Home

RobOff – User Manual Software for Robust Offsetting, Habitat

Contents

1. ssssssssseseeseeeneneeennene hene enne 23 2 5 Aggregation of conservation value sssssseeeeenenne 25 2 5 1 Uncertainty Analysis seris 25 2 5 2 Aggregating conservation value sssee 26 2 5 3 Weak and strong sustainability sseeene 32 2 5 4 Time Eeeevupece tc 33 2 5 5 Robust and opportunity performance indices sssusss 34 2 6 Optimizing resource allocation sssseeeen 35 2 7 Dealing with connectivity rescozon emen 38 2 8 ASSUMPTIONS and limitations 2 eee eee eee e eee EANAN ee me 39 2 9 TROTCL SIGS S ass aaa P E 40 Chapter 2 Framework Methods and Algorithms 2 1 The RobOff framework RobOff is a framework and software for conservation resource allocation What distinguishes RobOff from other frameworks and software systems for conservation planning is its emphasis on Uncertain responses of biodiversity features over time Multiple alternative actions with different effects costs and feasibility RobOff focuses on the question of what to do rather than on the question of where This question is directly relevant for example when investigating the effects of alternative actions in areas that face different threats RobOff fills a niche in that it concentrates on the uncertain effects through time that alternative conservation actions have on different biodiversity features in different
2. ssssesseee 111 6 1 Defining environments and actions in the setup section of the RobOff GUL et nee ERE 122 6 2 Summary of results for the minimal setup in the RobOff GUI 124 6 3 Feature responses of the Dam Forest River example in the RobOff QUI ert te te Othe te eR il e 128 vii List of Tables 2 1 Typical or suggested usages of RobOff eee eeeeeeeeeeeeeeeeeeaneneeeeeeees 18 2 2 Mathematical symbols aasian aesan a A aa a e aek e aiaa 28 2 3 Key terms for understanding conservation value aggregation in RRODOM ge E A E A N S 29 2 4 Scalability and optimality of the optimization methods supported by RObDOM re 37 2 5 Speed and ease of use of the optimization methods supported by ROBOM M XE 37 3 1 Example summary of results conservation value sssus 72 3 2 Example summary of results conservation performance ratios 72 viii Part I Introduction Table of Contents andre Pei EE 3 1 4 Aims and pUrpOSOQ eterno terat rs Eno e set epe unte e Dea 3 1 2 The RobOff framework in a nutshell sseem 4 1 3 RobOff inputs and outputs ssnin cusi asia aAa Fx a Ern ERU aaa 5 1 951 Inputs ie edo pe ccna ence se menceeer neers merrerr er rere S 5 WES 23 OUTPUTS e 5 1 4 A typical RobOff work flow 20 0 cece ce cece cece cece ae eeeeeeeeeeeeaeaaeaaeeeeeeeeeaeaaeaeneeeees 6 1 4 1
3. Environment oArea Action in environment Per unit area cost Available area Inputs Per area unit uncertain response nominal upper lower Aggregation of values Total cost Occurrence level of feature i n 1 in environment over time nominal in upper lower Uncertain total occurrence across environments for feature Amount implemented Cost 5 Optimization method 1 n 4 Candidate solution 1 1 Optimal solution 0 1 1 i n Robust conservation value iin Constraints and criteria Optimization The RobOff system records and updates specific characteristics of features in each environment and provides for the linked analysis of features within and across environments 2 3 Complementarity and scoring RobOff combines complementarity and scoring two approaches to conservation that have been seen as fundamentally different We use a two step aggregation 22 The RobOff output space process in which some features e g species are treated as individual biodiversity features in their own right while others are treated as score components that are first aggregated into uncertain scores before they are used to assess conservation value We call the former simple features and the latter scores Simple featu
4. 128 Index Symbols 64 bits 83 A Abundance 29 57 Action 21 113 see also allocation allocation 90 92 comparison 48 77 101 compensatory 48 cost 21 35 58 68 90 development 48 editing 90 92 GUI 90 92 101 mandatory 48 92 optimal 48 92 output 77 preset 48 92 response 62 scheduling 21 sets of 48 Actions overlap 30 Administrative region 110 Administrative unit 21 Aggregation across environments 32 33 35 across features 32 33 35 across time 34 conservation value 28 occurence levels 28 representation 28 Agri environment 112 Aim and purpose 3 Alleles 21 Allocation 113 see also Action mandatory 76 multi action 19 optimal 76 76 113 preset 76 spatial 3 19 38 41 Alpha 30 Alpha transparency 104 Alternative actions see Action GUI 101 output 77 Amount implemented 22 Analysis management level 18 Offsetting 18 113 see Offsets Restoration 18 see Restoration Targets 18 113 see Target types 112 Uncertainty 25 Area see Cost selection 41 target 113 B Benefit see utility function 40 60 Benefit function 59 92 see also utility function concave 59 constant 59 convex 59 editing 92 exponential 59 file 59 92 generalized 59 GUI 92 inverse sigmoid 59 linear 56 57 59 negated sigmoid 59 piecewise constant 59 piecewise linear 59 quadratic 59 sigmoid 59 Zonation 59 Binary see Operating system Biod
5. dam construction will result in complete loss of 300 ha of total forest area of 1000 ha management to control introduced pests could be applied in balance of forest ForestTypeA 0 r1 cc env MyddleForest act busuiness as usual Inundate r2_CC_env_MiddleForest_act_Inundate 0 0 Restore r3_ccC_env_MiddleForest_act_Restore 0 0 As an example response the response of the UpperForest environment to the Restoration action response_r5_CC_env_UpperForest_act_Restore csv Feature specific response file Saved automatically by RobOff Edit at your own FISKI assumes more rapid increase in ecological integrity as a result of intensive control of introduced browsers and predators Paes E EPROR 769703139057 0 TAIS 82 0 8051 0 8349 865 0 84515 0 88485 9 0 8752 0 9248 1925707895325 095415 94 ODIO 945 0 90535 0 98465 9457 09050 9996 9497090535 0 994515 295m OU OSS 0 sees iS le 61 e Cie ere eie c The feature weight function file defines equal weights for both features 1 and functions of time concave increase with diminishing return This file feature weight functiontypes csv Feature weight function types file Saved automatically by RobOff Edit at your own pias kal Format feature name weight benefit function comma separated list of parameters ForestTypeA 1 00 1 0 25 ForegtTypeB 1 007 Tr 0 25 Finally a budget allocation file can be used to enforce certain actions an
6. Figure 4 20 Setting optimization options Gua Roboff GUI View Results Tools Help eng rve e 4 Setup Results 9 Optimize Options Env Action Amount area Amount equivalent money Criterion Y 3 MiddleForestA Weak features Creme ur do nothing 300 o i Robustness v G MiddleForestB Strong features j do noth 358 d O Nominal o nothing Weak environments Restore 400 2000 O Opportunity Strong environments v UpperForest Restore 30000 150000 Use of preset allocations Enforce preset actions Budget resolution 1 00 2 Max time 300 00 7 Seconds 5 E Keep 1 2 best solution s Method Exhaustive search 2 Run 0 Save Load G G9 Log Feature response file 4 home fedemp example setup dam response r4 CC env MiddleForestB act Restore csv Ok Feature response file 5 home fedemp example setup dam response r5 CC env UpperForest act business as usual csv Ok Feature response file 4t 6 home fedemp example setup dam response r6 CC env UpperForest act Restore csv Ok Found 2 actions subject to optimization in 3 environments E 4 5 Preferences The dialog of general RobOff GUI preferences can be accessed from the main menu Tools gt Preferences The most relevant settings are Level of verbosity in the log window general information and error reporting level of verbosity report only errors or errors and warnings or errors warnings and notices It is also possible
7. sssseeeeee 91 4 5 Editing uncertain responses in the GUI ssseseeeeee 91 4 6 Editing allocations in the RobOff GUI sssseseeeenee 93 4 7 Editing benefit functions in the RobOff GUI ssesseseesesss 93 4 8 Editing costs in the RobOff GUI ssessseeeemmee 94 4 9 Editing time discounting model and parameters in the RobOff GUI 94 4 10 Editing score features in the RobOff GUI ssssseesseess 95 4 11 Visualizing results summary sese 97 4 12 Visualizing results through time ssmm 98 4 13 Visualizing results across environments eseeeeeeeee 98 4 14 Visualizing results across features sseseeeee 99 4 15 Visualizing results across features within environments 99 4 16 Visualizing results across actions see 100 4 17 Visualizing results as a function of the degree of uncertainty 101 4 18 Comparing actions ve deditio el eed ect due de vd ee ra ua 102 4 19 Visualizing results as a function of the available budget 103 4 20 Setting optimization options ssssesee ee 104 4 21 Editing RobOff GUI preferences sssssssesee 105 5 1 A possible simplified sequence of steps required to define a RobOff i o EET 110 5 2 Flow of definition of RobOff setups
8. p 70 for details on the output files that are generated by the RobOff command line or by saving results in the GUI Alternatively see Section 4 3 Results p 96 for details on the visualization functionality of the results section of the GUI RobOff software for allocation of conservation effort with multiple actions Loading configuration Reading RobOff setup file toy offsetting ro setup Processing RobOff setup file No discounting configuration file found Discounting model is Quasi hyperbolic and rate 0 Reading environments file environments csv Environments 4 1 Environment environment 1 weight 1 total area 3 active area 3 condition 1 No budget allocation file name specified skipping loading of budget allocations Reading features weights function types file feature_weight_functiontypes csv Reading specific responses reading file features_present_environment 1 csv Looking for 3 feature response files starting from compensation Feature response file 1 response compensation csv Ok Feature response file 2 response development csv Ok Feature response file 3 response donothing csv Ok Setup validated Found 1 environments 1 features 1 presences 3 responses and 1 actions subject to optimization Budget allocated 0 Set of actions of this setup succesfully validated Everthing seems correct Starting core calculations of conservation value and sustainabi
9. Optimize Looking for feature response file setups setup simple demo response_r_6 3 txt 2j Regenerating uncertainty graphs for uncertainty parameter 1 Regenerating uncertainty graphs for uncertainty parameter 0 8 Setup setups setup simple demo successfully loaded amp OpenMP support there are 4 threads in parallel regions 5 Lel 11 Quick start Figure 1 2 Example screen capture of the RobOff GUI editing responses in the setup section RobOff robust offsets calculator o Roboff GUI View Results Tools Help 9B vo te 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring 0 6 Response Estimate ower bounc Upper bound as r7_CC_env_OMT_act_business as usual 10 0 o o r4 CC_env OMT act set aside r1_CC_env_OMT_act_clear_cut_ _set_as UU og 5 r16 HG env OMT act business as usual 30 0 o 0 r13 HG env OMT act set aside 40 0 0803 0 00915 0 1514 o4 r10_HG_env_OMT act_clear_cut_ _set_i 50 0 1612 0 01875 0 30365 r25_LTT_env_OMT_act_business as usual 722 LTT env OMT act set aside 60 0 242 0 0298 0 45425 es r19_LTT env OMT_act clear_cut set 70 0 2505 0 0428 0 4582 r34_FS_env_OMT_act_business as usual 80 0 2698 0 0628 0 47685 r31 FS env OMT act set aside a 128_FS_env_OMT_act_clear_cut_ _set_a 90 2905 CHUAN WEEE r43_TTWO_env_OMT_act_business as us 100 0 2934 0 0725 0 5143 r40_TT
10. a Given a proposed set of amounts areas on where actions are performed the total representation of feature fj in in the landscape environment E is K FO gt s ol n j tyke b The conservation value in this environment is Vf w jg fo 0 Endfor Ne The retention of representation of feature fj BIS R O DO rl The conservation value for feature fj is V f w jg KR f Endfor 31 Weak and strong sustainability Lik The total conservation value across all features is V v 0 J71 An important practical consideration is that in the implementation of RobOff described in the next chapters conservation value across features or environments is by default divided by the sum of feature weights or environment weights respectively If all the feature or environment weights are 1 this is equivalent to the number of features or environments Hence the total conservation values produced by RobOff are in fact weighted average conservation values across features or environments This allows for example to compare conservation values obtained for different setups or different variants of a same setup that may have a different number of features From the values derived above the relative performance for feature i e a ratio between the conservation value obtained for the selected actions and the value obtained when no action is performed is V 0 I Vig These ratios are measures of the conser
11. p 62 Here is an example of this type of file 59 Feature weights utility functions file Types of functions 1 concave increase with diminishing returns pow R x ip 2 Aue Sie RO sie dep ge db ge dsb St bk 3 exponential decay exp p t a dL c ofone ilo Se ub NS AY 5 sigmoid 6 inverse sigmoid 7 quadratic polynomial 8 piecew constant 9 piecew linear 10 piecew interp Features weight function type function parameter s comma separated list ic db abun db 9 Sip pow oe fleo S iso ISO ap X539 The example includes a list of supported benefit functions Note that the number of parameters required depends on the type of function In any case the parameters must be provided after the function type code and separated by commas For the piecewise constant piecewise linear and piecewise interpolated functions the number of parameters is unlimited and these must be provided as a list of pairs of x y values the number of parameters is always even Examples of some types of benefit functions are shown in Figure 3 3 Example benefit functions p 61 The panels in the figure illustrate benefit functions of type power panel A function 1 target B 2 quadratic C 7 inverse sigmoid D 6 two different piecewise constant functions E amp F 8 an exponential decay G 3 a generalized benefit function H 4 and a piecewise linear function I 9 Piecewise co
12. score1 0 5 f2 0 7 f3 0 6 f4 To calculate an arithmetic mean like score use the operator for all the components To calculate a geometric mean like score use the operator for all the components Note that the weights need to be adjusted if you intend to calculate an arithmetic mean The next example defines two score features score_add f1 f2 f3 and score_mult f1 0 5 f2 f5 RobOff beta release Score features file table list of score components fl score add ty 1 EZ Score addi p 1 f3 score add t L fLl score mult ar dE E2 Score mult 545 ES secre mult ws dL Note that it is valid to use a biodiversity feature as a score component of two or more score features You can alternatively enter a same feature multiple times with different names in the biodiversity features present file and use those different names for different score features in the score features file N Note Remember that when you include a biodiversity feature in the score features file of an environment it will become a score component and will not be used as a simple feature in that environment RobOff identifies it as a score component It is possible to use the same feature both as a simple feature and a score component You just need to enter the same feature twice or more times with different names but using the same responses weights and benefit functions which effectively replicates the same feature with d
13. Chapter 4 RobOff Graphical User Interface p 87 for a description of the RobOff GUI where the same results can be visualized interactively Note that the same output files described here for the command line interface of RobOff can be generated from the GUI Figure 3 5 Set of output files and their equivalent GUI dialogs Adapted and expanded from Pouzols amp Moilanen 2013 Text files automated processing GUI interactive visualization Default General information on RobOff run Summary of conservation value and sustainability ratios Visualization across time Summary of results Visualization against uncertainty Log of warnings and errors er Visualization across environments Conservation value across time and uncertainty degree Visualization across features Performance sustainability indices across time and uncertainty degree Visualization across features within environments Visualization across actions Visualization of pairwise comparison of actions Results of uncertainty analysis best possible performance against uncertainty Visualization of budget analysis performance against resources available Results of comparison of actions resources against uncertainty Results of budget analysis performance against budget Optimal allocation results from optimization list of action amount pairs File of feature conservation values across time File of environment conservat
14. Local experts can then apply their insight to specify exactly where action should be taken accounting for any external information about land availability and socio political considerations Third RobOff methods can be used in a first stage to come up with area objectives or targets for different actions Then in a second stage one can use software for spatially explicit planning such as Zonation Marxan Marxan with Zones or ConsNet to suggest explicit spatial allocations based on potential management scenarios Overall these ways of accounting for connectivity and location provide a practical solution to the dimensionality problem without requiring a fully explicit spatial model It should be noted that in fact there is a gradient of options the extreme cases would be a the whole landscape aggregated into a single environment and b a specific environment for every single site In practice there is an important difference in that in the second option areas with different degrees of connectivity are dealt with differently in RobOff calculations possibly using different responses etc In short optimization results would be more fine grain area targets in the second case with targets specific to areas with certain degrees of connectivity 38 Assumptions and limitations 2 8 Assumptions and limitations Conceptual simplifications RobOff is a spatially implicit conservation planning framework As such it does not
15. Sets the discounting model that should be applied to costs Together with the discounting rate see above this defines the relative weights of costs over time which are used to calculate the corresponding net present cost The following options are available quasi hyperbolic hyperbolic and exponential Default quasi hyperbolic For simple uses of RobOff this option is not needed It sets whether to enable rescaling of responses based on the start value and end value fields in files of features present for details on how to use this feature see Section 3 3 4 Set of files biodiversity features p 62 Options no or any other value different to yes default and yes Unless yes is specified the start and 55 General settings file end value fields are ignored Note that this option does not change the way RobOff calculates conservation value it just modifies the values of responses and it is provided for convenience responses scale Scale in which the response values are expressed in the response files The possible scales are range 0 to 1 absolute and proportional gain Default range 0 to 1 This option controls the way in which benefit functions are applied to occurrence values The default is in principle the most convenient choice for most problems Proper use of the last option responses scale requires a good understanding of how conservation value is aggregated across features and environment
