and give some possibilities of cases where this type of analysis would be useful Instructions for how to include species interactions into Zonation can be found in section 3 5 6 1 Interaction type 1 resource consumer predator prey host parasitoid present to future distribution etc type interactions The general idea is that one wishes to i protect the resource independently ii protect a part of the resource distribution which is available to the consumer iii emphasize protection of the consumer at areas that are within foraging distances from the resource Item i can be achieved simply by entering the resource as an independent layer into the Zonation analysis Items ii and ili are linked and they can be implemented via application of the distribution smoothing technique The figure below illustrates the principle of positive con
161. ocations that are close to well connected to the distribution of the invasive species or source of pollution 4 Modeling of food chains or food webs The interactions can be calculated between multiple species in the analysis multiple resources can be connected to one consumer etc By appropriately chaining connectivity effects between distributions it should be possible to model more complicated relationships than just simple pair wise interactions When Zonation is started the species and uncertainty layers are read in one by one For each distribution discounting is done first and then distribution smoothing if used Interactions are implemented after all species layers have been read and discounted smoothed The interactions are read in one by one and performed between the layers specified in the interaction definitions file Each interaction transforms one of the species layers that were read in This means that if for example a habitat quality layer a connectivity layer and interaction layer are to be calculated and used based on the distribution map of one species then the distribution map needs to be entered three separate times into Zonation once as a plain unsmoothed habitat quality layer a second time smoothed to implement the connectivity calculation and a third time to be used in the interaction as a focal layer Note also that a species map may be transformed by more than one interaction If the same layer is the focal
162. of them listing the exact coordinates of the records for one species SSI_sp8 txt and SSI_sp9 txt We have also defined all the species specific parameters in an additional SSI species list file SSI_list txt 2004 2008 Atte Moilanen 119 Zonation User manual i SSI species 1 ii SSI species 2 Two maps showing the point occurrence data of our two SSI species respectively The analysis is run as in Exercise 2 except that now we need to activate the SSI species option and give the name of our SSI species list file You can either do this manually in windows interface or run the program from command prompt using the modified settings file set_SSl dat You can call the program yourself or use the do_ssi bat file to run the analysis As you can see the inclusion of the two extra species brings hardly any changes to the results This is because the points cover only a very small fraction of the study area and it is therefore easy for the program to include them to the top fraction without altering the spatial distribution of high value cells in our solution Also because the area covered by the points is so small the cells with SSI species receive very high values Thus it follows that the full distribution ie all the points of SSI species are practically always included to the top fraction This is evident also from the SSI species graph in the species info window from which we can see that the entire dis
163. ogram will not overwrite the solution from the previous exercise Batch file do_w bat Rank 1 0 Rank 0 15 Remaining 30 Top 15 Area 29 882 Area 16 543 BL A 0 648 BL A 0 932 Cost 29 882 cells Cost 16 543 cells av prop 36 6 av prop 20 6 min prop 29 9 min prop 14 6 As you see the weighting of species alters the spatial distribution of the highest value cells Here more importance has been given to the areas where the two endemic species occur e g west and south coast and less to those areas that have a high representation for the other species e g the peninsula on the east and the north east region A difference between the basic and weighted Zonation runs can be seen also in the species distributions curves in the Species info window If you compare the two solutions you can find that the relationship between landscape loss and average biodiversity protection for all species is quite similar in both cases blue line However when species 2 and 3 are weighted they retain a relatively higher proportion of their distributions through the cell removal process compared to the basic Zonation run where no weights are used In turn the minimum proportion retained species that has the lowest protection is smaller when using species weighting The two graphs below show the differences for species 2 black line between basic and weighted solution O 2004 2008 Atte Moilanen Tutorial amp
164. om it which is the component of the equation having the summation across neighbors k upriver from focal site k ENP Similarly loss of unit influences the upwards connectivity of units downriver from it Note that the present version of Zonation uses species and unit specific predictions of occurrence p and species specific connectivity responses h However connectivity up and down is based directly on the numbers of grid cells in planning units in a non species specific manner meaning that Fis Oj and A are taken as the same for all species which assumption might be relaxed in a later version of Zonation Equation 1 is simply the fraction of distribution lost for one species which does not account for how lost representation is translated to lost in conservation value When deciding which cell can be removed with smallest loss of conservation value O is aggregated across species according to the cell removal rule which now is for core area Zonation l w mar i C j Q a 2004 2008 Atte Moilanen Methods amp algorithms 30 and for the additive benefit function Zonation 5 Lw Sfr o s 1 0 5 8 aa 2b in which w is weight of the species j c is cost or area of planning unit and q is the fraction of distribution of species j remaining before removal of cell and V is the function translating increasing representation into increasing conservation value se
165. on target of protecting 10 of all species occurrences this can be done by typing the target proportion here 0 1 in to the Remaining field and pressing the Remaining button Note that using this option changes the coloring of the map Here the red color does not present the best 2 of the landscape but those areas that are needed to protect at least 2 of all species occurrences and so on lt is equally important to understand that in this procedure the program displays areas that achieve at least the given target for all species simultaneously Thus some species may have much higher proportions of their ditribution remaining in the solution ply Here in gradual colors is shown the area required for representing at least 30 of distribution of all species In black are those areas which are not included in the solution When a certain proportion of the landscape is selected based on either ranking or remaining the program displays a small information window about the selected top fraction This window includes the following information 2004 2008 Atte Moilanen Zonation User manual ZONATION Selection data BL 11574 Area 11029 BL A 1 049 Cost 11029 000000 minprop em 0 122787 mean prop_rem 0 140026 weighted_mean_prop_rem 0 140026 average 90 cell count 6219 50 cell count 2608 BL Boundary length of the selected area Area Count of cells in selected area BL A Boundary length Area r
166. ons with several a values to determine how the spatial pattern behaves with increasing uncertainty a can be either zero no uncertainty or any positive value 2 Define in the run settings file whether errors in species occurrences are uniform value 0 or proportional value 1 Uniform error is the default setting and works for most of the data sets but in some cases it is more appropriate to use proportional errors see e g Ben Haim 2001 3 Define whether uncertainty map layers containing species cell specific errors Wsc are used value 1 or not value 0 in your run settings file 4 If distribution uncertainty map layers are used type the name of your uncertainty analysis weights file in to the run settings file E g Info gap weights file UCweights spp 2004 2008 Atte Moilanen ZIG The Zonation software 88 In the windows interface 1 Select the distribution discounting option Give a value to the uncertainty parameter a oS Select whether you wish to use uniform or proportional errors Select the List of error weight map layers option and give the name of your uncertainty analysis weights file if you wish to include the uncertainty map layers into the analysis 4 11G2 Iterative cell removal done Showing removal rank Maps Run settings Species info Memo Landscape identification Solution comparison About Run mode w GAG A oS Pa E Species file list is in file
167. onse curve First the radius The effect of habitat loss can be localized the species only has a narrow edge effect or so or the effects can extend over a longer distance as could be the case with a timid larger animal that avoids human proximity habitat loss could negatively influence such a species from a long distance away By setting a species specific radius one can model how close habitat loss influences the species Note that inside the radius only the loss of those cells that have data on the particular species in other words cells that are not marked as missing data can influence the value of the focal cell 1 component Radius size Large radius Small radius Focal cell 2 component Response curve 10 1 0 E 08 E 0 8 Strong response to neighboring habitat loss S 06 5 06 D Q 2 04 Low response to 04 gt neighboring habitat loss Ls 02 S 02 Oo o 0 0 0 0 00 02 04 06 08 1 0 00 02 04 06 08 10 Proportion of neighboring cells lost Proportion of neighboring cells lost Figure clarifying the two components used when specifying the BQP for a species O 2004 2008 Atte Moilanen 27 Zonation User manual The second component the response curve specifies the kind of effect neighborhood habitat loss has on the species First there could be an absence of effect which would be modeled by a flat line no effect Then the species could suffer variable degrees of loss in local
168. op the computation and try again after closing all other programs to save memory or with a smaller data set or with a computer that has more memory See section 3 8 for more detailed information related to computer memory capacity It is not ok to have Zonation running using virtual memory the hard drive because that will simply run too slow Operating system Zonation v 2 0 is Windows 32bit software which should be operational for example on Windows XP and 2000 Check the memo Some warnings or error messages appear in the memo Read through the text to check for any information that might give a clue to solving your problem a values Check the a values for any errors Remember that these values have to be in same unit of length as the cell size given in the species distribution map file It is very easy to get these values wrong at first calculation be sure to verify computations Decimal points and commas Always remember to use only decimal points NO commas Zonation assumes decimal dots and commas will result is undefined errors Empty rows at the end of your input files Check that you do not have any empty rows at the end of your input files These may cause some unexpected software behaviour Differences in grid sizes cell sizes All raster files should have the same grid size This means that in all files the number of columns and rows as well as the size of cells should be equal You can check these information from t