16. Setup Results Summary Time Uncertainty Environments Features Env Features Actions Compare Actions Budget V Robust E i Opportunity NI Nominal mme Features 6 Add O Strong Solid 2 amp Weak Dashed x Clear Env feature Weig Conservation value v 047 v GE MiddleForestA VUE o B ForestTypeA v B3 MiddleForestB ForestTypeA Conservation valu o r E 0 46 3 UpperForest ForestTypeB DT r T T T T 1 o 20 40 60 80 100 Time Sustainability ratio r T T r T T r T r 1 o 20 40 60 80 100 Css D Time Optimize 99 Sensitivity of results to uncertainty Action specific results are conditioned to individual or groups of actions and shown in a tree that groups actions into environments where they are possible Figure 4 16 Visualizing results across actions g_RobOff GUI 1 0 0rc3 censored_name second_level_offset home fedemp example setup dam Roboff GUI View Results Tools Help eu vo o e 4 Setup Results Summary Time Uncertainty Environments Features Env Features Actions Compare Actions Budget V Robust NE Opportunity NN Nominal mme Features 9 Add O Strong Solid 2 amp Weak Dashed Clear Env Action Conservation value 0 49 4 Mandatory a Preset 0 485 J Optimal vea Dy fedi P B3 MiddleForestA X75 3 MiddleForestB do nothing Z 0474 Restore 4 v Bl UpperForest HG do nothing
17. Specification of aims ecological model and data 6 1 4 2 Getting a RobOff analysis running ssseeeeee 8 1 5 Software installation and quick start sssssssseeeeen 9 1 6 Major Teatile5 ettet bessere Edge ne sae sadsisavediendecs 13 Chapter 1 Introduction 1 1 Aims and purpose RobOff is a framework and software for conservation planning It can manage alternative conservation actions and their uncertain effects on biodiversity features in different environments through time the costs and feasibility of actions budgetary constraints time discounting and robustness requirements The RobOff framework and software are specifically intended to complement the many methods that are most appropriate for spatial reserve selection based on static biodiversity patterns see Section 2 7 Dealing with connectivity p 38 In RobOff actions produce different uncertain responses for features in different environments through time The focus is specifically on time and uncertainty and biodiversity pattern is therefore explicitly assumed to be dynamic To allow for the complexity of problems that arise in habitat maintenance management restoration and offsetting we make the simplification that any explicitly spatial aspect of analysis is dropped although workarounds for this simplification are provided RobOff analyses are intended to answer questions about how much
18. The discounting weights allows easy exploration of different types of preferences like equal weights throughout time final discounting no discounting etc The default value for this option is all the weights equal Note that these weights are applied in addition to the discounting factor which is by no means replaced by the discounting weights For the details on how the discounting factor is calculated please see Section 2 5 4 Time discounting p 33 Note that the length of the sequences of time discounting rates and time discounting weights must be consistent with the number of time intervals considered in the planning problem that is the number of rows in the response files If the length of the sequence is too short RobOff will report an error If it is too long RobOff will issue a warning message 3 3 7 Budget allocation file Here is an example of budget allocation file that includes mandatory and preset allocations of resources 65 Set of files score features Budget allocation file One line per environment with allocated actions Environment action area_allocated mandatory allocation environment 1 development action 8000000 environment 2 development action 2nd 800000 preset allocation A few possibilities environment 2 benign neglect 2nd 2000000 environment 1 active restoration management 2000000 environment 2 active restoration management 2nd 6000000 An allocation i
19. This setup is already complete and consistent You do not need to know how to define it or change its core components If you want to inspect the setup contents you can do so by editing the general settings file with a plain text editor From this file you can see what other files are included in the setup Run the RobOff command line tool by using the s or setup option and the general settings file of the first contact setup Inspect the results generated in the output directory Several files are generated see Section 3 4 Standard RobOff output p 70 for details Some results are only generated if certain command line options are provided to the command line tool This includes the uncertainty analysis optimized allocation of resources pairwise comparison of actions and the sensitivity to budget availability See Section 3 4 1 Optional output files p 75 for details on the available options and the optional files that are produced To obtain an optimal allocation of resources use the optmize alloc Or p option You will obtain the corresponding optional file 7 optimzal allocation csv see Section 3 4 Standard RobOff output p 70 for details This will use the optimization options defined in the setup Try different optimization alternatives by modifying the optimization options of the general settings file of this setup For details on these options such as optimization criterion and optimization method 6 4 A mi
20. Tks Oy We Qu 535 Wn O55 52 Panel I Piecewise linear Feature SOX Aly Sy Wy WA Orsi abe aly Oye Oasys dict MO 225 3 3 4 Set of files biodiversity features present in an environment For every environment a file must be provided that contains a listing of the biodiversity features that are present in the environment These files specify not only what biodiversity features are present but also their specific responses to actions A file of this type must be provided for every environment that appears in the environments file see Section 3 3 2 Environments file p 58 If you forget some of them RobOff will detect it and report the error Every row or line specifies characteristics of a different feature Features not included in a file of this type are considered not to be present in the corresponding 62 Set of files responses of biodiversity features environment If you forget to include a feature in a per environment file RobOff will not be able to detect your error The per environment biodiversity features present files must be named according to this convention concatenate the environment files prefix with the name of the environment and the extension csv by default For example if the environment files prefix option is set to feat env and there are two environments named forest and lake then two files must be provided eat env forest csv and feat env lake csv Here is an example
21. benefit function 92 biodiversity feature 90 cost 92 environment 90 First contact 120 main window 87 optimization 103 output 96 preferences 104 response 90 results 24 96 score feature 92 setup 89 time discounting 92 visualization 24 97 GUI features see RobOff H Habitat 21 21 110 degradation 21 Habitat suitability 29 Horizon of uncertainty see uncertainty Hyperbolic see Time discounting Inadequate data 82 Info gap 25 Input mandatory 49 optional 65 Input files 49 biodiversity features 62 90 budget allocation 65 92 costs 68 92 environments 58 90 feature weights utility functions 59 92 responses of features 63 90 score features 66 92 time discounting 65 92 Installation 9 Installer see RobOff Interactions 82 115 Interchangeability 30 see also substitutability Inverse sigmoid see Benefit function 131 J jpeg see format jpg see format L Land ownership 21 Land use 112 Library Open icon 83 Qt 83 Qwt 83 Linear see Benefit function Linux see Operating system Location file see File path Lower envelope 5 30 113 Maintenance 21 see Action Management 21 see Action Mandatory actions 48 see also actions Mandatory input files 49 Marxan 19 38 Marxan with zones 38 Minimax see optimization Multi core 83 104 Multiplier 114 N Negated sigmoid see Benefit function Nonlinear 35 O Occurrenc
22. conservation and or development actions costs discounting rates and scored features Typically each of these types of entities are edited in separate tabs This part of the RobOff GUI allows for defining setups without the need to edit input text files 87 Main window In the results section results can be visualized across different dimensions Note that it is not required to perform any optimization step in order to visualize results for mandatory and preset resource allocations In the optimize section it is possible to find optimal allocations A number of optimization methods and options can be selected Figure 4 1 RobOff main window E a o ES RobOff GUI 1 0 0rc4 Dam example second level offset home fedemp example setup dam river Roboff GUI View Results Tools Help eD j vo e 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring Global parameters Time planning horizon C File names and prefixes Name second level offset Regular intervals Robust offsetting Mi D Custom intervals ix Environments 5 2 f TA From o Features st 12 2 To 100 IS Rescale responses Time discounting Step 10 Discount rate 3 50 2 Uncertainty hori Budget limits ncertainty horizon ne Total budget 100000 00 Value 1 00 T Range Preset budget lo 00 From 0 00 Economic discount
23. csv files that are generated into the output directory These files include notices about suspicious events detected throughout the RobOff calculations as well as warnings and errors regarding inconsistencies in input data Other output files generated by RobOff into the output directory include information about the evolution in time of conservation value ratios of sustainability and optional results for uncertainty analysis optimization comparison of actions optimized allocation of resources etc These are further described in Section 3 4 Standard RobOff output p 70 3 2 1 Sets of actions When using RobOff for resource allocation optimization it is important to keep in mind that three different sets of allocations of resources to actions are considered in the RobOff software An allocation is simply an amount of area where an action is undertaken in a given environment As such individual allocations are specified as three parameters environment action and amount area extent The allocation sets are Mandatory These are forcibly included actions A typical example would be unavoidable development actions Another possible case is actions for which resources have already been allocated beforehand Preset This set is considered for convenience The resource allocations included in this set are added to the mandatory allocations before performing any optimization This allows the user to for example explore what is the im
24. in general the correspondence between habitats and environments can be complicated There are multiple criteria to define environments differences in the set of possible actions differences in the features that are present or differences in the responses of features to actions connectivity considerations condition or level of degradation different threat levels land ownership issues borders of administrative units or other socio political differences different degrees of uncertainty etc Biodiversity features can be any of those typically considered in conservation planning including species vegetation classes habitat properties communities genes alleles surrogates socially relevant factors etc These can be defined as simple features or as components for scoring and can be present in different environments at different levels Environment specific per area unit responses of features to actions are represented as three time dependent values estimated upper scaling envelope and lower scaling envelope Actions in environments are defined by their costs the availability of areas where they can be performed and their implications for features environment specific responses An action can mean any activity or intervention or combination of activities related to either development or conservation The term conservation action is used here to denote multiple and heterogenous types of actions in diverse contexts such as manage
25. try to generate the opportunity optimal solution and compare them c Perform an uncertainty analysis You need first to define the range of variation of the uncertainty horizon The output of this analysis will show the sensitivity of results to the degree of uncertainty d Compare pairs of actions with the actions comparison option or tab in the graphical interface This comparison will be shown for the range of variation of the uncertainty horizon previously defined e Analyze the sensitivity of results to the available budget To this end you need to define a range of available budget Interpretation and post processing of outputs Note that RobOff is spatially implicit Explicit spatial allocation of actions would require to use additional tools such as Zonation The optimal allocations produced by RobOff can be seen as area targets which can be used as inputs to spatial planning tools see Chapter 5 RobOff analysis setups for common planning needs p 109 Evaluation of proposed resource allocations Quite often at least part of the available conservation resources have been allocated a priori If you need to evaluate particular predefined allocations it is possible to include them as mandatory and or predefined actions with certain amounts allocated to them You can also use this feature to analyze how the optimal solutions allocations change when you forcibly allocate resources to some actions e g following expert re
26. 1 1 0383 Strong features 0 96315 1 1 0383 9 Optimize E a Setup validated Found 1 environments 1 features 1 presences 3 responses and 1 actions subject to optimization Budget allocated 0 Set of actions of this setup succesfully validated Everthing seems correct B Setup home fedemp roboff example setups setup minimal toy offsettina toy offsetting ro_setup successfully loaded Log D 6 5 A simple biodiversity offsetting example Dam Forest N Note This setup can be found in the RobOff software distribution under the folder directory setup dam forest offset This setup considers 2 environments 2 biodiversity features 1 per environment 2 environments 1 feature and 2 actions per environment The general settings file DamForest ro_setup contains the following definitions and options 124 Dam Forest RobOff setup saved automatically Edit at your own risk name Dam_example_offset_analysis robust offsetting 1 environment types 2 allow overwriting 0 planning horizons 0 10 100 info gap alpha 1 info gap alpha range 0 0 1 1 time discounting model quasi hyperbolic time discounting rate 3 5 feature responses 5 environments file environments environments files prefix features_present_ budget allocation file budget_allocation These ones are not needed unless you don t like the default names response files prefix response_ feature wei
27. E 1 03 Time Optimize Feature specific results are conditioned to individual or groups of features globally across all environments and shown in a list of features 98 Visualizing results across different dimensions Figure 4 14 Visualizing results across features RobOff GUI 1 0 0rc3 censored name second level offset home fedemp example setup dam im Roboff GUI View Results Tools Help Demg 2Zvwo o9 4 Setup Results Summary Time Uncertainty Environments Features Env Features Actions Compare Actions Budget Add j V Robust MEE Opportunity NN Nominal ME O Strong Solid M Weak Dashed ForestTypeA Conservation value ForestTypeB ch C id x 0468 2 0 466 Clear Sustainability ratio ooo SSg Sustainability ratio o PPPPPPH e 90000 8S288 o 20 40 60 80 100 Time 9 Optimize Results conservation value and sustinability uncertainty analysis comparison of actions budget anal ysis etc Features within environments Env Features tab are conditioned to individual or groups of features within a single environment and shown a tree that groups features into environments where they are present Figure 4 15 Visualizing results across features within environments RobOff GUI 1 0 0rc3 censored name second level offset home fedemp example setup dam Roboff GUI View Results Tools Help eg zvo o s Q 4
28. Elster 1992 Choice Over Time Russel Sage Foundation New York USA 40 Spatial prioritization and reserve selection Spatial prioritization and reserve selection 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 Proceeding of the Royal Society of London Series B Biological Sciences 272 1885 1891 Watts M E Ball I R Stewart R S Klein C J Wilson K Steinback C Lourivald R Kircher L and H P Possingham 2009 Marxan with Zones Software for optimal conservation based land and sea use zoning Environmental Modelling amp Software 24 1513 1521 Moilanen A Meller L Leppanen J Montesino Pouzols F Arponen A and H Kujala 2012 Zonation spatial conservation planning framework and software v 3 1 User manual 287 pp http www helsinki fi bioscience consplan Conservation resource allocation Kukkala A and A Moilanen A 2013 The core concepts of spatial prioritization in systematic conservation planning Biological Reviews 88 443 464 Moilanen A Wilson K A and H P Possingham 2009 Spatial Conservation Prioritization Quantitative Methods and Computational Tools Oxford University Press Oxford UK Margules C R and S Sarkar 2007 Systematic Conservation Planning Cambridge University Press Cambridge UK Approaches to conse
29. Robust and opportunity performance indices The calculations and ratios defined above correspond to the case where there is no uncertainty such that the uncertainty horizon of the info gap model is a 0 0i OL j and LT In order to account for the sensitivity of the models to uncertainty the representation values in the algorithm above are replaced with their corresponding info gap models derived from their expected values upper and lower envelopes and a variable uncertainty horizon By maximizing or minimizing conservation value in the info gap models it is then possible to calculate maximal conservation values Vj across features V across environments and minimal conservation values V across features V across environments To calculate the robust version of the performance ratios across features V 0 are replaced with Vr 2 U and V gW with Viol 2 U Vig are obtained by maximizing conservation value in the info gap model whereas V are obtained by minimizing in the same interval That is the ratios are calculated between the minimum conservation value that can be obtained within the region of uncertainty around the nominal values for a proposed set of actions and the conservation value that could be obtained if no action is performed Similar reasoning applies to the ratios for environments In the simple case that benefit functions are increasing monotone maximum values are attained for the upper envelope values of
30. SITS ORS pO LOE TER Os VISA Os IZ4SG Ws ISAT Os S355 lib E AE e OSG99S 0 5 SIVES OTS O 208240 TOS 27 7 0 12456 On 19527 OSS oil OSTIA OACII5 O s 3756 A A 0 c 09532 21 0 0722315575 0 79527 A S SN AFA c 22 30 289 0 Ee e ae 0 8 7 2 59 19525 oO 0 470527 On VAESG 05 1IS2 TO iso O gt WZ1LAS OROSCOPO ES p Qs SX02721 Os TISAI Os 7 22485Y 10 19 21 3 510 e SSS LALO c 22230288 Os GIS O53 166 Results for robust robustness 0 70s DOG On VIZ On AAG AW VISA O O ATA 5 72MM a I9 S 04 S719 0 soa po O PR RETA E 0 97 2 3 Os WIS OGIO Z5 06 TALS OSs VS 05 Sel VAS 6 optimization csv The results of the optimization of resource allocation analysis p or optimize alloc command line options are written into this file This file contains information about optimization criteria the optimization algorithm used etc Roboff optimization results Method used Exhaustive local search Budget resolution 3 Trials calculated 3262623 Budget spent in optimal allocation 1 80178e 08 Optimal allocation see file 7 optimal allocation csv can be visualized in the optimization section of the GUI Tie ce cie ce c db 7 optimal allocation csv This file is in the same format as the budget allocation input file see Section 3 3 7 Budget allocation file p 65 it lists the amounts allocated to different actions and can actually be used as an input allocation file The file is generated if the opti
31. Some of the concepts mentioned here may not be clear at this point if you are using RobOff for the first time They are described in depth in Chapter 2 Framework Methods and Algorithms p 17 These general results consider all the features defined and are produced in three variants All actions and amounts allocated No actions at all do nothing or business as usual scenario A subset of mandatory actions which can include for instance planned development actions that are unavoidable In addition specific results can be obtained by conditioning the three types of results above to a subset of environmental types features or actions 1 4 A typical RobOff work flow RobOff has a broad applicability to protection management maintenance restoration and offsetting Two parts can be distinguished in a typical RobOff workflow First the broad aims of analysis need to be define Then specific RobOff setups have to be prepared to match the analysis aims 1 4 1 Specification of aims ecological model and data The first step in conservation planning is to define the aim of the computational analysis Next one should gather together and organize the required data into one or several RobOff setups In practice most of the time invested in a planning Specification of aims ecological model and data project will likely be used to clearly specify the aims and collect the relevant data while the computational analysis would t
32. be found in the summary dialog of the results section of the RobOff GUI 71 Standard RobOff output Table 3 1 Example summary of results conservation value Robust Nominal Opportunity No action 0 6884 0 7715 0 8226 PE a Mandatory actions 0 6012 0 7459 0 8407 Optimal actions 0 8407 0 9462 1 017 Table 3 2 Example summary of results conservation performance ratios Robust Nominal Opportunity Weak 1 072 1 198 1 425 environments Conservation Strong 0 9655 1 001 1 153 performance environments ratios Weak features 1 090 1 226 1 486 Strong features 0 6863 0 9974 1 000 4 Roboff summary of results Info gap uncertainty horizon alpha 1 range from 0 to 1 Conserv value discounted weak across features Minimum Nominal Maximum No action 0 9047 0 9126 OR SET T Mandatory actions 0 8129 0 8198 0 8243 Optimal actions OS S95 0 8435 0 8474 Conservation performance ratios discounted Robust Nominal Opportunity Weak for environments 0 9148 0 9242 073370 Strong for environments 0 8193 0 8279 0 8423 Weak for features 0 9148 0 9243 Qo SD Strong for features QOIS 0 8279 0 8423 Mandatory Allocation Environment 0 MiddleForest area 500 0 200 0 Action 0 do nothing 300 0 Action 1 Inundate 0 000 Action 2 Restore Environment 1 UpperForest area 3 000e 04 3 000e 04 Action 0 do nothing 0 000 Action 1 R
33. before the second comma at the end of the character string the string will be trimmed to restoration expensive This includes normal spaces and tabs Note that spaces in between words matter The names restoration expensive single space in between the words and restoration expensive double space in between the words are different Similarly restoration expensive is different than restoration expensive This type of file is related to the option environment types of the general settings file The number given in that option should be consistent with the number of environments rows specified in the environments file A Note In RobOff input files there is no explicit distinction between conservation and development actions Normally conservation actions would have positive costs whereas development actions would have zero cost and be specified as mandatory Development actions could also have negative costs However their definition and interpretation is up to the user 3 3 3 Feature weights utility functions file For every feature included in a RobOff setup you need to specify its weight and benefit or utility function see Chapter 2 Framework Methods and Algorithms p 17 for more information on the benefit function approach if you are not familiar with it To see how to indicate that biodiversity features are present in different environments see Section 3 3 4 Set of files biodiversity features
34. conservation value 96 copy to clipboard 96 104 dimensions 97 robustness requirement 96 save as 96 104 sustainability 96 uncertainty 100 png see format Portable document format see format Portable network graphics see format postscript see format Preference see Time preference Preferences see GUI see RobOff Preset actions see Action Process based planning 39 82 Protected area network 41 Protection 21 see Action Q Qt see Library Quadratic see Benefit function Quasi hyperbolic see Time discounting Quick start 9 Qwt see Library R Randomization 36 36 55 see Optimization References 40 Representation 29 see Aggregation Representativeness 82 Reserve design incremental 48 Reserve selection 41 Resolution see Budget Resouce allocation optimization 35 Resource allocation see Allocation Response 26 29 53 63 63 editing 90 end value 56 envelope 64 GUI 90 preprocessing 63 rescaling 56 scale 56 57 63 start value 56 Responses scale see optimization absolute 56 57 proportional 56 57 range 56 57 Restoration 21 114 see Action Results summary 71 visualization 96 97 Retention 29 Return on investment 77 102 RobOff see results analyses 4 112 applications 17 assumptions 39 binaries 83 command line 46 computation time 39 conceptual diagram 22 data 4 dimensions 24 education 18 features 4 87 flow of use 111 framework 4
35. eaa Esi 48 3 3 Input files and settings ese 49 3 3 1 General settings file sesssssessessss 50 3 3 2 Environments file sssini a 58 3 3 3 Feature weights utility functions file 59 3 3 4 Set of files biodiversity features 62 3 3 5 Set of files responses of biodiversity features 63 3 3 6 Time discounting file sseeeeseeeeees 65 3 3 7 Budget allocation file seseesees esses 65 3 3 8 Set of files score features seesssssss 66 3 3 9 Set of files costs of actions sssssss 68 3 4 Standard RobOff output sssssssseee 70 RobOff 3 4 1 Optional output files eee ee ceeteeeeeeeeeeeeaeaaee ees 75 3 5 What RobOff does not do directly ceceeeceeseseeeeeeeeeeeeaee 82 3 6 Implementation details about RobOff ssseuuusss 83 3 7 Data limitations and system requirements 83 3 8 Troubleshooting ccccceeeeeeeeeeeeeeeceeeeeeeeeeeeeeseeeeeeeeaeeees 84 IV RobOff Graphical User Interface ssessssssessene 85 4 RobOff Graphical User Interface sssesseeeee 87 4 1 MairiWIDIdOW 5 sists ais hide CH eee ea ie ert abe cedi ere eh axes 87 4 2 Setup eee est ce o Ge em ed e bo peres 89 4 3 RESUS
36. environments For references on the RobOff methods and related literature see Section 2 9 References p 40 The results produced by RobOff are in short of two main types Quantitative evaluation of the outcomes of conservation actions over time measured as conservation value or sustainability indices Optimal allocations of resources to multiple alternative actions Using RobOff requires information on the responses of biodiversity features to different alternative actions whether from data warehouses models or expert knowledge It is also required to select different criteria and parameters related to preferences and human decision making RobOff has been designed as a framework of very broad applicability It can be applied in very diverse contexts where goals and terminology may be different but all of which share a common problem structure Very different types of planning problems can be conceptualized within the RobOff framework Some examples are summarized in Table 2 1 Typical or suggested usages of RobOff p 18 These possible and suggested usages are divided into three main categories the first two of which are education and problem clarification These are key considerations since RobOff is a novel framework that complements and differs significantly from traditional reserve selection and related tools An important component of problem clarification is what can be gained from setting up a RobOff analysis deve
37. is related to the concept of stock In RobOff the term occurrence level is used to denote the local occurrence values of features i e per area unit values Representation Occurrence levels integrated summed across the environment or all environments where the feature occurs Conservation Benefit converted from representation via a benefit function value transform such as convex increase with diminishing returns or target functions In the context of ecosystem services conservation value would correspond or be related to the concept of flow Retention In conservation biology one original definition of retention is what remains in the landscape without conservation action Here we use retention in a slightly expanded sense meaning what remains of a feature in the whole landscape under a certain set of actions Note that retention applies to representation and conservation value Substitutability In ecological economics these terms refer to the potential for interchangeability the susbstitution of certain forms of capital for different forms 29 Aggregating conservation value Term Meaning in RobOff of capital The major forms of capital are social human natural and manufactured In particular it is an important debate to what extent natural capital can be substituted for man made capital This concept is key for differentiating between weak and strong sustainability Weak This t
38. large problems in reasonable time We can distinguish two types of outputs 1 evaluation of the outcome for a particular allocation of resources to different actions and 2 optional allocations of resources i e a list specifying the quantity of what resources to allocate to what A typical RobOff work flow actions In the first case one could evaluate e g a business as usual scenario an allocation of conservation resources recommended by experts or an optimal allocation of resources found by using RobOff second case In the second case RobOff finds an optimal allocation of resources according to different criteria including time preference degree of uncertainty and robustness requirements When evaluating an allocation of resources RobOff generates diverse outputs in two complementary forms as text tables of values and plots in a graphical interface The general output of RobOff comprises the following elements Development through time of conservation value and the weak and strong conservation performance ratios across features and or environments and for the nominal robust and opportunity cases These curves are obtained conditioned on a particular uncertainty horizon for a given time discounting model and rate Time discounted conservation performance ratios as a function of the uncertainty horizon Optimal division of resources between any two given actions as a function of the uncertainty horizon
39. notice that many lines in the following example files start with a sharp or number or pound symbol These are comment lines which 49 General settings file are ignored by RobOff and you can use freely Anything written after a is ignored by RobOff You will also see comment lines like these in output and setup files generated by RobOff Note All these input files must be written as plain text files In windows systems their extension would typically be csv although this is not strictly necessary In principle RobOff imposes no limitation on the names of these files Their length is not limited and any character set can be used as long as it is supported by the operating system However there is a risk to run into trouble with other software or your operating system if you use disproportionately long or not so common characters including spaces accents parenthesis etc 3 3 1 General settings file A example general settings file rcoboff setup in the example command line above is shown below This file defines all the main properties of a setup but most of the information is actually provided in additional files Each option must be set in one line using the character to separate the option name and its desired value The definition list given below specifies the meaning and possible values for every option To start using RobOff you do not need to know about most of these The comments included in the exa
40. of what kinds of conservation measures should be allocated to which environment types In this role these results provide an important tool for target setting for systematic conservation planning The relative simplicity and flexibility of RobOff should be emphasized It is key that RobOff builds on uncertain discrete time responses that can be coded into text files with three columns of numbers Not only is this basic building block very easy to understand it is very flexible as it allows operation on responses that have no predetermined specific functional form The uncertainty in responses to actions can be derived as a combination of statistical information if available and expert opinion While setting up analyses using the proposed approach requires specification of a potentially significant amount of information the very act of collecting the information and thinking about the resource allocation problem will be constructive in terms of understanding the ecology conservation objectives and limitations of the particular planning case Specifying the problem makes the decision process transparent to those who utilize the results The RobOff conceptual model has been defined in a way that reduces the amount of information required from the user while still accounting for a number of important factors Nevertheless the model is sufficiently flexible so that domain specific subtleties can be effectively incorporated In summary the a
41. of conservation value or sustainability careful evaluation of the consequences of conservation resource allocation requires more in depth analysis and consideration of multiple factors and viewpoints The outputs generated by RobOff reflect this complexity In general the RobOff output space can be seen as the Cartesian product of multiple and heterogeneous dimensions as shown in Figure 2 3 Dimensions of the RobOff output space p 24 Some of these dimensions are directly related to ecological entities such as environments and biodiversity features while some others lie in the domain of preferences values and human decision making such as different time perspectives different views on substitutability robustness requirements or alternative variants of aggregation of value across environments or features 23 The RobOff output space Figure 2 3 Dimensions of the RobOff output space Adapted and expanded from Pouzols amp Moilanen 2013 Dimensions mainly related to values socio political factors and human decision making e Variant of Robustness Time Substitutability X adoreadiign X Allocation ees Across features Across No action Robust Mandatory Discount Preset Opportunity rate scenario Expected environments Optimal Nominal Dimensions mainly related to ecological entities nwronments X Features X UM X Landscape spatial scope Planning Info gap horizon alpha amp discou
42. reduced dimensionality greedy local search can be useful for quick tests but it is not recommended at all Finally random will randomly allocate fractions of the budget to different actions trying to allocate as much as possible of the total budget Its results can be used as a reference for evaluating the outcome of other optimization methods Note 54 General settings file optimization criterion optimization robustness requirement economic discount rate economic discount model enable rescaling of responses that a random allocation may generate worse results than if no resources are allocated at all In contrast all the other methods will provide results that are equal or better than the starting point Optimization criterion or performance index that should be maximized by the optimization process The options are weak features strong features weak environments strong environments These options correspond to the different ratios of sustainability defined in the RobOff computational model see Section 2 5 Aggregation of conservation value p 25 Optimization robustness requirement Options nominal opportunity Or robust Default robust see Section 2 5 Aggregation of conservation value p 25 and Section 2 6 Optimizing resource allocation p 35 Specified as a percentage Example economic discount rate 2 5Ssetsa discounting rate of 2 596 per unit of time
43. representation For decreasing monotone benefit functions similar reasoning can be applied However in a more general case these quantities must be calculated by maximization or minimization in an interval Similarly for calculating the opportunity version of the performance ratios V f is replaced with V A a U and V eo with V oft a U which likewise require minimization and maximization in intervals Robust weak sustainability across features nf by jo Nec mp Fats go j1 34 Optimizing resource allocation Robust strong sustainability across features V 0 feat J N strong mr go E Robust weak sustainability across environments Erio env il N rO Ne wea x Ly Lro il Robust strong sustainability across environments vi Va N strong min Finally opportunity indices are calculated as ratios of maximal value for a proposed allocation of resources against minimal values for the business as usual allocation 2 6 Optimizing resource allocation In RobOff aggregation of conservation value is done across environments features and time Once a RobOff setup has been established it is not trivial to find an optimal allocation of resources for the different actions available In this case optimal allocation refers to an allocation of resources such that the conservation performance ratios defined above are maximal given all the required parameters and choices of crite