169. orizon of uncertainty 31 86 host parasitoid interaction 39 including areas 61 including areas to the solution 3 7 91 inclusion cost 37 information gap theory 31 35 initial removal 53 input data 3 input file 58 input files 49 58 installation 8 integer programming 4 interactions 53 Invasive species 39 Ls land cost 60 landscape identification 95 landscape statistics 97 limitations 4 42 107 litterature cited 15 loading old solutions 48 102 logit space 53 loss curve 25 28 loss curves 63 loss function 25 management landscapes 95 97 map colors 6 7 map layout 67 map window 67 maps window 67 marginal loss 16 17 mask layer 53 61 memo window 67 memo window 67 memory use 107 metapopulation connectivity 25 Metapopulation Research Group 8 MRG 8 multiple batches 46 N name annotation 53 neighborhood quality penalty 24 28 new features 11 nominal model 31 NQP 24 28 59 63 64 83 O objective function 4 opportunity 31 optimality 4 5 optional input files 58 output 67 output files 73 overlap in landscapes 99 P Pareto optimal boundary 31 planning units 24 53 59 point distribution data 58 pollution 39 post processing analyses 95 power function 18 predator prey interaction 39 printing options 67 proportional error 53 proportional selection 67 Q quick start 8 R radius 25 rank file 73 ranking map 67 rarity map 67 raster map 49 references 15 18 remai
170. ot care how the nominal estimates and associated error measures are obtained any statistical method or expert evaluation can be used as a basis for developing those quantities 2004 2008 Atte Moilanen Methods amp algorithms 34 But how to determine the horizon of uncertainty a As mentioned above the relative error wsc can be any error measure related to the predicted species distribution lt can for example be a statistical error e g the length of the lower half of the 95 confidence interval or a probability of future anthropogenic threat or both The value of a on the other hand is unknown and has no correct value The way forward is to investigate how different reserve network candidates perform under increasing uncertainty In practise the way is to try out several levels of a to see what areas are selected If an area is always selected irrespective of the value of a then the area is important for sure If the area is selected with low a but not with high alpha then the area is selected because of the occurrence of an uncertain biological feature If the area is not selected with low a but is selected with high alpha then something of interest occurs in the area with relatively low density but high certainty Based on such uncertainty analysis one can split the landscape in three categories of areas i areas that are good for sure ii areas that are never selected and thus have low priority and ili area
171. ot move to the next run The description so far cover the basic use of batch files but another more complicated example follows below If you want to alter more than one parameter simultaneously then one way forward is to use multiple nested batch files For instance you might want to run the above batch do_zig2 bat with different species weights defined in separate species list files To do this we first create a new file called myruns bat and then adjust the settings from our example above E myruns bat txt Notepad File Edit Format View Help call zig2 r settingsi dat 1 outSl_ 2 txt 0 call zig2 r settings2 dat 1 outs2_ 2 txt 0 call zig2 r settings3 dat 1 outs3_ 2 txt 0 Then we create another batch file which in turn calls the my runs batch BP do _myruns bat txt Notepad Ela fx File Edit Format View Help call myruns speciesl spp spl call myruns species2 spp sp2 call myruns species3 spp sp3 Here the first parameter after my runs defines which species list file is used when running the my runs batch and the second parameter sp1 sp2 sp3 defines a part of the output file name to distinguish which species file has been used in each run In my runs batch file these parameters will be referred as 1 and 2 When running the do my runs batch the program first calculates solutions using species1 spp file with three different settings and giving each of the output files an ending sp1 outS1_sp1 outS
172. ou are using target based planning as your cell removal rule each species has a defined target e g 25 of original distribution which the program seeks to retain during the removal of landscape However as all cells will be eventually removed these targets will be inevitably violated Thus in curves txt file the program simply reports when the targets of particular species have been violated e what fraction of landscape was still remaining when the proportion of species original distribution dropped below the given target If target based planning is not used as a cell removal rule this column has only dummy values After the list you can find columns representing more detailed information of how large proportion of each species distribution is remaining when landscape is iteratively removed P output curves txt Notepad File Edit Format View Help List of species and weights used in analysis in order of columns distribution sum TviolationFractRem MapFileName weight 11 000 000 000 5 Prop_landscape_lost 3280 76 0 29142 O 33307 7 1 0 33116 94 0 2832 80 0 0 36691 1 34740 27 1763 0520 0562 speciesi species2 species3 speciesd speciess species species asc asc asc asi asc ase asc pp target violated at fraction 0 823679 of landscape removed 0 176321 remaining min_prop_rem cost_needed_for_top_fraction ave_prop_ rem ext 1 ext 2 prop for each
173. ould still be used If weights are implicitly taken as equal then that is a weighting also In analysis a species with say unique taxonomic history should have higher weight than a species that has 200 rather similar sister species iii Increasing the weight of one feature species will increase the fractional representation levels for that O 2004 2008 Atte Moilanen 51 Zonation User manual feature But this positive effect comes with the cost of small fractional losses for some or all of the other features The a value of species specific scale of landscape use parameter of negative exponential This parameter is only needed when you are using distribution smoothing as a part of your analyses if not used a dummy 1 0 will do fine The value indicates how species use the surrounding landscape and can be calculated based on for example the dispersal capability or the home range sizes of the species The a value can be calculated as 2 Cell size in km Use of landscape km Input cell size E g if the known guesstimated mean dispersal capability of a species A is 3 km and the cell size in species distribution files is 1 0 km note not 1000 m then the value of a for species A is 0 67 2 1 a 3541 0667 The last part of the equation Cell size in km Input cell size is needed to keep the a value in same unit of length as the cell size given in the species distri
174. ould you prefer to lose the cell from A If you have two otherwise identical species but one has a larger range remaining then you prefer to lose from the species that has the larger range B If you have two otherwise equal species but one has relatively higher weight then you prefer to lose from the distribution of the species with a lower weight C You have two presently equal species with equally wide distributions Then you prefer to lose from the species that has had a smaller historical reduction in the range dashed line D Within the distribution of a species one prefers to lose from a location with a relatively low occurrence density light gray Additive benefit function This section is mainly based on Moilanen 2007 Compared to core area Zonation the additive benefit function takes into account all species proportions in a given cell instead of the one species that has the highest value The program calculates first the loss of representation for each species as cell i is removed and the di value of the cell is simply the sum over species specific declines in value following the loss of cell O 2004 2008 Atte Moilanen 19 Zonation User manual l l 2 AP wi gt PONS F Oi 5 in which Q S is the representation of species in remaining set of sites S and Q S indicates what remains after cell has been removed Here w is the weight of the species j and c is the co
175. ously Thus you do not need to change the parameters manually in the species list file after each run If you do not wish to multiply the a values set this factor to 1 O 2004 2008 Atte Moilanen ZIG The Zonation software 80 Es es Command Prompt Cirscd Zonat ion C onationcall zig2 r settings dat sp_list spp output txt 6 6 1 1 0 A In the windows interface 1 Select the distribution smoothing option 2 Give a factor for multiplying the species specific a values 4 11G2 Iterative cell removal done Showing removal rank Seles Maps Run settings Species info Memo Landscape identification Solution comparison About A my_splist spp transform species occupancy probabilities from logit values loutput_DS dat Annotate outputfile name Run mode A Ce a P Species file list is in file Calculate new solution C Load old solution rank file KE Removal rule Start of output file name fe Original core area Zonation Use cost file Additive benefit function cost asc Use incl excl mask Use SSI file mask ras asc ISSI_list bet IPLU_file asc C Target based planning C Generalized benefit function Use planning unit layer Use interactions file interact_ spp Warp factor 100 0 z 10 250 Settings for generating spatial aggregatiop thto the solution Resampled species count NM Only remove from edges Info gap settings M Use dis
176. oval done Showing removal rank COX Fi Maps Run se ings Species info Hemo Landscape identification Solutiog comparison About Comparison of solution overlap Compare 0 2 fraction of present solution to 0 2 fraction of the solution in file old_sol rank_asc Compare solutions and output to loverlap1 ras asc W Show overlap on map Hote fractions refer to a proportion of landscape on top of intial removal Color key for comparison map rellow m both solutions Light green in present solution only Dark green in older loaded solution only Output The program produces a map showing the results of the comparison Here all overlapping areas are shown in yellow where as the light green areas are only included in the present solution and the dark green areas are only included in the older solution The rest of the landscape not included in the selected top fraction is colored as blue Note that this map is not automatically saved If you wish to save it double click on the image 2004 2008 Atte Moilanen ZIG The Zonation software 100 3 6 4 Picture of the output map More detailed results can be found in the Memo window Here the settings of your analysis and the similarity of the two solutions f1 and f2 is shown Average order diff is the average difference in removal rank order between the two solutions The Solution comparison also produces a basic raster output file