44. response file setups setup simple demo response_r_6 3 bt Regenerating uncertainty graphs for uncertainty parameter 1 Regenerating uncertainty graphs for uncertainty parameter 0 8 Setup setups setup simple demo successfully loaded E OpenMP support there are 4 threads in parallel regions Q Tip By default the plots of the results section will be updated every time a correct setup is loaded into the GUI The update is done by adding the new curves to the curves already displayed By loading a sequence of setups and changing the plot properties color and line style you can build plots that display results for multiple setups simultaneously You can restart the process at any time for any of the result plots by using the clear buttons This way it is for example easy to generate plots of the sensitivity to uncertainty for different setups or variations of a same setup for different resource allocation options see next sections 4 3 1 Visualizing results across different dimensions The RobOff output space can be seen as the Cartesian product of multiple and heterogeneous dimensions see Section 2 4 The RobOff output space p 23 In the results section of the GUl itis possible to visualize the evolution of conservation value and sustainability performance ratios over time in the following tabs time environments features and actions On the left a tree of environments features or actions normally grouped
45. sustainability ratio per features 05 101 5 2 9923 3L 0 O27 OP ORS ZB EnO SANAB ZIP OO ZITPN US Oy S03 60 05 199527 kp 0 20532 30707052 Tr Al 0 AS SIA EEO ONO c 1193532 2 7 1 5 00 rnc oak 13 sustainability ratios per environment csv Format time sust ratio featurel sust ratio feature2 sust ratio last feature Similarly to the previous file per feature sustainability ratios this file is analogous to the file 4 sustainability ratios csv but contains sustainabiliy ratios for individual environments as listed in the file header This file is generated only if the command line option e or per env results is used Here is the beginning of an example file 80 Optional output files Roboff results sustainability ratios per environment All values are discounted Robustness requirement robust Info gap alpha 1 Format time sust ratio environmentl sust ratio environment2 sust ratio last environment Names of environments MiddleForest UpperForest MiddleRiver UpperRiver LowerRiver SH oce HEHE db Results for No action sustainability ratio per environments Oey ples eile Z9 abr abr dE corsi dE bOr tyt SOIA lp pile ial sl 1 1 1 1 1 100 1 1 1 1 1 1 i ab sustainability ratio per environments time discounted Oy iby ab pike L7 dl gor b bod 219 716 SOE SAOP papal pig yak 1 1 1 rl 1 1 100 1 1 1 1 1 1 1 1 Results for mandator
46. time step all conservation values will be heavily discounted as obtained more than two thousand years ahead of present time and most likely results will be nonsensical time discounting model Defines the discounting model that should be applied Together with the discounting rate see below this defines the relative weights of conservation value over time see Section 2 5 4 Time discounting p 33 The following options are available quasi hyperbolic hyperbolic and exponential Default quasi hyperbolic time discounting rate Time discounting rate is a percentage Example time discounting rate 2 5 sets a discounting rate of 2 5 It is also possible to specify a sequence of discounting rates over time variable discounting rate see Section 3 3 6 Time discounting file p 65 for the details environment types Number of environment types defined This optional parameter is used to check the consistency of RobOff setups RobOff will check whether this number is consistent with the total number of environments defined in the global environments file and whether all the corresponding environment specific presence of biodiversity feature files can be found feature responses Number of feature responses defined This optional parameter is used to check the consistency of RobOff setups RobOff will check that this number is consistent with the total number of different responses that are used in the environ
47. versus long term goals Influence of Expected value of decision versus associated uncertainty uncertainty analysis of opportunities and robustness or best and worst on decisions case sensitivity of decisions to uncertainty Implications of different forms of sustainability Dependence of decisions on tradeoffs and different views on substitutability between biodiversity features and environments Management oriented analysis complicated level Balancing of restoration options Balancing budget allocation between actions to generate an outcome that is balanced and efficient across features and environments Biodiversity Analysis of best budget allocations to compensatory offsetting actions that robustly offset in a cost efficient way specific losses caused by development actions Extraction Focus on finding optimal budget allocation which is of targets converted into targets for spatial systematic conservation planning The types of analysis defined in the table as management level analyses have relevance for decision making either via direct generation of management recommendations or via investigation of the robustness of decisions to assumptions We called simple management level analyses those that focus on the influence of one factor on the decision Complicated management level 18 The RobOff framework analyses would include allocation of habitat restoration offsetting and extract
48. we denote the aggregation of the environment specific conservation values of all the features present in environment E with V the performance ratios for environments are as follows yp NOD j zd x Nweak Yo i 1 Strong sustainability across environments yo Volt D N strong minr Note that strong sustainability as calculated here can be seen as a variant of minimax optimization where minimum retention is maximized or maximum loss is minimized 2 5 4 Time discounting Evaluation of offsetting situations and conservation action in general should also consider when compensation is attained To this end RobOff integrates various models for time discounting For any of the considered criteria weak or strong and variants across features or across environments the last stage in the conservation value aggregation process consists of applying a time discounting model which produces a scalar conservation performance ratio Ysnvariant 9 Bee PIQUE criterion N discounted 7 Va Q t 33 Robust and opportunity performance indices where r are discount factors that follow a given discounting model The most common models are hyperbolic quasi hyperbolic and exponential all of which are supported in RobOff For a concise review of time discounting models please refer to Green and Myerson 2004 These depend on the discount rate parameter which has to be set depending on time preference 2 5 5
49. 0 0 I Inundate 0 20 0 UpperRiver L 3200 0 0 l ObSEruct 0 200 0 ObstructElusTI 2 0 200 0 LowerRiver 1 O0 OW i Alter row 0 ESCORT The feature presence files for the river environments are as follows RobOff file of biodiversity features present in an environment MT feature name present estimate not used response no action list_of_quartets action response init value env value accounts for effects of increased flow variability downstream of the dam due to fluctuating generation River 0 r11 CC env LowerRiver act business as usual AlterFlow ri2 CC env LowerRiver act AlterFlow 0 0 127 Dam Forest River RobOff fille of biodiversity features present in an environment MT feature_name present_estimate_not_used response_no_action list_of_quartets action response init_value env_value River 0 r6 CC env MiddleRiver act business as usual Inundaete r7_CClenv_ MiddleRiver act_Inundate 0 0 feature_name present_estimate_not_used response_no_action list of quartets action response init value env value River 0 r8 CC env UpperRiver act business as usual Obstruct r9_ CC env _ UpperRiver_act Obstr ct 0 Obst rust Pl sT r10_CC_env_UpperRiver_act_ObstructPlusTT 0 0 Remember that the responses of features to actions and all the entities required to define a RobOff setup can be created from scratch and edited in the RobOff GUI The following screenshot
50. 000 500 1 0 development action 0 150000 benign neglect 400 15000 active restoration management 1000 15000 environment 2 1 0 25000 2000 1 0 development action 2 0 150000 benign neglect 2 200 20000 active restoration management 2 800 15000 N Note Note that fields are separated by commas within each line of the environments files like in normal comma separated values csv files This is a general rule in RobOff input files Note that the example file contains two effective lines but these are wrapped because they are too long for this manual In the input text files the specification of each environment should go in a single line In addition to the environment name you need to specify a weight the total area the area that is not available for any action the condition typically but not necessarily in a scale from 0 to 1 and a 58 Feature weights utility functions file list of actions that are possible For each action three parameters must be given the action name its cost per unit of area and the total area extension where the action can be undertaken available area N Technical note about character strings In comma separated files spaces at the beginning and at the end of strings are ignored by RobOff String fields will be automatically trimmed For example if you type restoration expensive note the spaces after the first comma at the beginning of the character string and
51. 0462 E 0 455 3 r T T T T r T r T T 1 o 20 40 60 80 100 Time Sustainability ratio o 20 40 60 80 100 Lx D Time 9 Optimize 4 3 2 Sensitivity of results to uncertainty In the uncertainty tab it is possible to visualize results against the uncertainty horizon or degree of uncertainty This is the core of the anlysis type uncertainty see Part V RobOff analysis setups for common planning needs p 107 for more details 100 Comparison of alternative actions Figure 4 17 Visualizing results as a function of the degree of uncertainty Roboff GUI View Results Tools Help ego vo o S 4 Setup Results Summary Time Uncertainty Environments Features Actions Compare Actions Budget Analyze Strong environments Ml Solid i Add Clear 124 1 Uncertainty horizon Info gap alpha 4 3 3 Comparison of alternative actions In the compare actions tab it is possible to visualize a special type of result trade offs between alternative actions as a function of the degree of uncertainty Two alternative actions can be selected in the left and right trees respectively For convenience these trees group actions into their respective environments The pairwise actions comparison plot shows on the vertical axis the percentage of resources allocated to the action selected on the left as compared to the amount of resources allocated to the action on the right as a functio
52. 082 0 9082 0 9082 0 9082 0 9082 SO 00 0 SOWA 0 80120 3072 Uo M072 0 S012 0 GOA 0 D012 0 D002 0 00125053012 On 902 90700 0791016271090 627 0 9016270 90 6277 07 91016 27 0 9016272090 627 0 590 62 079062710 901627201 90 62 IMO 50 0520525 053052 0 9052 0 J0s2 Wi SOSA Oss 0 0520 90S4 0 0S4 OA SOS 27 0505 Min conservation value across features CO NON 0 SL 08 0190 270180 71 4 22 0 fey 10 3 Chale ORTOS QTIORTUZ2 NOSOP iste abe 0 iS aka 2 10 Chal IO 5 OO SALTO 0 OXINIEO 0 SL ALGO 0 Sal LG 0 SLI Sak Wy SALI ORCI BITES RT 1 Obs Shih 0 4 SL LO These matrices are generated for each different set of allocations in the following sequence no action no allocation of resources at all mandatory preset and optimal For each of these options three robustness variants are considered nominal average opportunity maximal and robust minimal In addition for each robustness criterion the matrix of average weak values is written first followed by the matrix of minimum strong values the matrix of discounted average values and finally the matrix of discounted minimum values Therefore the file can contain up to 48 4x3x4 matrices Note that we use the strong weak terms applied to conservation value in a similar sense as in the performance ratios Weak strong conservation value across features is equivalent to average minimum conservation value across fe
53. 1 General settings file p 50 for details Environments list file see Section 3 3 2 Environments file p 58 for details For each environment a file with the list of biodiversity features present see Section 3 3 4 Set of files biodiversity features p 62 for details Files with responses to actions see Section 3 3 5 Set of files responses of biodiversity features p 63 for details File of feature weights utility functions see Section 3 3 3 Feature weights utility functions file p 59 for details In the graphical interface there are different tables where all these entities and parameters can be entered interactively N Note In numeric fields whether in the graphical interface or input text files please use dots as a decimal mark or separator If your country locale uses commas as the decimal separator please avoid them and use just dots Normally commas will be detected as errors but there could be unexpected consequences which might be difficult to spot Time preference or discounting model and rate and degree of uncertainty and range of variation are probably the most relevant general settings and should be carefully considered When you have at least a first guess at these and other settings You can use the RobOff graphical interface to a enter them or b load a setup from files Once you are able to define a complete setup you will typically be switching between the three mai
54. 1 0 25 8 1 1 0 25 9 1 2 10 1 3 0 25 1 i 1 o ai 1 1 0 25 12 New featur l 1 0 25 13 New featur 1 1 0 25 14 New featur 1 1 0 25 15 New featur 1 1 0 25 16 New featur 1 1 0 25 Results Optimize Setup Figure 4 8 Editing costs in the RobOff GUI RobOff GUI 1 0 0rc4 censored name_second_level_offset home fedemp example setup dam BBA Roboff GUI View Results Tools Help DABitve o s 9 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scorin g P g g Actions Type of cost Env Action Cost Constant cost per area unit v G MiddleForestA Cost area curve di thi e nouang Time varying cost Inundate Y G MiddleForestB do nothing o Interpolation Piecewise linear 2 S mmm v W UpperForest f do nothing oj 10 3 25 Restore SE 20 1 2 24 30 0 3 1 40 016 3 154 5 50 0 14 J 1 nes 60 0 13 70 0 19 0 5 80 0 4 1 oJ SSS A Na 20 40 60 80 100 Spatial units Results Optimize Figure 4 9 Editing time discounting model and parameters in the RobOff GUI on Roboff GUI Edit View Tools Help Demg zv o se 92 4 Setup General settings Environments Features Responses Actions Functions Costs Discounting Scoring Discount model Type of discoun
55. 124 127 examples 119 GUI 89 Sharp character 50 Sigmoid see Benefit function Simple feature see Biodiversity feature Site 21 41 Site selection 41 Social factors 21 Spatial allocation see Allocation Spatial planning 5 82 Spatial prioritization 41 Spatial reserve selection 3 41 Species 21 see Biodiversity feature Species richness 29 Species weight see Weight Spreadsheet 47 58 71 Stochastic search 36 Stock see Ecosystem services Strings see spaces see tabs trimming 59 Strong sustainability see Sustainability Substitutability 30 see also interchangeability Surrogates 21 134 Sustainability 32 preference 113 criterion 103 range 52 ecological 30 sensitivity 100 economic 30 Upper envelope 5 30 113 index 96 103 Utility per environment 80 function see Benefit function per feature 79 Utilityfunction see Benefit function ratio 32 34 74 strong 30 32 96 V variant 96 103 Vegetation classes 21 weak 30 32 96 Visualization see results svg see format action 100 Systematic conservation planning 3 allocation 103 alternative actions 101 T budget 102 Target 3 20 113 environment 98 extraction 113 feature 98 99 Threat 21 feature within environment 99 Time 112 see Cost optimization 103 discounting 114 time 97 preference 10 40 114 Uncertainty 100 weights 114 Time discounting 33 40 65 92 W editing 92 Weak exponential 34 53 sustainabili
56. 17 graphical user interface 87 GUI 24 GUI features 87 GUI output 96 inputs 5 49 installer 9 learning 18 library 83 limitations 39 logs 88 133 main window 87 memory requirements 39 operating systems 83 output 70 122 output files 70 output space 23 24 outputs 5 preferences 104 104 results 48 96 running 46 settings 49 setup 20 89 setup components 20 110 stages 111 teaching 18 usages 17 work flow 6 Robustness 34 Robustness requirement 96 optimization ROI see Return on investment S Scalable vector graphics see format Scale of analysis 110 Scale of responses see Response Scheduling 21 83 Scope of analysis 110 Score see Score feature component 66 Score feature 66 editing 92 file 66 92 GUI 92 Scoring 22 23 41 see Score feature Selection area 41 reserve 41 sites 41 Sensitivity budget see Budget uncertainty see Uncertainty Settings budget 54 budget allocation file 54 cost files prefix 57 103 see economic discount model 55 economic discount rate 55 environments 53 environments file 57 feature responses 53 file 8 50 file extension 58 file suffix 58 info gap range 52 mandatory budget 54 per environment files prefix 57 planning horizons 53 preset budget 54 response files prefix 57 score features files prefix 58 time discounting model 53 time discounting rate 53 variable costs 57 Setup editing 89 example
57. 3 1 General settings file p 50 Roboff results Budget analysis Allocation Format budget_level_ money sust_weak_features sust_strong_features sust_weak_environments sust_strong_environments consval_weak_features consval_strong_features consval_weak_environments consval_strong_environments 0 0 89789 0 78124 0 71544 0 0 84837 0 71286 0 69194 0 37573 20000 0 89926 0 78124 0 71624 0 0 84967 0 71286 0 69272 0 37573 40000 0 9006 0 78124 0 71703 0 0 85094 0 71286 0 69348 0 37573 000007090192 0 WIZ 0 5 WTS Oy On SS2m Ol MAS 0 OHS OS SIS HOO O VOSA 0 E24 0 MUSING OM OROSII Oi WIA 0 GWG 0 S73 1e 05 0 90447 0 78124 0 7193 0 0 8546 0 71286 0 69568 0 37573 10 conservation_values_per_feature csv This file is generated only if the command line option f or per feature out is used lt contains conservation values specific to features across time The file consists of several matrices with the columns corresponding to different features and rows corresponding to the different time intervals of the planning horizon indicated in the first column Each of the matrices written in this file corresponds to one combination of resource allocation and robustness requirement as in the 3 conservation values csv file However in this case conservation values are reported only for one degree of uncertainty the preferred degree of uncertainty specified in the general settings file see Section 3 3 1 Gener