177. pe Cells that include a part of the distribution of a valuable species high weight remain later in the iterative cell removal process than cells only containing low weight species assuming everything else is equal between the occurrences Weighting influences the fraction of a species distribution retained at any point of the cell removal When using weighting high weight species retain a relatively higher proportion of their distribution Also note that the balance in representation levels developed by Zonation is such that narrow range species typically have a larger fraction of their ranges protected compared to initially wide ranging species Note that weights can also be used to test the efficiency of surrogate species This is done by weighting the surrogate species normally e g by 1 and giving a weight of O to all those species that are NOT used as surrogates This way the program will not use the non surrogate species in the selection of the next site to remove but it will monitor the decrease of the distributions of these species as well Thus you can test how well a reserve network selected using surrogates will protect all species Output Here is an example of how species weighting can influence the final solution The pictures show the results of two basic Zonation runs for seven species In the first picture no species weights have been used where as in the second picture one of the species has received a weight of 3 0 20
178. population density if there is neighborhood habitat loss For example you could have a relatively insensitive species which loses half of the population density when the focal cell has lost all it neighbors from inside the species specific buffer Then again the species could be very sensitive to neighborhood fragmentation all local value could be lost when only half of the neighboring cells have been lost And finally a species could even favor fragmentation which would be modeled by a response curve that goes above 1 at some levels of habitat loss For the exact way of analyzing a habitat model to get the species specific response see Moilanen and Wintle 2007 Note also that the BQP curves need not be derived from habitat models One could also use expert opinion to guesstimate the response of the species to neighborhood habitat loss The hypothetical curve would then be entered into Zonation it makes no difference for the Zonation process how the BQP curves were obtained Note that there are major differences between distribution smoothing and the BQP even though both induce aggregation in a species specific manner First from a practical point the BQP is much slower to run as the effect of removing a cell is not only local but extends over the neighborhood area which needs to be accounted for in computations Second the BAP is biologically better justified The BQP definitions can be based directly on species responses in statistica
179. prevents the acquisition of other biologically much more valuable locations At C gt CIk the focal site becomes included in the optimal set and inclusion cost becomes zero Note that the resource here should be understood the proportion of landscape retained in the Zonation The figure also shows an example of how exclusion and inclusion can be expected to behave qualitatively With a small resource lt CE x the exclusion cost a of a site is likely to be zero because the site would not be in the optimal set in any case At a level CE x the site becomes part of the optimal solution With a resource slightly higher than CE it is likely that the exclusion of the site can be compensated with small cost at least if there are many selection units However when the available resource is large sites of less importance are included in the solution and the exclusion of a high quality focal site has a clearly positive cost Inclusion cost b behaves differently When a site that would not belong to the optimal solution is included in the network it generates an increase of cost even when the resource available is small because the resource is spent on suboptimal areas With increasing resource availability the inclusion cost gradually decreases At a level C k the site would already become part of the optimal solution and inclusion cost thus becomes zero O 2004 2008 Atte Moilanen 39 Zonation User manual 2 7 Species i
180. put map for the species i CAZ Histogram of local occurrence levels in cut ii ABF The habitat quality histogram can be used for example for investigating the difference between i core area Zonation CAZ and ii the additive benefit function ABF analysis it can be expected that core area Zonation produces solutions that include relatively more of the areas with highest habitat quality In contrast the additive formulation allows the representation of the species to be 2004 2008 Atte Moilanen 73 Zonation User manual 3 4 2 summed from many areas with middle to low occurrence levels This difference is illustrated in the picture below where the species specific habitat quality in top fraction for one species changes substantially when switching the cell removal rule from CAZ i to ABF lii Note the different scaling in histograms Memo window The Memo keeps track of what the program is doing or has been doing Also certain results of analyses and error messages are printed here Check the memo for information about options used and about any warnings or errors that occurred during the run the present Zonation version does not stop execution at every suspicious input 7162 Iterative cell removal done Showing removal rank Maps Run settings Species info Memo Landscape identification Solution comparison About 110280 BE Miscellaneous information about the prog
181. r J Lt oS Go co ae H A CMA O oo fo co ao oo oo Lae J fn fi planning un A AAA A A e LA A 1J bd Go LA bent jam co lab GO Es Bo Las E m ima l GA d o t co co C gt g 4G MO Go 4 5 aia La bad i O o O O t 1 co LEA LA ca Er co Lt Cr Pi D oo Lo Es AA Go do FTP A pu ji ee CS Go A AS ca a C3 M h Ma LA EA SS CO MO NO ca ES EC Ca Ca a la EA LEA E CA CA co EA O 2004 2008 Atte Moilanen ZIG The Zonation software 102 3 6 5 The results of the analysis will appear in the Memo window The first column shows the value of a which is the info gap uncertainty parameter The a value represents the uncertainty of species occurrence in cell in other words the uncertainty that the species does not occur in a specific cell although our data suggests that it does The higher the a value the higher the uncertainty In the picture above an a value of 0 000 on the first row indicates that the observed occurrence of a species in a given cell is trusted to be completely accurate The increasing a results in rapid potential loss of biological value in target cells as neighbourhood habitats are lost The following three columns indicate for any given a the loss of biological value in the focal cell when 87 5 50 or 0 of the neighbourhood habitat is remaining respectively The fifth column shows how large proportions of species distribut
182. r matrix This is due to computational reasons and the lack of these no data rows leads to a situation where the program automatically transforms the values on edge rows to missing data This in turn may alter species distribution information which possibility one should be aware of Missing data in the species distribution rasters do not necessarily need to be congruent between all species the program will run if cells marked as no data for one species has values for other species However if missing data is not aligned there are implications for the use of the BQP Remember also to use decimal points not commas in all the input files B Speciel asc Notepad Ea ix File Edit Format View Help incols 649 Nr ows 555 xllcorner 294205 0283004 6 00 NODATA_walue 9999 9999 9999 9999 9999 9 9999 9999 9999 1 1 1 999 111111111131 lt Picture of species distribution map file in this case with presence absence information coded as 0 and 1 99 9999 9999 999 999 GO 9999 1 1 1 1 000000011 9999 99 9 9999 0110 EIA E F 1001 Species list file A spp file containing a list of all species distribution map files asc raster files which will be used in your analysis Each species file is on a separate row This file always contains six columns except when using generalized benefit function as a cell removal rule when there are nine columns only in version 2 0 The six column version used by Zonatio
183. r six species This conclusion stays the same even when using presence absence data Generalized benefit function Cell removal rule number four is a generalized benefit function form that allows very flexible function shapes The function is defined in two pieces each a power function R w if R lt T j V R w w 2 117 lt R lt l O 2004 2008 Atte Moilanen 23 Zonation User manual In this equation R is the fractional representation level of the species the fraction of the original distribution remaining 7 is a nominal target level for the species at R T the representation of the species gets value w4 x is the parameter of the first part of the power function When T lt R lt 1 the function continues as another power function with parameter y and at R 1 the representation of the species gets value w w Thus by giving different values to parameters you have practically an endless number of options for the shape of the benefit function Note however that the definition of how marginal value is calculated does not change from that of additive benefit function Also with this cell removal rule species occurrences are considered as additive and the cell that has the lowest marginal value summed across all species will be removed next Some shapes that generalized benefit function can assume are listed in the following table JD ji 0 DA 0 aa 0 1 0 1 iii co gt
184. ras asc file ras asc file The values in the matrix are as follows 0 The cell is not included in the top fraction 1 The cell is included in both solutions overlapping areas marked as yellow 2 The cell is included only in the present solution light green areas 3 The cell is included only in the older solution dark green areas Remember that this file as any of the ASCII files produced with Zonation can be imported to GIS programs However when importing this file select integers as the format of your cell values Fragmentation uncertainty anlysis The theory behind fragmentation uncertainty analysis is explained in section 2 5 2 This analysis is also described in Moilanen amp Wintle 2005 Biological Conservation 129 428 430 It has been widely argued that habitat fragmentation is bad for population persistence and that a high level of fragmentation is a similarly undesirable characteristic for a reserve network In this analysis it is assumed that cells deep inside a reserve will not be influenced by habitat loss around the edges of the reserve Cells close to the edge on the other hand may lose some of their biological value due to known and unknown negative effects of nearby habitat loss and fragmentation The higher the proportion of habitat lost around the cell the less certain one would be of the remaining conservation value Fragmentation uncertainty analysis provides a method to see how different conservation network
185. re needed every time you run Zonation 3 3 2 1 Species distribution map files A standard GIS raster file asc file of species distribution one file per each species These files need to include the standard GIS raster fill header ncols Number of columns nrows Number of rows xllcorner X coordinates of the low left corner yllcorner Y coordinates of the low left corner cellsize Cell size used in the raster file NODATA value Definition of no data values In example files no data has either value 1 or 9999 After these rows comes the species distribution matrix Each value in the matrix describes species occurrence in a specific cell Values can be of any form of data e g probability of occurrence presence absence data number of population etc The data does not have to be the same for all species i e you can simultaneously input different data types for different species Note that value O in the matrix indicates that the species does not occur in the cell with certainty whereas lack of data must be marked as 1 or 9999 or with a similar value indicating no data 2004 2008 Atte Moilanen ZIG The Zonation software 50 3 3 2 2 It is important that all species distribution rasters have the same grid size This means that in all files the number of columns and rows as well as the size of cells should be equal It is equally important to have at least one row of no data on each edge of the raste