58. 4 eia itte Gri rire ied 96 4 3 1 Visualizing results across different dimensions 97 4 3 2 Sensitivity of results to uncertainty 100 4 3 3 Comparison of alternative actions 101 4 3 4 Sensitivity of results to Budget variations 102 4 4 OptUmization zoe oe eh Genere dti i 103 4 5 Preferences ee eee ea p D ER RE ERRARE 104 V RobOff analysis setups for common planning needs 107 5 RobOff analysis setups for common planning needs 109 5 1 General remarks eessssssssseeeenee mene 109 5 2 Analysis typos i rete tee eere aee adn 112 5 3 Uncertalnty 3 ocio eet teet eere reb cen 113 SM 114 b bz OffSetlirig ec e eb tec et coated 114 5 5 1 How much compensation is enough 114 5 6 SINTSFACTIONS ys tie ceo ote rettet tere ho o e Po cete unie ots 115 VI Tutorial and examples see 117 6 Tutorial and examples ece eare ean e e EELER EAEE AA EENAA 119 e ALIN ite a aaa st Stance A EM 119 6 2 A first contact with the RobOff GUI aeee 120 6 3 A first contact with the RobOff command line interface 120 6 4 A minimal example sssesee m 121 6 4 1 Minimal set of input files seseesessssss 121 6 4 2 Output obtained esssssssseeeeeee 122 6 5 DamrForest uiti a i E RU
59. P ERO Ies 124 6 6 Dam Forest River eo eaten ee EE AEE rE AREE EERE 127 MION irese aA S NE ATAA AA AE 129 vi List of Figures 1 1 Example screen capture of the RobOff GUI summary of results 11 1 2 Example screen capture of the RobOff GUI editing responses in the Set p SOCIION ooi ans e n e ean Es 12 2 1 Basic components of a RobOff setup sssssssseeeeeee 20 2 2 RobOff conceptual diagram Adapted from Pouzols Burgman amp Moilanen 2012 mreana e na petet io Pee eie v ee ee 22 2 3 Dimensions of the RobOff output space ccceceeeeeeeeeeeeeeeeeeeaeaaeaees 24 2 4 Flow of aggregation of occurence levels and conservation value in RBODOIT sis seers toeIR xai petimus idm 27 3 1 Running RobOff stages from inputs to outputs ssessssssse 46 3 2 Set of input files and their equivalent GUI dialogs 49 3 3 Example benefit functions 2 00 eee cece cece teeter eee ee ee cena eeee e 61 3 4 Example of per area unit cost as a function of area extent 69 3 5 Set of output files and their equivalent GUI dialogs 70 4 1 RobOff main window ssssssseeeenene memes 88 4 2 Editing a setup general settings see 89 4 3 Editing environments and actions in the GUI ssseesssesssss 90 4 4 Editing biodiversity features in the GUI
60. ROBOFF USER MANUAL Software for Robust Offsetting Habitat Restoration Maintenance and Management Eu version 1 0 4 Federico M Pouzols orn Atte Moilanen z lt RobOff User Manual Software for Robust Offsetting Habitat Restoration Maintenance and Management Analysis of alternative land use options and allocation of conservation resources to multiple actions RobOff User Manual Software for Robust Offsetting Habitat Restoration Maintenance and Management Analysis of alternative land use options and allocation of conservation resources to multiple actions by Federico M Pouzols and Atte Moilanen Technical and language editing Victoria Veach Cover page Aija Kukkala Atte Moilanen Revision 1 0 Publication date March 2012 Copyright 2011 2013 Biodiversity Informatics Conservation Group University of Helsinki ISBN 978 952 10 8720 2 paperback ISBN 978 952 10 8721 9 PDF This user manual and the RobOff software are distributed in the hope that they will be useful but WITHOUT ANY WARRANTY without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE This manual is Copyright C 2011 2013 Biodiversity Informatics Conservation Group University of Helsinki This manual is licensed under the Creative Commons Attribution ShareAlike 3 0 Unported License To view a copy of this license visit http creativecommons org licenses by sa 3 0 o
61. Re e files prefi response features files prefi st function files pre t_unit Benefit function files pre defined benefit function Feature weight functic counting file name i Data files extension suffix cs t ir typ Budget allocation file name budget allocation The core of a RobOff setup includes four types of entities environments actions biodiversity features and the responses of features to actions in environments see Figure 2 1 Basic components of a RobOff setup p 20 in Section 2 2 RobOff Setups p 20 In the RobOff GUI setup section these entities can be created and modified in the following tabs 89 Setup Environments Actions see Figure 4 3 Editing environments and actions in the GUI p 90 The information that can be edited in this section is contained in the environments file see Section 3 3 2 Environments file p 58 Features see Figure 4 4 Editing biodiversity features in the GUI p 91 The information that can be edited in this section is contained in the set of files of biodiversity features present see Section 3 3 4 Set of files biodiversity features p 62 For each environment there is a row in the table of environment and a specific table of features present which correspond to one file of biodiversity features present Responses Uncertain responses can be edited and visualized simultaneously in the responses tab see Figure 4 5 Edit
62. See also the section called Quick start p 10 84 Part IV RobOff Graphical User Interface Table of Contents 4 ron Graphical User Interface eeessessssssssessseeeeeeenenee enne 87 a pBM si scctccaseeciddannccueeas ret dcny EEEE SE ETRA UAE EKRAR AANA 87 42 egre EE 89 Lu mEI cdmm c 96 4 3 1 Visualizing results across different dimensions ssss 97 4 3 2 Sensitivity of results to uncertainty sssssessseeee 100 4 3 3 Comparison of alternative actions eessseeseeee 101 4 3 4 Sensitivity of results to Budget variations ssssssessss 102 Le Eee ERR 103 4 5 Preferences conu etes a E AAAA cessus eo epe dae 104 Chapter 4 RobOff Graphical User Interface The RobOff graphical user interface GUI allows utilization of RobOff without the need to edit input files directly or use the command line interface All the features of RobOff are supported by the GUI as well as the command line If you prefer command line interfaces or need to use commands in automated scripts see Section 3 2 Running RobOff from the command line p 46 Listed below are some of the general features of the RobOff GUI Loading existing RobOff setups Editing and management of RobOff setups Saving modified setups Calculation and visualization of results in different windo
63. WO_env_OMT act_set_aside 110 0 2831 0 066 0 5002 01 r37_TTWO_env_OMT act_clear_cut_ _s na mna am r v 120 0 2688 0 05685 0 4807 l I 130 0 2592 0 0491 0 46925 o mm F EE as m r T T T T T 1 ID o 50 100 150 200 250 300 Results C Optimize amp ra Feature response file 53 home fedemp postdoc hy mrg papers roboff clearcuts example_finland_metso_clearcuts_setup response_r8_CC_env_MT_ai Feature response file 54 home fedemp postdoc hy mrg papers roboff clearcuts example_fintand_metso_clearcuts_setup response_r9_CC_env_VT_ac Found 6 actions subject to optimization in 3 environments D 9 E gt a RobOff outputs can be visualized in its graphical interface but they are also written into text files This way it is possible to inspect results with simple text editors do automated post processing using scripts or other software load results into spreadsheet software etc Output files are generated into a subdirectory folder and includes among other files A Read me file with basic information about the RobOff analysis A summary table of main results for different allocations of resources no allocation mandatory actions predefined allocation and optimal allocation if available A log of how the analysis proceeded including warnings and notes Evolution of conservation value over time Conservation performance ratios i e indices of relative sustainability Alog f
64. a 1 n Biodiversity feature oweight benefit transform Per unit area cost Available area Uncertain response of feature over time nominal upper envelope lower envelope Calculations that apply to these entities and their interrelations utilize additional information concerning the available budget costs of actions availability of areas conservation benefit transformations time discounting model and an uncertainty model The way these computations are performed and the alternative approaches to optimization are described in the next sections Here we concentrate on the input data which are defined as follows 20 RobOff Setups An environment is an aggregation of sites that in practice are dealt with as a block for planning purposes The state and evolution of an environment is calculated based on the features present actions taken and environment specific responses of features to actions In some cases environments can also correspond to sites or conservation units Depending on planning objectives and constraints environments can correspond to habitat types combinations of habitats or parts of a habitat An additional reason to split a habitat into multiple environments is the need to define different classes of connectivity see Section 2 7 Dealing with connectivity p 38 for an explanation Whereas in some planning cases environments can directly represent habitat types
65. ake a relatively short time The steps involved in developing a RobOff analysis are described in more detail below 1 Make a decision on the broad aims of the planning project Is it about evaluating the influence of uncertainty or time preference Is it about evaluating an offset agreement of finding and optimal one What kind of activities will be considered protection management maintenance restoration At first it is also important to give careful consideration to what biodiversity features and habitats are relevant 2 Gather and develop the information related to social and human related factors and uncertainty This includes factors such as costs availability and feasibility of actions resources budget available time preference different criteria to find optimal allocations etc What actions are possible and or compulsory 3 Develop the ecological model which defines a large part of the RobOff setups that will be generated later on What actions are relevant for your planning project What information is available about the responses of features to these actions What habitat types are relevant and what data is available about them what is their spatial extent how much is available for different actions should or can they be managed as one block or is it not possible because of implementation administrative or other reasons These last questions can help in answering whether habitat types can or should be merged or split in
66. al settings file p 50 Features are listed in the same order as they are specified in the feature weights file see Section 3 3 3 Feature weights utility functions file p 59 and their names are shown for convenience in a comment line at the beginning of the file The following listing shows part of an example 78 Optional output files Roboff results feature specific conservation value Format time cons_val_featurel cons_val_feature2 cons_val_last_feature Names of features ForestTypeA ForestTypeB River SH HOHE HE se Results for nominal robustness conservation value per features 0 0 91423 0 91468 0 99924 20 WSO 0 5 Sil We 703 99924 4070 SOMO 27 Or I2 nO 2 O99 92 4 60 0 89419 0 92414 0 99924 A 25 PAA PEIN een 100 0 88004 0 93029 0 99924 conservation value per features time discounted 0 0 91423 0 91468 0 99924 220 PSOE A PSOE OR 0L B992 200 909 89105918619 10759992 6070 90396170 9173970 99924 HOO SE HO Sala ish 0 lt SIS LOO Op SOV2s Oasis OL AIIE Results for opportunity robustness conservation value per features On Oc Sil DIG MORSOIIGSICNDIOO 11 conservation_values_per_environment csv This file is generated only if the command line option e Of per environment out is used This file is analogous to the file of conservation values per feature 10 conservation_values_per_feature csv see above but it contains results per environme
67. amage given limitations to budgets and to the amounts of areas where different compensatory actions are feasible Type VI Extraction of targets An optimal allocation of resources output from analysis type IV consists of amounts areas assigned to different alternative actions This set of amounts of areas can be seen as a set of area targets which are specific to particular actions in individual environments Area targets obtained this way can then be spatially allocated by further analysis for example using target based site selection methods 5 3 Uncertainty lt Note This section and the next ones are under construction and will be extended over the coming weeks Keep an eye on our website for updates http www helsinki fi bioscience consplan The degree of uncertainty in the responses of features to actions or info gap uncertainty horizon parameter or alpha value see Section 2 5 1 Uncertainty Analysis p 25 determines how much the upper and lower envelops expand from the nominal response In RobOff there is an uncertainty horizon parameter that can be changed in the general settings tab of the setup section of the GUI see Section 4 2 Setup p 89 or alternatively in the general settings input file option info gap alpha see Section 3 3 1 General settings file p 50 This parameter will be used as the reference value of uncertainty and it sets the degree of uncertainty for which most results ar
68. aning of the required parameters and how to specify them in the feature weights function file The examples listed in this file correspond to the panels shown above in Figure 3 3 Example benefit functions p 61 Note that the same sequence of values is used as the list of parameters for panels E and but the resulting functions are quite different from each other More examples of utility functions in RobOff Features weight function type function parameter s comma separated list Panel A power function feature with diminishing return 1 0 1 0 3 Panel B target function at 0 8 or 80 if the reference is 1 0 feature with target 1 0 2 0 8 Panel C convex function defined with a quadratic polynomial polynomial 0 8x 2 2x 0 feature iejolly7 leVr Wp Cer 2 0 Panel D 1 sigmoid 3 parameters as in sigmoid r K PO with r 7 K 1 P0 0 00001 TEMPO ARAE gt cexcloibiciesiavers MAR EPO eIpIr o veleretbliaiatsiavers il weciethes dicuep Inor fs Ws dio OO OOOO Panel E piecewise constant function score inScietias Sel dL 87 Of O25 U S Onis 205 3 355 3 50 Panel F another piecewise constant function score eee xe dis Gu Uo OS 55r 1 Io O 5 Bod 05225 Panel G exponential 0 8 exp 1 2 occurrence feature exp decay 1 0 3 1 2 0 8 Panel H GBF Generalized benefit function as in Zonation Parameters wl 0 25 x 0 3 T 0 9 w2 0 5 y 0 2 feature Clole
69. any methods mentioned above that are most appropriate for spatial reserve selection based on static biodiversity patterns We introduce a framework in which actions produce different uncertain responses for features in different environments through time The focus is specifically on time and uncertainty and biodiversity pattern is therefore explicitly assumed to be dynamic To allow for the complexity of problems that arise in habitat maintenance 19 RobOff Setups management restoration and offsetting we make the simplification that any explicitly spatial aspect of analysis is dropped although we outline workarounds for this simplification The present analyses are intended to answer questions about how much of what kinds of conservation measures should be allocated to which environment types In this role these results provide an important tool for target setting for systematic conservation planning 2 2 RobOff Setups In RobOff a setup is specified to define a particular conservation problem consisting of a number of entities and their interrelations The conceptual model for these is depicted in Figure 2 1 Basic components of a RobOff setup p 20 using standard entity relationship model notation The core concepts in the setup are environment types biodiversity features actions and the responses of features to actions in different environments Figure 2 1 Basic components of a RobOff setup Environment Are
70. apter 4 RobOff Graphical User Interface p 87 Alternatively Chapter 3 The RobOff Software and Command Line Interface p 45 describes how to define a RobOff setup in plain text files Environments actions biodiversity features and responses Figure 5 1 A possible simplified sequence of steps required to define a RobOff setup p 110 depicts a top down sequence of steps to define a RobOff setup for a given planning problem this is just one possible approach First environments need to be defined To this end one has to identify the set of habitat types relevant for planning In principle environments and habitat types are very closely related but an environment is defined in RobOff as an aggregation of sites that are dealt with as a block for planning purposes In practice this means that when an environment is defined in a RobOff setup it is possible to define actions a list of biodiversity features present and responses of features that are specific to that environment If planning is to be performed exactly on a habitat by habitat basis then it is sensible to define an environment for each habitat But more in general additional factors should be considered in order to decide whether some or all of the sites 109 Environments actions biodiversity features and responses belonging to a habitat should be merged or split into different environments Some factors that may suggest merging or splitting habitats i
71. ations further complicate resource allocation The effectiveness of conservation action over time and the impact of development actions are uncertain The aim is to evaluate the sensitivity of conservation action to this uncertainty To this end different approaches can be applied such as various probabilistic models An effective way of addressing uncertainty is to use a non probabilistic info gap model formulation of uncertain responses Info gap is a decision theory for decision making under severe uncertainty As a particular case an envelope bound info gap model contains different upper and lower envelopes which delimit the information gap The uncertain development over time of the representation levels of features is given by upper and lower envelopes around the estimated values These three sequences nominal and upper and lower envelopes need to be specified for each discrete interval Info gap models depend on the horizon of uncertainty parameter which indicates the degree of uncertainty For a 0 the responses will exactly match the expected values For a 1 responses span the whole range between the specified envelopes The envelopes do not need to be symmetric around the nominal values and scale around them as a function of the uncertainty horizon With an info gap formulation it is possible to account for negative consequences of uncertainty robustness analysis as well as positive aspects opportunity 25 Aggr