186. re used in this first exercise the last three parameters are given a dummy value of 1 In your first run settings file set dat you can find all necessary parameters to run the basic core area Zonation Here we use edge removal but you can also try to run this analysis without this option to see what kinds of effects it has to the result The warp factor here 100 has been adjusted so that the analysis runs smoothly but with fair accuracy You can also test different warp factors to see how much of a difference it makes to the outcome and the running time If you are operating the program from the windows interface remember to adjust the settings identical to those found in set dat file To run the analysis you can call the program yourself or just use the batch file do_zig2 bat that comes together with the installation package You can run the batch file from Windows by double clicking it If you are using Zonation windows interface just press the Run button Note that the batch file is a text file that can be edited using notepad The outcome of this analysis should look like this Batch file do_zig2 bat Rank 1 0 As you remember the biologically most valuable areas are shown here as red Since we have the restriction of 15 on the area that we can protect we now have to find those areas that compile the best 15 of the landscape Thus enter the target value 0 15 into the Rank field and press the button The program will now sho
187. ress of analyses nl Startins Zonation Tun SSI spp count ee a E a EEE rs sms For conditions of use of this software ses 232 the disclaimer in the sbout be in ESSA 10 102 apogu A A A A A A EE A A E A HH matrix x dimension A A Sk nr ee ee A aR ge eee ey carne D MST mmm QUE Le Zons on lutoral pus Cepa Low count in species file 7 species asc 335 rows read in total Xinduis element conti 740073 NOTA SS COUM ee Sum of probs 35280 7626873725 Rd de de Mains dimension 649 pi Fair qr Masraneamen ST AE MONA 202 matrix y dimension cells with species data sino HEREDAR ss REMOVAL RULE Onisinal basic core area Zonation missing data cells FE BW se eta a wrt hee EX EII E NOT USME TET 21100 thine planning units serres NOT using Info gap distribution discounting uncertainty analysis Loading feature aE species data layers asc 353 rows read in total sum of non missine data elements 35280 7626873725 AAG UT Leer IO Picture of Memo window The fields on the right display aggregate information about input data which could be useful for error checking The memo displays info about the progress of the analysis Automated file output In addition to the visual output the program automatically produces six different output files for each run In the command line or in the windows version s Run settings window you have specified the output filename e g
188. rtain species e g rare species of high conservation value or commercially valuable species 4 Type the call for Zonation in the command prompt and press enter to initiate the computation See section 3 2 1 for how to use the command prompt Command Prompt Cinca Zonation CisZonation call zig2 r settings dat sp _list spp output txt 6 6 6 1 6 A Picture of basic Zonation call In the windows interface 1 Go to Run settings window and check that no additional analyses are selected 2 Adjust the settings as described in command prompt above 3 Adjust the species weights in species list file if you wish to prioritize certain species 4 Define the names of your species list file and output file and press Run to initiate the computation O 2004 2008 Atte Moilanen ZIG The Zonation software 78 4 1162 Iterative cell removal done Showing removal rank Maps Run settings Species info Memo Landscape identification Solution comparison About Run mode ee fie ket iain Gl x ecies file list 1s in file f Calculate new solution E C Load old solution rank file Removal rule Start of output file name Original core area Zonation ES z Use cost file Additive benefit function Use incl excl mask Use SSI file Target based planning f Generalized benefit function Use planning unit layer Use interactions file 1100 0 Z 0 250 Warp factor Resampled speci
189. ry low e g 0 00001 Note that here the costs do not need to be measured in terms of money you can use other measures of economical loss as well For example in economical fisheries the fishing intensity of a landscape can be used as a cost layer the higher the fishing intensity the higher is the cost of protecting the particular site The cost layer is an optional file the program can be run without cost data as well If land costs are not included in the analysis all cells implicitly have an equal cost value of 1 2004 2008 Atte Moilanen 61 Zonation User manual 3 3 3 4 P costs asc Notepad e Formal Vew ncols 649 nr ows 555 xllcorner 294205 yllcorner 6283604 6 cellsize 200 NODATA_value 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9999 9 9999 9999 9999 200 200 200 200 200 200 200 200 200 9999 200 200 300 300 300 300 300 3 999 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 300 300 450 450 450 450 El NT gt Picture of cost layer file It is important that the cost layer raster has the same grid size as the species distribution map rasters This means that in all files the number of columns and rows as well as the size of cells should be equal It is equally important that all those cells which have data of any of the species used in the analysis that is to say the cells that are NOT marked as no data in species d
190. ry quality penalty BQPcurves txt BQP definitions file name BQP mode BQP aligns missing data C Potential habitat is species specific MW Use directed NQP PLU connectivity load from file my_tree txt Boundary Length Penalty 0 000 n Additional edge points Output As NQP uses directed connectivity and planning units instead of singular grid cells the output of this analysis should look significantly different in comparison to basic runs The aggregation of areas depends on the average size of your planning units large planning units lead to a high level of aggregation where as a small planning unit size gives more fine graded solutions nd directed connectivity have been included in the analysis Picture of output map when planning units a Note that the use of large planning units will automatically cause at least an apparent decrease in the quality of results The reason for this is that large planning units will probably contain both areas that are good and bad for conservation Consequently the performance curves will suggest lower protection levels than what can be obtained if selection is based on individual grid cells However this effect can be counteracted by the relatively high levels of connectivity afforded by the use of the planning units With respect to the computation time needed by Zonation use of planning units is likely to cut computation times with larger reductions produced by use of large
191. s 0 Note potential difficulties in interpretation of results if multiple aggregation methods are used simultaneously use tree connectivity Determines whether NQP is used value 1 or not value 0 Unlike the other aggregation methods NQP has a directed connectivity measurement Note that planning units need to be always used together with NQP tree connectivity file Indicates which directed connectivity description file will be used use interactions interaction file annotate name when including NQP into the analysis Determines whether species interactions are included value 1 into the analysis or not value 0 Indicates which interaction definition file will be used With this option you can mark your output file names to show which analyses have been used to produce them value 1 The program will add letters and numbers in the middle of your output file name depending O 2004 2008 Atte Moilanen 97 Zonation User manual logit space on the used analyses CAZ ABF TBF GBF shows whether you have used the basic core area Zonation CAZ the additive benefit function ABF the target based function TBF or generalized benefit function GBF as your cell removal rule M mask used C costs used E edge removal used A edge points added Sxxx distribution smoothing used The following numbers show the factor that has been used to multiply the species specific a
192. s method is qualitative in the sense that the estimated conservation value of individual cells or consequently the conservation value of the entire reserve network is not influenced by the degree of fragmentation but rather aggregation is induced via a penalty given for the boundary length of the reserve When using the boundary length penalty the hierarchy of cell removal is based on both species occurrence levels in cells and the increase decrease of boundary length that results from the removal of a cell The boundary length penalty can in the context of core area Zonation be formulated as 5 ma LV wf Ci where A BL A is the change in boundary length area ratio of the reserve network following removal of cell i and f is a constant defining the strength of the boundary length penalty If cell removal decreases boundary length A BL A receives a negative value and the value of di for cell i decreases indicating that it is relatively advantageous to remove the cell because removing it reduces fragmentation 6 ACBL A AL 0 AL 2 AL 2 Above is a picture showing how different cell removal options would influence the boundary length The boundary length is calculated in the terms of cell edges Removing the gray cell in the first example results in no changes in boundary length for as two edges are removed while another two are gained In the second example the cell removal leads to the loss of one edge but also to the gain of t
193. s once more as in Exercise 5 but this time we include the mask see set_maski dat Thus call the program with a new batch file do_rs bat O 2004 2008 Atte Moilanen 125 Zonation User manual Batch file do rs bat Rank 1 0 Rank 0 15 Remaining 30 Top 15 Area 30 830 Area 16 543 BL A 0 271 BL A 0 369 Cost 30 830 cells Cost 16 543 cells av prop 31 6 av prop 14 4 min prop 29 9 min prop 11 4 Compare the solution to the results from Exercise 5 How has the representation of species altered now that the old reserves are included in the top 15 fraction Note the differences at the end of the species distribution curves As you see the solution received with the old reserves is suboptimal since the program is forced to include poor quality areas in the top fraction Dh 00 O IN NO oO 0 0 1 02 03 04 05 06 07 08 0 9 1 proportion of landscape lost proportion of distributiogs remaining Average and minimum performances when old reserves are included to the top fraction In some cases the situation is the other way around where areas can not be included to the reserve network for some reason Thus these areas need to be first masked out from the landscape before ranking the rest of the area See how this kind of masking would change the results in our study area by running the analysis with the batch file do_towns bat where we have two imaginary r
194. s see Moilanen A 2001 Simulated evolutionary optimization and local search Introduction and application to tree search Cladistics 17 512 525 2004 2008 Atte Moilanen Introduction 6 1 4 Atypical Zonation work flow This section outlines a typical sequence of steps that would be done for the Zonation analysis of one data set 1 Get the basic analyses running 1 1 Install Zonation and get the basic zonation running with the example files provided 1 2 Decide your cell removal rule 1 3 Produce a new settings file species list file etc for your own data and check that you are able to run the basic analysis without aggregation methods or uncertainty analysis 1 4 Try variants of the basic analysis by adding unequal species weights aggregation methods and uncertainty analysis You can use solution comparison to check how big a difference does the addition of one complication cause into the solution These preliminary analyses can as well be run using high warp factors 100 1000 to reduce runtimes ii Identify your base analysis There are endless options of what species weights to give what species specific parameters exactly to use in the aggregation method distribution discounting or boundary quality penalty and what parameter s to use in the uncertainty analysis You cannot run all combinations of everything and indeed it is not useful to do so Therefore after getting the basic Zonation runnin
195. s that are selected with some levels of uncertainty these areas may need further investigation before a choice of conservation priority for them can be fixed Distribution discounting uses the following algorithm for finding robust optimal reserve designs 1 Specify robustness requirement a 2 Distribution discounting Read in species information For every species and cell apply Eq 3 or its analogue depending on type of data uncertainty model 3 Use any reserve selection algorithm here Zonation to search over spatial patterns The robust optimal design X at the given level of a is the one achieving the highest possible conservation value The advantage of this approach is that the worst case probability set Eq 3 has to be calculated only once item 2 and thereafter the contributions of cell to representation levels psc do not change in the reserve selection process Testing several a values allows you to outline how different reserve structures behave in increasing uncertainty Some designs are always bad some are good according to nominal habitat model predictions but bad if uncertainty is incorporated into the models Others have intermediate nominal performance but have a good robustness to uncertainty The robust optimal designs are always at the Pareto optimal boundary with respect to the target as demonstrated below The Pareto optimal boundary Target achieved proportion of populations Here each thin li
196. s that in that particular management landscape the species occurrence is less than 0 01 of their full distribution Note that if any of the species have a larger proportion than 1 of its distribution located in the landscape the program automatically prints a list of those species and the precise proportions of their distributions on the next rows The second part shows you how large proportions of species distributions are remaining in the whole landscape all management landscapes together that was initially included in to the analysis see section 3 6 1 Percentage of landscape The program also automatically calculates an average of these proportions Fz Liout_species txt Notepad as Of x Fils Edit Format Help ge proportion remaining over all spp in networks 0 259218 proportion remaining for species 1 asc oa ASC weak 156 24 45C 20 asc a speciesb asc 23 species asc 24 The third part contains a list of all management landscapes area in number of cells and the proportions of distributions for each species in the respective management landscape The species are listed here in the same order as they are in your species list file Sj Llout_species tut Notepad E 10 x File Edit Format Help Biological data of 32 networks spots 799 Networks x species matrix Nw_number area cells sp_data 328 0 0000 0 0000 O O0DO 8041 0963 0 1576 0 1393 0000 0 0000 O 0000 ar 0016 0 0000 0 0010 20
197. s to the top 10 fraction of the whole landscape Thus this option can be used for setting criteria for the management landscapes Note that if the inclusion minimum is equal to or larger than the percentage of landscape all spatially separate areas will be included in the management landscapes where as if the inclusion O 2004 2008 Atte Moilanen ZIG The Zonation software 96 minimum is smaller then only areas with sufficiently high ranked cells are included 4 21G2 Iterative cell removal done Showing removal rank T E Edd Maps Run settings Species info Memo Landscape identification Solution comparison About a Percentage of landscape 20 e Inclusion minimum 20 Maximum distance cells 0 Maximum difference in species composition 02 Network output file name Llout ras_ase Network species data file Llout_species txt 5 Run landscape identification Cost 1 0 Get LS fraction corresponding to cost 4 5 Finally define the names of your output files to Network output file name and Network species data file fields 6 Note that if you need the selected fraction of landscape to correspond to a certain cost you can easily found out the correct percentage of landscape by entering the desired cost into the specified field Program will then calculate how large of a top fraction can be achieved with the given cost This in turn can be used to run the landscape identification Output
198. s will be included into the analysis 2 Also define the name of your interaction file in the run settings file Run settings dat Notepad File Edit Format View Help settings removal rule 1 warp factor 10 edge removal 1 add edge points 0 use SSI 0 SSI file name S5I_list txt use planning unit layer 0 planning unit layer file plu asc initial removal percent 0 0 use cost cost file cost asc use mask O mask file mask ras asc use boundary quality penalty 0 BOP profiles file BOPcurves TXT TOP mode 1 0 E tree e connectivity Tile tree txt interactions 1 interaction file interact txt annotate name set logit space 0 treat zero areas as missing data Z g 25 sample species 0 Info gap settings Info gap proportional 0 use info gap weights 0 Info gap weights file UCwelghts spp Note that you do not need to make any changes to the actual call in the command prompt for including species interactions to your analysis 2004 2008 Atte Moilanen 94 Zonation User manual 3 5 7 In the Windows interface 4 ZIG2 Iterative cell removal done Showing removal rank Maps Run settings Species info Memo Landscape identification Solution compariso Run mode 3 Species file list is in file fe Calculate new solution Load old solution rank file A Removal rule Start of output file name Original core area
199. se settings are required if you are including uncertainty in species distributions into your analysis Info gap proportional Determines whether the errors in species occurrences are uniform errors value 0 or proportional errors value 1 Uniform error is the default setting and works for most of the data sets but in some cases it is more appropriate to use proportional errors see e g Ben Haim 2001 use info gap weights Determines whether species specific distribution uncertainty map layers are used in the info gap analysis value 1 or not value 0 Info gap weights file Indicates which uncertainty analysis weights file will be used It is very important that all the parameters in your run settings file are written exactly like presented here lf there are errors in the spelling of parameters the program can not find them and will use default settings instead The order of parameters on the other hand is not obligatory If necessary you can enter comments in your species list file on separate rows starting with the symbol Remember also to use decimal points not commas in all the input files Optional files These files are needed only when certain options are used SSI list and coordinates The SSI species Species of Special Interest are the second kind of species occurrence information that can be entered into Zonation The input for a SSI species is a probably relatively short list of observation locations instead
200. species remaining at level of removal 0 0000 1 10 28e 05 1 000 1 000 0 000 0 000 1 000 1 000 1 000 1 000 1 000 1 000 1 000 0 0010 1 101 e 05 1 000 1 000 0 000 0 000 1 000 1 000 1 000 1 000 1 000 1 000 1 000 0 0020 1 1006e 05 0 999 1 000 0 000 0 000 0 999 1 000 0 999 1 000 1 000 0 999 1 000 0 0030 1 0995e 05 0 998 0 999 0 000 0 000 0 998 1 000 0 999 1 000 0 999 0 999 1 000 0 0040 1 0984e 05 0 998 0 999 0 000 0 000 0 998 1 000 0 998 1 000 0 999 0 998 1 000 0 0050 1 0973e 05 0 997 0 999 0 000 0 000 0 997 1 000 0 998 1 000 0 999 0 998 1 000 O 1 0962e 05 0 997 0 998 0 000 0 0 997 1 000 0 997 1 000 0 999 0 997 1 000 0 00 1 0951e 05 0 996 0 998 0 000 0 0 996 1 000 0 997 1 000 0 998 0 997 1 000 0 004 1 094e 05 0 995 0 998 0 001 0 0 995 1 000 0 996 1 000 0 998 0 996 0 999 0 0090 1 0929 05 0 995 0 997 0 001 0 0 995 1 000 0 996 1 000 0 998 0 995 0 999 Picture of output curves txt file The first column gives the proportion of the landscape removed If you have initially removed some parts of the landscape before running the program initial removal the file contains only those areas that are included in the analysis The second column shows the cost of remaining landscape If land costs are not included in the analysis this column represents the number of cells that is remaining in the landscape The third column shows the minimum proportion of species distribution that is remaining in the landscape thus the situation of the worst off species
201. st or area of planning unit i Again the cell that has the smallest value will be removed 00 02 04 06 08 10 proportion of distribution remaining Above is a picture of a benefit function for species j When a grid cell is removed from the landscape the representation of each species occurring in the removed cell goes down by a small fraction ARj and the respective value for that species goes down by AVJ The total marginal loss in value is simply a sum over species specific losses Note that here the species has a standard weight of 1 0 but as with core area Zonation it is possible to weight species differently when using additive benefit function The effects of weighting are seen on the scale of the y axis which will go from 0 0 to species weight w instead of going from 0 0 to 1 0 Because the additive benefit function sums value over all species the number of species in a cell has a higher significance compared to basic core area Zonation For example using additive benefit function might lead to situation where species poor cells are removed even though they have a high occurrence level for one or two rare species because the di value of these cells is smaller than that for cells that have many common species with high representations Thus using the additive benefit function typically results in a reserve network that has a higher performance on average over all species but which retains a lower minimum proportion of original d
202. st and previously removed areas showing as buffer zones In this way landscapes can be zoned according to their potential for conservation and differing degrees of protection maintenance or restoration effort can be applied to different zones The purpose is not necessarily to produce a detailed conservation plan for a large region but to identify priority areas of the landscape that could be subjected for more detailed analysis and planning that accounts for other land use pressures than nature conservation The Zonation software has been geared towards using large grids as input data facilitating a direct link between GIS statistical distribution modeling and Zonation It is particularly simple to input modeled species distributions or land cover types into Zonation First do statistical habitat models for species then predict species occurrence across the landscape grid then feed the grids into Zonation The Zonation software can be run with relatively large datasets on an ordinary desktop PC The Zonation software is intended for the analysis of biological data with the aim of finding out spatially good conservation solutions In this process Zonation purposefully ignores other land use planning considerations such as commercial or recreational values except for what can be entered as simple cost mask layers Thus output of Zonation should be seen as an analysis of conservation value which feeds into a broader land use planning framework