72. atively variable costs can be defined interactively in the GUI see Section 4 2 Setup p 89 For an example of variable cost in the GUI see Figure 4 8 Editing costs in the RobOff GUI p 94 N Note As an exception the names of cost area curve and time dependent cost files cannot be numbers whether integer or fractional This is because RobOff will first try to parse a number and if it succeeds the cost field will be interpreted as a constant monetary value Only names that cannot be parsed as numbers will be considered as names of cost files For example if in the cost field of an environments file you indicate the value 30 it will be interpreted as constant cost 30 RobOff will not look for a file named 30 even if you provide it A better name if you want to use a variable cost would be cost 30 for example Using numbers as file names is in general a very error prone practice that we strongly discourage 69 Standard RobOff output 3 4 Standard RobOff output RobOff generates a set of files in the output directory o or output command line option Some of them are generated regardless of the options specified whereas some others are optional depending on what options have been enabled in the command line A diagram that summarizes the different types of output files is shown in Figure 3 5 Set of output files and their equivalent GUI dialogs p 70 The figure also includes the equivalent GUI dialogs see
73. atures Note that this is different from the performance ratios which are ratios between conservation values for different robustness criteria 4 sustainability_ratios csv This file contains for each of the 4 possible sets of allocations whenever available sustainability indices as ratios of conservation values see Chapter 2 Framework Methods and Algorithms p 17 The file contains up to 4 matrices of values over time of discounted weak and strong performance ratios across features and environments The first lines of an 74 Optional output files Roboff results sustainability All values are discounted Robustness requirement robust Info gap alpha 1 0 Format discounted strong_perf_envs SHH HEHE SHE Results for No action example file of this type are shown below ratios Horizon weak_perf_feat strong_perf_feat weak_perf_envs O709955377 05 99463 87099553670 994638 107 07 20 0 30 0 40 0 5070 60 0 70 0 992099 079904 0 992098709904 AAA O ase HSE 0 LEE Aor 110 5 i ei on Results for mandatory actions 0 0893746 0 7910067 0 893721 Os TEMG NOOR SO 2Z AT 016921407892221 0789214 20 0 890944 0 787636 0 890919 0 787636 993698 0 992384 0 993697 0 992384 990728 0 988678 0 990727 0 988678 989568 0 987204 0 989568 0 987204 987802 0 984921 0 987801 0 984921 987151 0 984065 0 98715 0 984065 986626 0 983367 0 986625 0 983367 3 4 1 Optional output files Some additio
74. austive environment pairs within 1 resolution used search hour Exhaustive Similar as with exhaustive Optimal for convex problems search search local search Stochastic Tens or hundreds of action Not guaranteed but currently global search environment pairs within tens the only practical and effective genetic of minutes or 1 hour option for large non convex algorithm problems Table 2 5 Speed and ease of use of the optimization methods supported by RobOff general properties Adapted and expanded from Pouzols amp Moilanen 2013 limited Several repetitions are recommended to verify convergence Method Speed of computation Ease of use Random Extremely fast seconds High Greedy Extremely fast seconds High Grid based Fast with small dimensions High exhaustive but becomes exponentially search more time consuming with the number of actions Exhaustive As with exhaustive search but High search an order of magnitude or more local search slower Stochastic Fast but growing with the Medium global search number of environments genetic features and actions algorithm Computation time can be 37 Dealing with connectivity 2 7 Dealing with connectivity The RobOff framework considers alternative conservation actions and their uncertain effects on biodiversity features in different environments through time costs and feasibility of actions a budget time discounting an
75. below for a more detailed explanation Type 1 concave increase with diminishing returns requires the exponent Type 2 target requires the target level Type 3 exponential decay requires two parameters the inverse time constant usually called lambda and the initial value Type 4 generalized benefit function requires four parameters w4 x T wo and y This type of function is in fact a flexible family of functions that was introduced in Zonation 3 see the Zonation manual which can be found online at http www helsinki fi bioscience consplan for details w and w are weights for two parts of the generalized benefit function T is the target level that divides the two parts and x and y are exponents corresponding to the two parts Type 5 sigmoid uses the formulation of the logistic function as defined for modeling population growth in ecology This function requires three parameters r growth rate K carrying capacity and PO initial value Type 6 inverted sigmoid or 1 sigmoid uses the same parameters as the sigmoid function type 6 Type 7 quaaratic polynomial requires the three coefficients of a second order polynomial starting from the higher power 61 Set of files biodiversity features Types 8 and 9 piecewise constant and piecewise linear respectively require an unlimited number of parameters given as pairs of x y values for every interval The following example clarifies the exact me
76. biodiversity features present in an environment Section 3 3 4 64 Time discounting file Set of files biodiversity features p 62 In these files the name that identifies the response is just like business as usual 3 3 6 Time discounting file This is an optional type of file that defines options related to time discounting including discounting model discounting rate and the weights for each time interval The options included in this type of file can all be specified in the general settings file By defining different alternative time discounting files and using them from the general settings file or with different t ime disc command line options it is possible to analyze the influence of time discounting on a base case analysis The following example of a time discounting file defines a discounting rate that is variable in time Different weights of different time intervals are also considered the 5 first intervals say years are discarded whereas the last 5 intervals are equally weighted RobOff time discounting file beta release Time discounting model quasi hyperbolic OR hyperbolic OR exponential Default quasi hyperbolic time discounting model exponential Example of discounting rate variable in time time discounting trate S 2 13 Susp Se25 2 Lekap doa ds295 Ly 05275 Example of different weights over time time discounting weights 0 Ws 0 O OW iW i a ik il
77. by environment are shown By selecting individual entities or groups of them using the shift and control keys it is possible to visualize results specific to subsets of features environments or actions 97 Visualizing results across different dimensions Figure 4 12 Visualizing results through time auod Roboff GUI View Results Tools Help ep v e 4 Setup Results Summary Time Uncertainty Environments Features Actions Compare Actions Budget z Im _ I Feat c M Robust E vi Opportunity NN vi Nominal NN 20S Plotting mode CEU TEN Add O Strong Solid 2 amp Weak Dashed Simple c Clear Conservation value a4 33 22 81 o r T T T T T 1 o 50 100 150 200 250 300 Time Optimize Results conservation value and sustinability uncertainty analysis comparison of actions budget analysis etc Environment specific results are conditioned to individual environments or groups of them Figure 4 13 Visualizing results across environments uou Roboff GUI View Results Tools Help egz7zveo o e 4 Setup Results Summary Time Uncertainty Environments Features Actions Compare Actions Budget Features c M Robust NN lt Opportunity N Yi Nominal NE Plotting mode O Strong Solid 2 Z Weak Dashed C Simple MiddleForestA Conservation value MiddleForestB z 092 8 0 915 UpperForest os 0 905 Sustainability ratio
78. ces among pre defined sets of options such as the type of aggregation across features or across environments whereas some other dimensions are continuous or numerical Such as the degree of uncertainty and are typically associated to sensibility analyzes 2 5 Aggregation of conservation value In RobOff conservation value is aggregated across different dimensions features environments actions and time Fully aggregated results are useful for a global evaluation of a particular RobOff setup However partially aggregated results are needed for a more in depth understanding of the consequences of resource allocation to specific environments or features RobOff generates fully aggregated results as well as partially aggregated results which are calculated across either features environments actions time or a combination of some of these dimensions The next sections define the RobOff computational model for calculating conservation value See Section 4 3 Results p 96 for details on how to use the RobOff graphical interface in order to visualize results across these different dimensions Alternatively see Section 3 4 Standard RobOff output p 70 for details on the output files of RobOff where these results are generated 2 5 1 Uncertainty Analysis Uncertainty about the development of biodiversity features through time when landscapes and environmental conditions change and various actions are applied in different loc
79. changed environments file environments feature weight function type file feature_weight_functiontypes per environment files prefix features_present_ response files prefix response_ Score features files prefix score features present variable cost files prefix cost_units_curve_ budget allocation file budget_allocation data files extension csv Not needed if default discounting options are ok discounting file discounting txt Budget budget 15000000000 mandatory budget preset budget optimization method genetic algorithm optimization criterion weak features robust default OR opportunity OR nominal optimization robustness requirement robust 51 General settings file S Note In all input files file paths are relative to the setup directory location of the main setup file Let us now define each of the options key value pairs supported in the general settings files name info gap alpha info gap alpha range planning horizons Defines the name of a RobOff setup This will be used to identify a setup and for reporting Any text string is accepted Sets the uncertainty horizon alpha for info gap calculations see Section 2 5 1 Uncertainty Analysis p 25 This value will be taken as the reference value It is also possible to specify a range of values see below Range for the uncertainty analysis The range is specified by three numbers s
80. commendations social or political preferences etc 1 5 Software installation and quick start If you are using RobOff for the first time you would typically follow this sequence of steps Installation RobOff can be downloaded from http Awww helsinki fi bioscience consplan and is available for windows systems as a self installing binary or zip package The windows distribution of RobOff includes the following files The RobOff program graphical interface roboffGUI exe The RobOff command line program roboff exe A user manual as pdf and html browsable from the graphical interface Example setups Quick start Quick start First load some of the examples included in the RobOff distribution Familiarize yourself with the program options Some of them are rather intuitive whereas some others will require you to read at least parts of this manual To change a RobOff setup use the graphical interface Chapter 4 RobOff Graphical User Interface p 87 or follow the detailed description of options and input file formats given in Chapter 3 The RobOff Software and Command Line Interface p 45 RobOff setups can be defined via its graphical interface or in a set of input files To run a RobOff analysis you need to define a complete and consistent setup This can be done in the graphical interface or via text files If using text files at least the following input files are needed General settings file see Section 3 3
81. d by preprocessing the responses and that this proportional scale can lead to mathematically indeterminate results if initial values are 0 If absolute is selected occurrence levels will be aggregated as they are specified in the response files without any rescaling or averaging as an exception to the algorithms described in section Section 2 5 Aggregation of conservation value p 25 Responses in absolute scale can be useful if benefit functions are defined for total or global occurrence values across space and environments This however requires that the spatial range of analysis be known in advance when preparing the benefit functions and analyses for different problem subsets or subsets of environments would require readjusting the benefit functions to different scales of total occurrencies Also responses in absolute scale could be useful for example in problems with a single 56 General settings file feature where occurrence levels are transformed into conservation value by a linear benefit function In such cases it is possible to obtain conservation values that directly represent occurrence values Such as abundance In addition the following options define the names of different types of input files These options allow the user to change the names prefixes and extension or suffix of the input files If you decide to use the default values you do not need to specify any these options at all environments
82. d limits on the size of setups In principle there is no limitation on the number of environments features actions etc that a setup can have Processing thousands or tens of thousands of features and environments should be no issue with current commodity PCs Computation time limitations Computational time will in most cases be the limiting factor RobOff computation time will generally increase linearly with the number of features and environments The most constraining factor is the number of alternative actions When optimizing the allocation of actions the number of alternative actions can quickly become the major practical limiting factor Among the optimization strategies supported by RobOff exhaustive grid search methods can be useful for a reduced number of alternative actions typically less than 10 When the number of alternative actions is large you should definitely employ the genetic algorithm optimization strategy The graphical interface see Chapter 4 RobOff 39 References Graphical User Interface p 87 provides feedback that allows you to estimate how long an optimization process will take It is up to the user to decide to switch to faster optimization strategies if needed See Section 2 6 Optimizing resource allocation p 35 for more detatils on the effect of different parameters on the computation time of the supported optimization methods 2 9 References The RobOff framework and software P
83. d robustness requirements The compromise in making this combination of factors analytically operational was to omit the explicit consideration of space If the RobOff model was spatially explicit the already moderately large dimensionality of the data would have been multiplied by the dimensionality of the landscape Assuming a GIS derived grid based landscape description this could result in a data dimension in the order of 10 to 10 times its present size making data demands prohibitive The obvious drawback of the non spatial representation is that the present analysis will not suggest in which locations conservation action should be undertaken The spatial element is important Rather than ignoring it we can suggest three approaches that provide spatial context for the analyses above The first is to account for connectivity by defining different variants of environments for different categories or degrees of connectivity One could have different responses in say isolated forest fragments moderately connected forest fragments and well connected forest areas that are part of a semi continuous forest landscape This is a compromise solution that has the implication that data dimension is multiplied by the count of connectivity categories added As asecond approach it is possible to combine the analysis with expert opinion The present analysis will suggest how much of which kinds of actions should be undertaken in what environments
84. d to constrain the optimization process The following is an example budget_allocation csv 126 Dam Forest River mandatory allocation MiddleForest Inundate 300 MiddleForest Restore 200 preset allocation MiddleForest Restore 200 6 6 A more elaborated biodiversity offsetting example Dam Forest River Aj Note This setup can be found in the RobOff software distribution under the folder directory setup dam forest river offset This setup is an extension of the one presented in the previous section It includes 5 environments and 3 biodiversity features The third feature River is present in the three environments that have been added LowerRiver MiddleRiver and UpperRiver 7 different actions are included The general settings file DamForestRiver ro_setup is similar to the one shown in the previous setup example The environments file now looks like this Note that in some of the new environments only one action related to development is possible Environments file Saved automatically by RobOff Edit at your own risk Environment types file containing a table list of environment types Format env_name weight total_area area_no_action condition list of triplets action name pau cost area available pau per area unit cost SH SH ce SHE MiddleForest 1 500 0 0 1 Inundate 0 300 0 Restore 5 0 200 0 UpperForest 1 30000 06 0 1 Restore 5 0 300900 0 MiddleRiver 1 20
85. e probability 29 Occurrence level 26 29 53 56 57 63 see Aggregation Occurrence value see Occurrence level Offsets adequacy 113 example 121 124 127 multiplier 114 optimal 113 reliability 113 time discounting 33 Offsetting 112 124 127 Open icon library see library Open source 83 Operating system GNU Linux 83 Windows 83 Opportunity 34 Optimal actions 48 Optimization 76 76 103 113 computation time 36 convex 36 criterion 55 genetic algorithm 36 global search 36 greedy search 36 grid search 36 GUI 103 heuristic 36 local search 36 method 36 55 mnimax 33 options 36 55 103 random 36 36 55 robustness requirement 55 stochastic search 36 Optional input files 49 see Input files Optional output files 75 Output file 48 GUI 96 96 Output file 70 budget analysis 77 77 conservation value 73 78 79 costs 81 environments 79 80 features 78 79 log 73 optimal allocation 76 optimization 76 optional 75 performance 74 readme 71 standard 70 summary 71 sustainability ratio 74 79 80 uncertainty analysis 75 Overlap of actions 30 132 P Path see File path pdf see format Per area unit responses 21 Piecewise constant see Benefit function Piecewise linear see Benefit function Plain text 50 58 96 see format Planning incremental 48 Plot see aggregation see combined background 104 budget 102 comparison of actions 101