203. structures respond to surrounding habitat deterioration and thus helps you to evaluate the robustness of your reserve network candidates against negative effects of habitat fragmentation Thus this analysis is most relevant for solutions developed using no connectivity methods but it can as well be applied to solutions with connectivity Running Fragmentation uncertainty analysis The Fragmentation uncertainty analysis can only be done from the windows interface 1 Goto Solution comparison window and type the name of your solution rank file rank asc file one of the output files in basic Zonation analysis 2 Determine how large fraction of the landscape will be included in the analysis 3 Next set the buffer radius This determines the area around a cell in which any loss of habitat will influence the biological value of the focal cell Size of the buffer radius depends on your conception of how far negative effects of fragmentation might reach but also on the cell size of your data as the length of radius is given in number of cell 4 Finally press the Info gap solution button to initiate the computation O 2004 2008 Atte Moilanen 101 Zonation User manual 3 21G2 the Zonation software for spatial conservation prioritization Maps Run settings Species info Hemo i Landscape identification Solution comparison About Comparison of solution overlap Compare 0 2 fraction of present solution to
204. t cost a useful measure of site value for conservation planning Biological Conservation 132 336 342 See also the following references for the benefit function approach to reserve selection Arponen A Heikkinen R Thomas C D and A Moilanen 2005 The value of biodiversity in reserve selection representation species weighting and benefit functions Conservation Biology 19 2009 2014 Arponen A Kondelin H and A Moilanen 2007 Area Based Refinement for Selection of Reserve Sites with the Benefit Function Approach Conservation Biology 21 2 527 533 van Teeffelen A and A Moilanen 2008 Where and how to manage Optimal allocation of alternative conservation management actions Biodiversity Informatics 5 1 13 For those who would wish to familiarize themselves more broadly with recent literature concerning spatial conservation planning we recommend using Web of Science or a similar search facility with key words such as reserve selection reserve network design site selection algorithm area prioritization spatial conservation planning and spatial optimization Journals such as Biological Conservation Conservation Biology Ecological Applications Journal of Applied Ecology and Environmental Modeling and Assessment among others include many studies concerning quantitative conservation prioritization methods O 2004 2008 Atte Moilanen Methods amp algorithms 16 2 2 The Zonation meta a
205. tain to deviate from your nominal model The uncertainty model specifies a set of bounds that expand around the nominal estimate as a function of an uncertainty parameter a 3 A performance function This is a function measuring how well you are doing for example what is the proportion of species distributions that would be covered by a given set of areas 4 Robustness function This function measures how large can the horizon of uncertainty a be so that conservation goals are still met even if you take the most adverse choice of probabilities from inside the uncertainty bounds good reserve candidate is such that it achieves goals while allowing for high uncertainty a The robust optimal reserve candidate is the one that achieves conservation goals while allowing for highest uncertainty Ben Haim Y 2006 Info gap decision theory Decisions under severe uncertainty 2 edition Elsevier Academic Press At simplest distribution discounting is implemented as follows 1 Take your nominal estimates the normal input distributions for species 2 Develop a respective uncertainty map for each species The uncertainty layer could for example represent the standard deviation SD of the nominal estimate or the length of the lower half of the 95 confidence interval 3 Specify e g that the horizon of uncertainty a is 0 5 or 1 E g if a 1 and your uncertainty model is 1xSD then you would essentially subtract one SD of the nomin
206. th areas that are good and bad for conservation Consequently the performance curves will suggest lower protection levels than what can be obtained if selection is based on individual grid cells With respect to the computation time needed by Zonation use of planning units is likely to cut computation times The reduction will be the larger the more cells are grouped into planning units When using planning units remember to type into your Run settings file use planning unit layer 1 planning unit option selected and planning unit layer file my plu layer asc name of your planning unit layer file If you are running the program from windows version go to Run settings window select the Use planning unit layer option and type the correct path to your planning unit layer file if the file is in the same directory with Zonation program only the name of the file is required Note that when using planning units as input data the program automatically sets warp factor to one Cost layer A standard GIS raster file asc file on land cost information This file includes all basic raster information as explained in species distribution map files and a matrix of land costs in each cell The land cost value in the matrix can be any positive number larger than 0 Thus zero nor any negative value can not be used as land cost If areas with no land costs need to be included in the analyses the land cost value for these cells can be set ve
207. the south west and cheapest at the north east region Now we need to determine how much the protection of our proposed areas would actually cost To illustrate this we first load the two solutions calculated in Exercise 4 with a cost layer cost asc and compare them By loading a already produced solution the program can calculate the increase of cost as cells are removed in the exact same order as they were when producing the solutions in the Exercise 4 Use the two batch files load_costds bat and load_costbqp bat to see how large differences if any there is in the land costs between the two solutions Distribution smoothing Boundary Quality Penalty 0 10000 20000 30000 40000 50000 60000 costneeded to achieve given conservation value 0 10000 20000 30000 40000 50000 60000 costneeded to achieve given conservation value proportion of distrigutiogs remaining proportion of distributions remaining Another possibility would be to include the cost into the analyses during the cell removal process This way the program would calculate a solution that has both high conservation value and low 2004 2008 Atte Moilanen Tutorial amp Examples 124 4 7 demands for resources This is achieved by selecting cells that have a high conservation value cost ratio To do this we rerun the analysis from Exercise 5 using a cost layer instead of loading it with the cost layer as we did above Again we use distributio
208. tions used and about any warnings or errors that occurred during the run the present Zonation version does not stop execution at every suspicious input Command prompt Command Prompt a LC cd onation Ci onation call zig2 r settings dat sp list spp output txt 0 6 6 1 6 Picture of basic Zonation call when running a new solution without uncertainty analysis or distribution smoothing The Zonation can be run from the command prompt from the directory in which zig2 exe and input files are located To do this you need to 1 Open the command prompt from your Start menu 2 Use cd directory name to change to the correct working directory which contains the zig2 exe and all the input and settings files 3 Call the program with zig2 followed by e r if a new solution will be calculated or Ifilename if an existing solution is loaded the name of the rank asc file e g Loutput rank asc e Name of the Run settings file e Name of the Species list file e A name for the output files name txt Remember to write the correct suffix after each file name e Value of the uncertainty parameter a for the uncertainty analysis UCA If uncertainty is not included in the computation this should be set to zero e Value to determine whether distribution smoothing is used in the analysis parameter 1 or not parameter 0 O 2004 2008 Atte Moilanen 45 Zonation User manual 3 2 2 e Factor for
209. to Memo all calculations related to using species interactions option It is recommended to check from here that all interactions have been loaded correctly in the beginning of the analysis to avoid false results With respect to interpretation of curves assume a species or whatever feature has been entered into the same run as a raw distribution a connectivity distribution and as transformed as a spatial interaction to a source Then the curve for the raw quality layer tells the fraction of local habitat quality retained the curve for the smoothed layer tells about the fraction of connectivity retained and the curve for the interaction tells about the fraction of the potential for spatial interaction retained Inclusion Exclusion cost analyses using mask files This is a method that allows you to forcibly include or exclude specific areas to or from the final reserve design and to evaluate the costs conservation value or economical following such a constraint placed on the solution The theory behind the replacement cost analysis is explained in section 2 6 To find out more about replacement cost analyses see Cabeza amp Moilanen 2006 Biological Conservation 132 336 342 2004 2008 Atte Moilanen ZIG The Zonation software 92 Running the analysis with a mask file To include exclude areas to from the final solution you need a 1 Removal mask file in which the hierarchy of cell removal is determined This fi
210. to carefully consider the cell removal rule you use Analysis options that you might well wish to modify include edge removal the warp factor and adding of fake edge points See sections 3 5 and 3 6 for a full list of options O 2004 2008 Atte Moilanen 115 Zonation User manual 4 1 Exercise 1 Getting started with the basic Zonation Before starting see section 3 2 for how to operate the program either from command prompt or windows interface We start with a simple exercise by conducting the basic Zonation analysis Let us think that there is a area in a remote country somewhere which is the home of seven rare species We have been given a task to create a proposal for conservation network that will help to protect them However due to cost restraints the proposed conservation areas cannot be larger than 15 of the landscape We decide to use the Zonation program to identify areas that have high priority for conservation We also decide to use core area Zonation as our planning method variant because it best corresponds to our planning objectives see section 2 3 In your first species list file splist spp you have a list of seven species distribution maps and species specific parameters in front of the map names Here all species are given an equal weight but they have different dispersal values as the use of surrounding landscape e g home ranges differs between the species Because no other features a