86. e are no hard coded limits on the number of environments features or actions that can be analyzed The RobOff GUI was developed using the Qt libraries for cross platform desktop http qt project org It also uses extensively the Qwt Qt Widgets for Technical Applications library http qwt sourceforge net Also we used various versions of GCC the GNU compiler collection http gcc gnu org The Roboff GUI also uses icons from the open icon library http openiconlibrary sourceforge net These free open source software projects without which an effective implementation of RobOff would have probably been impossible are greatly acknowledged In the RobOff computational core calculations and aggregation of conservation value are done in a parallel manner in multi processor core shared memory systems by means of OpenMP directives http www openmp org The command line interface and the GUI share a common library that has been designed following a modular approach The core components perform the following tasks construction and validation or RobOff setups calculations of conservation value and sustainability ratios and search of optimal allocations of resources Additional components perform tasks such as loading and saving setups numeric results plots and allocations of resources Precompiled binaries are provided for both 32 and 64 bit systems These are included in the software distribution available as self installing setup binar
87. e ce inaa ener ANALE KVVV E AVAA KA 13 Il Framework Methods and Algorithms 0 cccceeeeeeeeeeeeeeeeeeeeeaeeeeeeeeeas 15 2 Framework Methods and Algorithms 0cceeeeeeeeeeeeeeeeeeeeeees 17 2 1 The RobOff framework sss 17 2 2 RODO Setups weve eee deo ee rated dee aa 20 2 3 Complementarity and scoring esssesesssssesse 22 2 4 The RobOff output space ssssssseee 23 2 5 Aggregation of conservation value seesesesess 25 2 5 1 Uncertainty Analysis sese 25 2 5 2 Aggregating conservation value sss 26 2 5 3 Weak and strong sustainability 32 2 5 4 Time discounting sss 33 2 5 5 Robust and opportunity performance indices 34 2 6 Optimizing resource allocation sssseesseess 35 2 7 Dealing with connectivity eese 38 2 8 Assumptions and limitations eeeeeeeee 39 2 9 References esed AE IA ee ene ee d eto unda 40 IIl The RobOff Software and Command Line Interface 43 3 The RobOff Software and Command Line Interface 45 3 1 Introduction and important general information about files 45 3 2 Running RobOff from the command line 46 3 2 1 Sets of actions ceti a
88. e described in the next sections As a general working principle the list or tree on the left of each of these tabs or visualization dialogs allows the user to select all groups of or individual entities features actions environments etc On the right two plots visualize results These two plots are separated by a splitter element that can be freely shifted up and down In general it is possible to zoom in out using the first second mouse button on all the RobOff plots The plot coordinates at any location can be obtained by clicking on the plot s canvas Also by right clicking on the plots their content can be copied into the clipboard or alternatively saved as an image and or document Common formats such as png svg jpg pdf and postscript are supported The upper plot shows conservation value whereas the lower plot shows sustainability performance indices It is possible to select different robustness requirements nominal robust or opportunity Sustainability indices can be weak and strong In general the plotted values are conditioned to the entities selected on the left Also plots are cumulative Plots are added one at a time and the content of both plot panels can be cleared at any moment to start creating a new plot from scratch For instance one can combine in the same plot the curve of conservation value over time for the whole set of features and a particular feature As another example it is possible to add in the
89. e file Roboff results reame txt RobOff 1 0 0rc4 Warning this is a candidate release http www helsinki fi bioscience consplan Generated on 20130109 at 14 38 27 on achernar This directory contains results for the setup Dam example second level offset and was generated for the following setup configuration DamForestRiver ro_setup General results available see l summary txt 3 conservation values csv and 4 sustainability ratios csv Uncertainty analysis results NOT available Optimization results available see 6 optimization csv Optimal allocation of resources results available see 7 optimal allocation csv Actions comparison results NOT available Budget analysis results NOT available Feature specific results available see the files 10 conservation values per feature csv and 12 sustainability ratios per feature csv Environment specific results available see the files l1 conservation values per environment csv and 13 sustainability ratios per environment csv Last but not least don t forget to check 2 log txt for possible warnings and errors SH oce cb cb db db 1 summary txt Summary of results presented as a table The table gives conservation values and sustainability ratios summarized as scalar values after applying time discounting The following tables and listing are examples of a results summary for a problem with mandatory and optimal sets of allocations This same output can
90. e reported including the summary table plots and time series of conservation value and sustainability ratios over time 113 Time In addition to the info gap uncertainty horizon parameter alpha it is possible to define a range of variation for the info gap uncertainty horizon Similarly this can be done in the general settings tab of the setup section of the GUI or alternatively in the in the general settings input file option info gap alpha range see Section 3 3 1 General settings file p 50 This range is used in the analysis of sensitivity of results to the degree of uncertainty See Section 4 3 Results p 96 for details on how to visualize the results of this analysis in the GUI Alternatively see Section 3 4 1 Optional output files p 75 for details on the output files generated by the uncertainty analysis 5 4 Time Time preference accounts for the change of relative values over time or how to compare values in the future to values in the present In RobOff the relative conservation values over time are defined by three parameters the discounting model the discount factor factor which can be constant or variable over time and weights for every time interval in the planning horizon In the general settings input file see Section 3 3 1 General settings file p 50 the relevant options are time discounting modelandtime discounting rate In the GUI these parameters can be defined in tables that ca
91. eatures Score feature files consist of rows or entries containing four columns Each row defines a score component and how to aggregate it into a score feature The columns included in every row score component are described below The first column indicates what feature is aggregated as a score component These features must have been listed in the biodiversity features present file RobOff will use them as score components not as simple features anymore If a feature listed here is missing from the corresponding biodiversity features present file RobOff will report an error The second column is the name of the score feature RobOff will create a new score feature for every different name found in this second column 66 Set of files score features The third column defines the operator used to aggregate this score component At the moment three operators are supported and which all correspond to basic arithmetic operations The fourth column indicates the component weight By using different component weights it is possible to generate combinations of simple features weighted differently The following listing shows a simple example of score features file RobOff beta release Score features file table list of score components 62 score ty OLS E3 scorel 105 7 il Scorel y Was In this example a single score feature is defined as a combination of three features Score components
92. eceeeeeeeeeeeeeeeeeeeeeeeaaeaeeeeeeeeeeeeanaaeaees 66 3 3 9 Set of files costs of actions sssssssssssseeeen 68 3 4 Standard RobOff output ssssssssssssssseeeenenne mene 70 3 4 1 Optional output files entente tte erede re Eran 75 3 5 What RobOff does not do directly ssssseeee 82 3 6 Implementation details about RobOff ssssssssesssee eee 83 3 7 Data limitations and system requirements ess 83 3 8 Tro bleshootlhig 2 ett rentes ceci eed eerte eheu ended 84 Chapter 3 The RobOff Software and Command Line Interface 3 1 Introduction and important general information about files This chapter describes how to use the RobOff command line There are two alternative approaches to using the RobOff software command line or graphical user interface If you are not familiar with command line tools or editing plain text files you should rather skip this section and use the RobOff graphical user interface see Chapter 4 RobOff Graphical User Interface p 87 The RobOff GUI provides custom dialogs for editing RobOff setups and different visualization options Using RobOff from the command line usually involves creating and editing a set of text files that contain the description of a RobOff setup The next sections describe Whattypes of analyses can be performed using the RobOff command line tool including a concrete list of confi
93. edited in a more interactive manner Figure 3 2 Set of input files and their equivalent GUI dialogs Adapted and expanded from Pouzols amp Moilanen 2013 Input text files for command n GUI interactive editing of setups line execution General settings Required File of general settings File of environments Environments and actions possible File of feature weights amp utility functions Responses of biodiversity features Set of files features present 1 file per environment Set of files responses of features 1 file per different response Biodiversity features present in environments and their responses Allocation of resources Feature weights and benefit functions Optional Costs constant or variable over time or area Time discounting options File of time discounting options File of budget allocation Set of files score features present 1 file per environment Set of files costs over time or as a function of area 1 file per different cost curve Definition of score features The input files are described by simple and self documenting examples first in the following sections From the examples you should be able to write your own input files assuming some familiarity with the RobOff conceptual model see Section 2 1 The RobOff framework p 17 For further details a definition of each of the accepted options in each type of file is provided as well N Note You will
94. ee eee tree HH 121 6 4 2 Output obtained neiseina e iE TEE E 122 0 5 Dam FOreSt eoo tent Ert Rx te E e E CEE N E AR Aa EAK PRI DUE RAN 124 6 6 Dam Forest River Chapter 6 Tutorial and examples 6 1 Aim This tutorial illustrates the practical use of RobOff in different planning contexts Some of the example setups included in this tutorial and the RobOff distribution can serve as a starting point when preparing new setups for modeling your own planning problems The following examples illustrate different types of planning problems Typically different variants of a same base case need to be explored This involves many different aspects or dimensions including ecological uncertainty and uncertainty related to human decisions More specifically dimensions that need to be considered include uncertainty horizon in responses of biodiversity features time preference different discounting rates and different models range of budget availability range of costs different schemes for weighting environments and species and alternative robustness requirements These trade offs are illustrated as far as possible but it is far beyond the scope of this tutorial to explore all the possible variants N Note This tutorial is currently work in progress Several conservation planning projects are currently ongoing and new material is expected soon Do not hesitate to contact us for further information or if you find any issue Keep a
95. egating conservation value analysis An optimal budget allocation can be determined for the robustness or opportunity analyses This can have a determinant relevance in evaluating the reliability of biodiversity offsetting agreements Time is particularly relevant in the context of offsetting where compensation measures are assigned to ecologically damaging economic activities Ecological damage is frequently immediate and certain whereas compensation via restoration will appear with a time delay and is not certain Finally it should be noted that nothing prevents from estimating the info gap uncertainty scaling from statistical analysis The uncertainty in responses to actions can be derived as a combination of statistical information if available and expert opinion 2 5 2 Aggregating conservation value Conservation value in Roboff is calculated through a number of stages that also generate a global performance index while accounting for uncertainties and time preference The basic elements required for the computations are the occurrence values of biodiversity features Discrete time sequences of values for the whole planning horizon must be specified A schema of the stages of computation is shown in Figure 2 4 Flow of aggregation of occurence levels and conservation value in RobOff p 27 These calculations are performed for a proposed set of actions In stage 1 score component features are evaluated following a scoring appr
96. eparated by colons The first number is the lowest value the second value is the step and the third number is the highest value of the uncertainty horizon Default 0 1 by steps of 0 2 Examples info gap alpha range 0 0 2 1 default info gap alpha range 0 0 25 2 Sequence of discrete time instants It can be specified either as a sequence of comma separated numbers or using a range as in the info gap alpha range option The length of the sequence must be equal to the number of columns in the specific response files If the length of the sequence is shorter RobOff might stil run but it will emit a warning message The default value is 0 1 2 columns in response files Note that the first value should be interpreted as units of time ahead of present time where present time can be also interpreted as the time when values are not discounted immediate or non delayed accrual If for example present time is year 2008 and the first planning step is year 2013 the first value of the planning horizon sequence could be defined as either 5 or O Using 5 implies that even the conservation 52 General settings file values accrued in the first year 2013 will be discounted considering a delay of 5 years from now Alternatively to avoid discounting values in 2013 use O as first step which effectively defines 2013 as present time This consideration is extremely important in practice If you use 2013 as first
97. erm originates from the field of ecological economics sustainability and refers to a view of sustainability in which economy is sustainable if the total capital does not decline Total capital includes all forms of capital manufactured human natural and social This implies that is it acceptable to interchange one kind of capital for another i e all forms of capital are substitutable Weak sustainability is also referred to as economic sustainability In RobOff this interpretation of sustainability is extended to biodiversity features meaning that features and their conservation values are interchangeable This means that weak conservation values and performance indices are average values across features Strong An alternative interpretation of sustainability that establishes sustainability that natural capital must be sustained Or in other words itis not acceptable to trade natural capital for other forms such as manufactured captital This is also called ecological sustainability In the RobOff framework this notion is extended to all forms of conservation value i e all biodiversity features must be sustained It follows that strong conservation values and performance indices are minimun values across features It is assumed that two actions cannot overlap geographically or in other words actions in RobOff can be bundles of actions that take place in a same location If in actuality two given actions overlap this s
98. estore Best Allocation Environment 0 MiddleForest area 500 0 0 000 Action 0 do nothing 300 0 Action 1 Inundate 200 0 gt Action 2 Restore Environment 1 UpperForest area 3 000e 04 1 020e 04 Action 0 do nothing 1 980e 04 Action 1 Restore Standard RobOff output 2 log txt This file contains a detailed log of RobOff runs It is strongly advisable to always have a look at this file in order to make sure everything went fine For instance warning messages concerning possible inconsistencies in the input files are recorded here Summaries about files found are also written into this file An extract of an example log is shown in the following listing Note that RobOff logs can be quite long Important messages are usually highlighted with various symbols and lines Roboff log Generated on mrg20 Tue Jun 26 17 36 55 EEST 2012 on setups setup simple demo is a directory Trying to append default filename itolbot Scop EO Reading configuration file No discounting configuration file found Discounting model is Quasi hyperbolic and ratet 00255 Environments 4 4 Setup setups setup simple demo successfully loaded OpenMP support there are 4 threads in parallel regions 3 conservation values csv This file contains conservation values across the time and uncertainty dimensions The file consists of several matrices each of them corresponding to one of the cells in the matrix shown abo
99. fferent time preference options For example how close is the optimal robust solution to the optimal opportunity solution How sensible is the best allocation of resources to the discounting rate 5 2 Analysis types The following is a brief list of different types of analyses that can be performed using the RobOff methods and software While RobOff puts a strong focus on biodiversity offsetting these and additional analyses are applicable to a wide range of situations involving both development and conservation actions such as land use planning or investment in agri environment schemes Type Analysis of uncertainty of conservation value In this analysis one investigates the uncertainty of conservation value through time for a pre specified set of actions One can ask for example how much uncertainty around the nominal value can be tolerated such that the conservation value of the landscape can be expected to not decrease from present This analysis does not necessarily involve optimization e Type Il Analysis of the time perspective Most studies about reserve selection and systematic planning have either explicitly or implicitly assumed static biodiversity patterns However in reality some actions may result in conservation losses or gains immediately while for others losses or gains may occur more slowly over time In this analysis one is concerned about how time discounting influences our perception of how beneficial or detri