211. to make a new copy of the program to the directory containing data files when starting a new project Quick start Here are instructions to run the basic Zonation for those who have already familiarized themselves at least to some extent with the program For more detailed instructions and additional analyses please see sections 3 2 Running Zonation 3 3 Input files amp settings 3 5 Main analyses and 3 6 Post processing analyses amp options Using Windows interface 1 Torun the program you need at least two sets of input files e Species distribution map files which are basic raster files asc exported from GIS programs These files define species distributions in the landscape The program can incorporate any kind of species distribution data such as presence absence probabilistic or abundance data or species specific population connectivity surfaces etc Zonation v 2 0 can also use point distribution data e The names of all the species files must be listed in a separate species list file spp each file on a separate row with the species specific parameters before the file name see section 3 3 2 2 Species list file for more detailed descriptions The species list file tells the program which species distribution files will be used in the analysis Remember to always use decimal points NOT commas in all input files 2 Double clicking on the zig2 exe icon starts the windows version of the program 3 Goto Run settings wi
212. tribution smoothing multiplier for dispersal kernel alpha 1 000 Use info gap distribution discounting 0 000 Use boundary quality penalty BQPcurves txt Info gap alpha BQP definitions file name Uncertainty model BQP mode fe Uniform error default C Proportional error List of error weight map layers load from UCweights spp BQP aligns missing data Potential habitat is species specific Use directed NQP PLU connectivity load from file tee txt Boundary Length Penalty 6 000 Additional edge points Output Note that using distribution smoothing should result a distinctively more aggregated solution compared to basic Zonation analysis Thus these two solution should probably not have e g a 99 overlap with each other If your solution with distribution smoothing does not show clear aggregation check the run settings for possible errors for example you could have a in different units than the cell size in the species distribution files See section 3 8 for troubleshooting 2004 2008 Atte Moilanen 81 Zonation User manual 3 5 3 ET Ts 7 L a cbr cs EE os Mo y i Ya x 5 i 4 s Picture of a typical output map when distribution smoothing has been included in the analysis BQP Zonation The boundary quality penalty BQP is a quantitative method that induces aggregation into reserve networks according to the needs of individual species
213. tributions for both SSI species are retained till the very end of the cell removal process prop S3b distributions remaining 0 3 04 05 06 07 08 09 1 proportion of landscape lost 0 2 0 0 1 Picture of SSI species curves in the species info window Note that the same information that is displayed in the graph is also outputted in SSI_curves txt file which the program produced during this analysis As the inclusion of the two SSI species contributes very little to the final result from now on we keep them separate from the main analysis and continue with the original seven species O 2004 2008 Atte Moilanen Tutorial amp Examples 120 4 4 Exercise 4 Adding aggregation into the analyses Now we have identified sites that have a high occurrence of our target species weighting the two endemic species But the areas are quite fragmented which is never a good quality in a reserve network Thus we want to produce a more aggregated solution To do this we try two different aggregation methods distribution smoothing and the boundary quality penalty Both methods favor the selection of contiguous cell groups rather than selecting more fragmented sets of cells This in turn offers advantages in terms of greater connectivity and can also promote more practical and cost effective management Note that it is not recommended to use several aggregation methods simultaneously due to difficulties in interpr
214. undary quality penalty 24 25 81 boundary quality penalty definitions file 63 BQP 24 25 28 63 64 81 BQP mode 53 Ce call 44 cell removal principles 17 cell removal rule 17 20 50 59 climate change 39 colors 67 command line 44 command prompt 8 44 110 competition 39 compulsory input files 49 connectivity 24 25 28 39 79 81 83 89 conservation value 31 37 consumer resource interaction 39 convex 20 convex optimization 4 coordinate files 58 core area Zonation 17 cost 60 99 cost curves 67 cost layer 53 60 curves 67 curves file 73 D directed connectivity 24 28 53 59 64 directory paths 108 distribution curves 67 distribution discounting 31 62 63 86 distribution maps 49 distribution smoothing 24 25 79 distributional uncertainty map layer 62 download 8 downstream connectivity 28 E edge removal 53 error weights 63 errors in data 31 evaluating conservation areas 3 7 examples 113 excluding areas 61 excluding areas from the solution 37 91 exclusion cost 37 extinction risk curve 67 extinction risk output 67 Tr file names 108 file types 49 fragmentation uncertainty analysis 35 100 framework 3 G generalized benefit function 17 22 genetic algorithm 5 GIS 2 H habitat loss rate 42 habitat model 2 habitat quality histogram 67 heuristic 16 2004 2008 Atte Moilanen 129 Zonation User manual heuristic definition 5 h
215. ur NQP connectivity description file Note that when planning units are used the program will automatically set warp factor to be one regardless what has been defined in the run settings oO N 2004 2008 Atte Moilanen 85 Zonation User manual 4 11G2 Iterative cell removal done Showing removal rank Maps Run settings Species info Memo Landscape identification Solution comparison About Run mode re x z Species file list is in file Calculate new solution C Load old solution rank file AAA Removal rule Start of output file name Original core area Zonation E A Use cost file C Additive benefit function Use incl excl mask Use SSI file C Target based planning C Generalized benefit function Use planning unit layer Use interactions file mn oo gt 620 Warp factor Resampled species count Info gap settings Use info gap distribution discounting 6 000 Info gap alpha Uncertainty model Uniform error default C Proportional error List of error weight map layers load from UCweights spp my_splist spp transform species occupancy probabilities from Jog output_ NQP dat Annotate outputfile na mask ras asc SSI_list txt PLU asc interact spp Settings for generating spatial aggregation into the solution M Only remove from edges Use distribution smoothing multiplier for dispersal kernel alpha 1 000 Use bounda
216. use planning unit layer to 1 in your run settings file and give the name of your planning unit layer file 2 Set use tree connectivity to 1 to indicate that NQP will be used 3 Also define the name of your NQP connectivity description file in the run settings file 4 Note that when planning units are used the program will automatically set warp factor to be one regardless what has been defined in the run settings file 4 E Run settings dat Notepad SEX A File Edit set ings removi rule 1 warp factor 1 tadoae removal 1 Ladd a dae points O Format View Help INSe 55 i SN file nas S5I_list txt usen lannina unit layer 1 planhing unit layer Tile my PLUS asc use cost 0 2 cost file cost asc use mask 0 ask file mask ras asc tse boundary quality penalty 0 BOPSproftiles file RQPcurves tT BOP mise 1 BLP use tree ones 3 tree connectivity file my tree file txt use interactions 0 interaction file intefact spp annotate name logit space E treat zerg areas as missing data O z 0 resafple species O 33 Info gap settings Info gap proportional use info gap weights Info gap weights Tile O UCweights spp In Windows interface 1 Select the planning units option from the Run settings window and give the name of your planning unit layer file Set use tree connectivity to 1 to indicate that NQP will be used Also define the name of yo
217. ution 2 Cells with a value of 2 are removed first These cells may be for example undesirable for conservation e g built up areas private areas areas ear marked for residential building or commercial fishing etc or they may have any other reason to be primarily excluded from the final solution E mask asc Notepad Fe Fat Format Ven Hep ncols 649 nr ows 955 xllcorner 294205 yllcorner 6283604 6 cellsize 200 INODATA value 9999 39999 9999 9999 9999 9999 9999 9999 9999 9 9999 9999 9999 000000000 9999 0 0 Picture of removal mask layer file It is important that the mask layer raster has the same grid size as the species distribution map rasters This means that in all files the number of columns and rows as well as the size of cells should be equal It is equally important that all those cells which have data of any of the species used in the analysis that is to say the cells that are NOT marked as no data in species distribution files also have a mask value Otherwise undefined program behaviour may occur Remember also to use decimal points not commas in all the input files 2004 2008 Atte Moilanen ZIG The Zonation software 62 When using the removal mask layer remember to type in to your Run settings file run mask 1 mask option selected and mask file yourmaskfile asc name of your removal mask layer file If you are running the program from windows version go
218. utomatically aligns missing data if different species layers happen to have missing data at different locations When aligning data if species A has missing data at location x y where any species B has a positive occurrence then the missing value for species A is replaced by a zero level occurrence Mode 2 indicates that the data no data matrixes are not uniform and aligned and that the program needs to calculate species specific buffers for each species cell separately Mode 2 is more realistic in the sense that fragmentation loss in habitats that are not suitable for the species will not influence the value of the focal cell But mode 2 also requires longer computation times due to more complicated species specific calculations Also use of mode 2 at least doubles the memory usage of Zonation thus decreasing the number of species that can be run in one analysis Thus mode 1 is a preferable when all species use approximately the same habitat type Uniform data matrixes Species specific data matrixes mode 1 mode 2 Species A Species B Species A Species B OG Gt No data Focal cell El Data Buffer Defines a penalty given for the boundary length of the reserve Use of BLP leads to more aggregated solution Try first a small value e g 0 01 to perceive the effect of BLP to the solution When including BLP in the analysis always use a warp factor of 1 If BLP is not used this parameter should be set a
219. vision multiple alternative outcomes for each location e g different levels of protection or restoration The currently available Zonation version does not allow for multiple alternative land uses partial solutions to this issue are under consideration e At present Zonation operates on input distributions of biodiversity elements species land cover types etc lt does not operate on environmental surrogate information This is not a conceptual limitation and will most likely be lifted in forthcoming versions of Zonation e The connectivity measures used in Zonation in distribution smoothing BQP or NQP are simplistic in the sense that they are based on species specific dispersal distances or scales of landscape use and influences of habitat loss However they do not include any explicit analyses of movement paths such as least cost path type analyses The directed connectivity of v 2 0 partially alleviates this issue e The Zonation software does not include a full GIS interface and only a limited set of analyses and graphs can be produced with the software The analysis output files can however be imported into GIS software for further processing e The present software can use at maximum 4 GB of memory on a PC which sets the ultimate limit of how many species and in how large landscapes can be handled This limitation is likely to be lifted in a forthcoming release of the software 2004 2008 Atte Moilanen Z
220. w you the top 15 fraction of the landscape for all species Check also how large an area would be needed to protect at least 30 of all species distribution Remaining button Double clicking the image lets you save it into a graphics file Remember also that the information used to produce these figures has been saved into the rank asc and prop asc files which you can import into GIS to prepare your own maps O 2004 2008 Atte Moilanen Tutorial amp Examples 116 Batch file do zig2 bat Rank 0 15 Remaining 30 Top 15 Area 27 582 Area 16 543 BL A 0 587 BL A 0 977 Cost 27 582 cells Cost 16 543 cells av prop 33 0 av prop 20 5 min prop 29 9 min prop 18 3 The cell ranking is only one part of the relevant Zonation output Another part is a set of curves describing the absolute performance levels of species The figure below shows the minimum red line and average blue line performance across all species for our basic Zonation analysis above With this data set a fraction of species distributions is lost already when only a small fraction of the landscape has been removed This reflects the fact that the species in the sample data are both relatively widespread and that some of them have mostly non overlapping distributions This is different from e g the analysis of British butterflies where the distributions of species were narrow and nested and a substantial fraction
221. when preparing your species list file keep in mind the following points 2004 2008 Atte Moilanen ZIG The Zonation software 90 Every interaction changes the loaded original distribution layer For example lets assume you have two species A and B and you wish to run an analysis with the two original distributions plus the connection of species A distribution to species B distribution To do this you need to list species A twice in your species list file The first layer will be used as it is original distribution and the second one will be transformed based on its connectivity to species B If species A were listed only once the landscape ranking would be done only based on species B distribution and species A connectivity to species B Thus species A original distribution would not be included e Make sure that you are not using already transformed layer to transform other layers Unless you absolutely want to However in that case the interpretation of results is outside the scope of this user manual Note that every file listed in the species list file will be used for landscape ranking If you want to use a layer to transform another layer but not to be included into the analysis itself you can do this by setting the weight of the layer to zero In addition you need to adjust your settings In command prompt 1 Set use interactions to 1 in your run settings file to indicate that species interaction
222. where political decisions are made that balance between different land use needs Note that Zonation can also be used for identifying the least important parts of the landscape those in which human activity would cause least harm to biodiversity value O 2004 2008 Atte Moilanen Zonation User manual 1 2 The Zonation framework Aim and purpose e To provide a tool for large scale high resolution spatial conservation planning using primarily GIS grid data Analyses e Identification of optimal reserve areas Identification of least useful conservation areas Replacement cost analysis for current or proposed reserves Planning methods Core area Zonation Additive benefit function Target based benefit function Data e Large scale grids with Presence absence data Probabilities of occurrence Abundance density data Cost and mask layers e Point observation data New in v 2 0 e Planning unit layers New in v 2 0 Features e Methods for dealing with connectivity needs of species Distribution Smoothing species specific Boundary Quality Penalty species specific Boundary Length Penalty Directed Freshwater Connectivity species specific New in v 2 0 Uncertainty analysis aiming at reliable conservation decisions Species weighting Clearly defined trade offs between species Species interactions New in v 2 0 2004 2008 Atte Moilanen Introduction 4 1 3 Zonation
223. wing command line Call 2192 Os 0 Q deb 0 r settingsfile dat specieslistfile spp outputfile txt In this command line give the names of your settings file and species list file and define a suitable name for your output files See section 3 2 1 for explanations for the four numbers in the call Note that this call can also be written into a normal text file using notepad If the extension of the file is renamed to bat a batch file the Zonation run can also be initialized simply by double clicking the file name in the Windows file manager This is an easy way to prepare and use batch files 6 Press enter to initiate the computation O 2004 2008 Atte Moilanen Introduction 10 Command Prompt C cd Zonat ion CisZonation call zig2 r settings dat sp_list spp output txt 4 0 6 1 8 The basic Zonation program can be used for example for identifying a best proportion of the landscape rank selection in Map window or for identifying the area required for representing a certain proportion of the species distribution remaining selection in Map window The program automatically produces six output files e jpg and bmp maps of the landscape ranking showing the order cell removal in different colors See section 3 4 1 for detailed interpretation of the colors e A curves txt text file containing a list of species and weights used in the analysis and columns representing how large proportion of original o
224. x J up J gin J pq Ej pq ken ki ki ki ki ke P a be A F p 4 gt D Ma a UE hal N a 9 Kj J up J down J up J gon keNj kj kj kj kj 1 in which p is the occurrence level of species j in cell Equation 1 describes the fraction lost from original distribution of species j following the removal of site i The loss consists of three components local loss loss upriver and loss downriver The assumption is that everything remaining locally is lost if a cell is removed and that loss accrued upriver and or downriver will depend on the size of unit for species j A loss of a larger unit implies greater influence on connectivity nearby The influence of connectivity on occurrence levels is mediated via functions h upriver and downriver which are response functions like those in BQP with the x axis reversed When the full landscape remains and nothing has been lost h 1 1 Importantly when calculating marginal loss the equation accounts for degradation that already has occurred This implies that if no local value remains due to past neighborhood loss further loss of connectivity has no local influence on the species Connectivity in Eq 1 is modeled separately upriver and downriver Quantities Fj and Oj up and down are the remaining and original connectivities of unit for species j both upriver and downriver respectively Loss of planning unit influences the downwards connectivity of sites upriver fr
225. y habitat in the neighborhood of the location Such a neighborhood influence essentially states that the species is somehow dependent on connectivity or edge effects or both Now ideally reserve selection would be directly based on nonlinear habitat models with neighborhood effects However this is not realistically possible because it would make reserve selection computationally very very slow Also implementing dozens of different habitat modeling techniques efficiently into reserve selection software would be an enormous task Herein enters the BQP The BQP is a mechanism for approximating the aggregate response of a species to edge effects and metapopulation size and connectivity lt can be seen as a way of exploiting the connectivity response that is present for a species in a habitat model Essentially one uses the habitat model for two things First one predicts an abundance or probability of occurrence into every cell in the landscape giving the standard input layer for one species Additionally one analyses the habitat model to find out what kind of an aggregate response to habitat loss and fragmentation does the species have And this response is transferred into Zonation as a standardized curve which mediates the boundary quality effect in Zonation Different species can have different responses to fragmentation and habitat loss which are entered into Zonation as two BQP components i a species specific radius and ii a resp
226. y of occurrence of a certain species e g very rare species The higher the weight the more strongly the program prefers cells with low uncertainty Species specific weights can have any positive value larger than 0 Thus zero or negative values cannot be used as weights If no species specific weighting of uncertainty is used as is most commonly the case this should be set as 1 0 all equal 2 Name of the species distributional uncertainty map layer If your uncertainty maps are in different directory than your weights file remember also to type the correct path in front of the names Note that the uncertainty layers for species have to be in the same order as the species distribution maps in the species list file these files are linked to each other solely via the order of listing in the two files When using the windows version select the Use info gap distribution discounting option in Run settings window under Info gap settings and type the correct path of your uncertainty analysis weights file on the field If you are running the program from command prompt remember to type the name of your uncertainty analysis weights file to your Run settings input file Info gap weights file yourweightsfile spp under the Info gap settings Note that there shouldn t be any empty rows at the end of the uncertainty analysis weights file This is because the program might interpret these as empty values or files that just don t have any
227. y scale 2004 2008 Atte Moilanen 105 Zonation User manual 3 6 7 The Memo keeps track of what the program has been doing At the end of the memo you can find a summary of your results e g how many cells where included in to the solutions in more than 90 of the cases how many cells were never included etc In the parentheses are the proportions of each cell group Note that the group lt 10 also contains those cells which were never included in the solutions where as the group 90 contains the cells which were always included 4 Jonation landscape identification summary Seles Image Memo About n LSl_cost ras asc 555 rows read m total Nodata element count 243913 Sum of probs 42078 LOADED LS cost ras asc UD Adding solution from LS OS ras asc LS DS ras ase 555 rows read in total Blodata element count 249973 Sum of probs 55896 LOADED LS DS ras ascl UD Adding solution from LS n ras asc LS nras asc 555 rows read in total Modata element count 249913 Sum of probs 4696 LOADED LSI nras ascl UDI Adding solution from LSl_blp raz asc L5l_bip ras asc 555 rows read in total Modata element count 249913 Sum of probs 41767 LOADED LSI blp ras asc Count of processed fles 4 sim datamode 1 show 01 po 1 Dutputtng to out sum asc SUMMARY INFO Missing values 249913 Included never 0043 0 63912 Included lt 10 of time 0043 063514 Included

Download Pdf Manuals

Related Search

Related Contents

  SIIG 04-0600B User's Manual  Bullet Tools 226 Use and Care Manual  2015 TRANSIT CONNECT Owner`s Manual  プロセスチェーンマネジメントのコンセプト  Ficha del producto    Instruction Manual - Magnetrol International  

Copyright © All rights reserved. Failed to retrieve file