100. ffsets Optimal division of resources between conservation actions Optimal biodiversity offsetting Extraction or area targets Data Definition of environments habitat types Actions feasible in each environment including costs and area availability Biodiversity features present in each environment Uncertain responses of features to actions in environments Available budget Features Environment habitat type and features priorities via weighting Methods for dealing with uncertain development of biodiversity features aiming at robust decisions Time discounting taking time preference into consideration Integrates complementarity and scoring approaches Possibility to account for both positive and negative consequences of uncertainty opportunity and robustness analyses RobOff inputs and outputs Automated comparison of alternative actions for varying uncertainty conditions 1 3 RobOff inputs and outputs 1 3 1 Inputs In RobOff emphasis is placed on two aspects time and uncertainty No explicit spatial information e g species distributions habitat suitability maps etc is required as it operates on per area unit characteristics of habitat types or environments RobOff is specifically intended to complement the many methods that are most appropriate for spatial reserve selection based on static biodiversity patterns A RobOff setup a complete definition of a planning problem in the RobOff framework consi
101. file Name of the global environments file Default environments See Section 3 3 2 Environments file p 58 for details on this type of file per environment files Prefix of the names of per environment prefix files RobOff will construct the names of environment specific files containing a list of features present by prepending this prefix to the environments names used in the global environments file see Section 3 3 4 Set of files biodiversity features p 62 Default features present response files prefix Prefix of specific response files RobOff will look for response files by prepending this prefix to the names of responses used in the environments specific files see Section 3 3 5 Set of files responses of biodiversity features p 63 Default response This option can be used for example to enforce the same prefix for all the response files which should make it easier to distinguish response files and or to use responses that are located in a certain directory folder variable cost files Prefix of files defining variable per spatial prefix unit costs RobOff will look for this type of file by prepending this prefix to the variable cost names used in the environments file see Section 3 3 9 Set of files costs of actions p 68 for details on the cost of actions files Default cost units curve Score features files Prefix of files defining score features One prefix of this type of fi
102. g file overrides time discounting options in setup a FILENAME input alloc Use allocation file as input overrides mandatory and preset allocations overrides resource allocation options f per feature out Output feature specific result files potentially big per environment out Output environment specific result files potentially big o INTEGER panels Number of parallel units threads An example of simple RobOff command line could be roboff V setup setup example setup output example output optimize alloc The command line options can be used in any sequence Some other simple examples are roboff setup setup example2 setup output example2 output budget analysis 47 Sets of actions roboff setup setup ex3 setup output ex3 output budget analysis uncertainty optimize alloc The output option specifies a directory name and RobOff generates at least a handful of files inside output directories It is strongly advisable to use different directory names for different commands or RobOff runs to avoid overwriting files As it can be seen certain types of analyses are enabled with specific command line options including the uncertainty analysis the comparison of actions the analysis of sensitivity to budget variations and the optimization of resource allocations After running RobOff it is always advisable to check the 0 readme txt and especially the 2 10g
103. ght function type file feature_weight_functiontypes score features files prefix score_features_ benefit function files prefix custom_benefit_function dE d d bob budget 100000 mandatory budget 0 preset budget 0 budget resolution 5 optimization method exhaustive optimization criterion weak features robust default OR opportunity OR nominal optimization robustness requirement robust The environments file environments csv contains two lines where environments are added to the setup Environments file Saved automatically by RobOff Edit at your own risk Environment types file containing a table list of environment types Format env_name weight total_area area_no_action condition list of triplets action name pau cost area available pau per area unit cost MiddleForest 1 500 0 0 1 inundate 0 300 0 Restore 5 0 200 0 UpperForest 1 30000 0 0 1 Restore 5 0 20000 0 3E E d OE The only feature present in the MiddleForest environment is defined in the presence file features_present_MiddleForest csv In this file it is also 125 Dam Forest defined what are the responses of the present features to the actions that can be undertaken in this environment RobOff file of biodiversity features present in an environment MT feature_name present_estimate_not_used response_no_action list of quartets action response init value env value
104. guration options and command line options The types of input files that are either required or optional The output files generated by RobOff for different types of analysis First we highlight a few important practical issues File names In principle RobOff imposes no limits on the length and characters used in file names and file paths Both the slash and backslash characters can be used as directory separators Be aware however that special characters such as spaces quotation marks brackets and punctuation marks in general are a very common source of problems You will likely experience issues with some operating systems and or third software We recommend to avoid such special characters when possible and to use them with care For example using the underscore character instead of spaces can save a lot of trouble File paths In RobOff setups all the file paths are interpreted as relative to the location of the main settings file This rule applies to every single file referred to in any file that is part of a RobOff setup Always keep in mind that paths are required if you use files located in different directories For instance if the response files are located in a responses subdirectory or folder then the responses must be named by their relative path such as responses response business as usual csv responses response restorationl csv etc Similarly it is possible to use 45 Running RobOff from t
105. he command line full paths such as C roboff responses response restorationl csv or home foo roboff responses response restorationl csv Some comprehensive examples can be found in the RobOff distribution and some of them are described in Chapter 6 Tutorial and examples p 119 3 2 Running RobOff from the command line The general syntax of a RobOff command line is just a sequence of options Whether you use RobOff from the command line or the GUI three main stages or functional blocks can be distinguished editing a setup visualization of results and optimization of resource allocation as summarized in Figure 3 1 Running RobOff stages from inputs to outputs p 46 Figure 3 1 Running RobOff stages from inputs to outputs Input as plain Input via text files GUI lt Error checking consistency Validated setup Output as plain text files Calculations conservation value sustainability GUI output tables and plots Optimization Optimal allocation 46 Running RobOff from the command line When you use the RobOff command line you will typically edit input files first To do so you can use your favorite text editor or spreadsheet software but always remember to save files as plain text The RobOff command line will first check the consistency of the input files you provide If no errors are found RobOff builds a validated setup and performs certain calcula
106. hould be modeled as three non overlapping actions one action the other action or both simultaneously For each environment at least one action must be specified that corresponds to the case where neither development nor conservation actions are undertaken i e no human intervention takes place Different actions have diverse effects on biodiversity features Typically the effectiveness of conservation actions over time and the impact of development actions are uncertain The aim is to evaluate the sensitivity of conservation actions to this uncertainty around the estimated outcome To this end uncertainty conditions are addressed by an info gap model formulation of uncertain responses We use an envelope bound info gap model with different upper and lower envelopes which delimit the information gap model The uncertain development over time of the representation levels of features is specified by lower and upper envelopes around the estimated values These 30 Aggregating conservation value three sequences need to be specified for each discrete interval in the planning horizon Info gap models depend on the horizon of uncertainty parameter a which indicates the degree of uncertainty For 0 the responses will exactly match the expected values For 1 responses span the whole range between the specified envelopes Generally as increases responses will proportionally deviate from the expected values within an expanding inte
107. ia Allocations mandatory etc Substitutability viewpoint 111 Time preference and robustness Time preference and robustness against uncertainty Two additional and very important aspects of any planning problem that need to be specified are time preference and requirement of robustness against uncertainty These two aspects are particularly subject to opinion and interpretation Once all the entities and parameters described above have been defined it is possible and common even advisable to evaluate results for different time preferences and robustness requirements Similarly for types of analysis that involve some form of optimization it is possible to calculate optimal allocations of resources for different variants of time preference and robustness requirements Time preference is defined in RobOff by setting the following inputs time discounting model and rate and time weights See Section 2 5 4 Time discounting p 33 for methodological details Related to uncertainty analysis three levels of robustness or robustnes requirements for optimization are considered in RobOff nominal robust and opportunity Results are provided for the three alternatives both in the console and graphical versions of RobOff However each optimized solution is calculated for a particular robustness requirement Often one will be interested in optimal allocations for different robustness requirements and possibly di
108. iable_cost2 15000 Below is an example of a variable cost area cost file These files describe a cost area curve and consist of two columns The first column corresponds to the area value whereas the second column corresponds to the cost in monetary units x and y coordinates respectively in the example Figure 3 4 Example of per area unit cost as a function of area extent p 69 If this file were specifying the variable cost of action active restoration 1 variable cost1 it should be named variable cost1 csv RobOff cost area file Format area value cost value PIS 00 000 5 45000 10 30000 Dye 0100 207 18000 5760 0 30 3510910 ES PEESIE 000 68 Set of files costs of actions Figure 3 4 Example of per area unit cost as a function of area extent Cost 1100 0 1000 0 Decreasing land availability 900 0 800 0 700 0 600 0 eni 20 40 60 80 100 Area In contrast time dependent cost files consist of one single column with one row per time interval in the planning horizon time intervals are implicit as in response files and the number of rows must be consistent with the length of the planning horizon If time dependent costs are defined economic discount will be applied by using the model and discount rates specified in the general settings file options economic discount rate and economic discount rate see Section 3 3 1 General settings file p 50 Altern
109. ies or compressed packages As of this writing precompiled binaries are provided for GNU Linux and Windows operating systems The software example setups and this manual are available from the consplan website of the Biodiversity Conservation Informatics Group at University of Helsinki http consplan it helsinki fi software projects roboff General information can also be obtained from the group website http www helsinki fi bioscience consplan 3 7 Data limitations and system requirements RobOff fully supports multi core 64 bits systems and imposes no hard limits on the size of setups Processing hundreds or thousands of features and environments is no issue with current commodity PCs Computation time increases linearly with the number of features and environments but RobOff supports optimization strategies that can cope with large sets of actions For such high dimensional problems more than 10 20 alternative actions the genetic algorithm method of optimization should be used for more details on optimization methods see Section 2 6 Optimizing resource allocation p 35 83 Troubleshooting 3 8 Troubleshooting This section is simply a concise list of common issues and pitfalls to take into account Always check the log window or readme and log files Most simple and common errors are easy to catch by carefully reading notices and warnings issued by RobOff Path of files and directories If RobOff complains tha
110. ifferent names Then you can use one or several of them as score components 67 Set of files costs of actions 3 3 9 Set of files costs of actions Costs can be constant values per area unit cost or variable There are two types of variable costs cost area curves and time dependent costs Constant costs are directly specified in the environments file Section 3 3 2 Environments file p 58 However variable costs require additional files If the cost of an action is variable itis specified as a name in the environments file as opposed to a numeric value in monetary units and the corresponding file must be provided one file per different cost In the case of cost area curves costs are defined as functions of the amount of area where an action is performed In cost area files two columns of values must be provided with one x y pair or area cost pairs in each row Cost values are interpolated linearly The same area and cost units as in other setup files must be used The following is an example of an environments file with some variable costs as functions of area allocated Environment types file containing a table list of environment types env name weight total area are no action condition list of triplets action area unit cost area available envi 1 0 25000 500 1 0 devell 0 150000 restoration variable coat r 15000 env2 1 0 25000 2000 1 0 devel12 2 0 150000 restoration 2 var
111. ing To 2 00 Rate 2 50 x 42909 f get allocation Step 0 20 v Medals CETME Results Optimize Feature response file 12 home fedemp example setup dam river response r9 CC env UpperRiver act Obstruct csv Ok Setup validated Found 5 environments 3 features 5 presences 12 responses and 3 actions subject to optimization Budget allocated O Set of actions of this setup succesfully validated Everthing seems correct AL Satun Thamalfadamnlawamnlancatun damarivariDamEaractDivar ro entun euceaeefullisclaadad b The log window is normally shown at the bottom of the graphical interface but it can be dragged and detached This window provides feedback about the operation of RobOff including warnings and error messages to the user The visibility of this window can be toggled on off from the main menu 88 Setup 4 2 Setup In this section RobOff setups can be defined in different tabs Each of the tabs correspond to one of the main entities in the RobOff framework environments features specific responses of features to actions actions and their costs time discounting scores etc General settings can be edited in the leftmost tab This includes general parameters such as the name or description of the setup and values that define the context of the RobOff analysis such as the uncertainty horizon planning temporal horizon time discount rate and budget
112. ing uncertain responses in the GUI p 91 Each of these responses is contained in one of the files of responses of biodiversity features see Section 3 3 5 Set of files responses of biodiversity features p 63 Figure 4 3 Editing environments and actions in the GUI Gg Roboff GUI View Results Tools Help em j vo e 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring Environments Actions Name Weight Total area Condition Active area Name Per unit cost Area available 1 MiddleForest d 500 1 500 1 do nothing o 300 2 UpperForest 1 30000 1 30000 2 Obstruct o 200 3 MiddleRiver 1 20 1 20 3 ObstructPlusTT 2 200 DY UpperRiver 1 300 1 300 5 LowerRiver 1 10 1 10 Results Optimize E a D 3 Optimizing budget allocation to 3 actions in 5 environments Genetic algorithm search finished Trials 150500 Budget spent in best solution 99920 Performance 0 866432 Optimization finished at 12 07 39 on Thu Nov 8 2012 took 8 479 seconds 90 Setup Figure 4 4 Editing biodiversity features in the GUI IT Reno CO Vic esate coe tel cdm Dem wo o 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring Environment Features MiddleForest Name 2resentestimate Response no action business as usual UpperFores
113. ion This procedure is applicable 114 Interactions to cases where the amount of resources that are available for conservation can be expressed as proportional to the amount or impact of development actions e g extent of urban development or amount of habitat loss 1 Set cost of action devel to a negative value that represents the availability of resources to compensate a development action This is done in the Costs tab of the setup section 2 Set the total available budget to 0 Inthe General settings tab of the setup section 3 Set as mandatory allocation a certain amount of area to the devel action This amount may have been decided in advance in a development plan or it may be under consideration In the latter case you might want to explore a range of amounts 4 Go to the optimize section of the GUI and obtain an optimal allocation of resources The budget available from the negative cost of the devel action will be best allocated to maximize the sustainability ratio that you select 5 By looking at the sustainability ratios in the summary table Summary tab of the results section you can easily conclude if no net loss is achieved Normally this will depend on whether the robust sustainability indices are equal or greater than 1 or a reference value obtained for satisfactory conditions and actions 6 To find the exact minimum amount of resources that would have been needed to achieve no net loss you need to ge
114. ion of targets for systematic conservation planning These analyses involve many or all of the optional components of RobOff analyses described in next sections The way conservation value is calculated is described in detail in Section 2 2 RobOff Setups p 20 We first describe here the motivation conceptual model Section 2 2 RobOff Setups p 20 and output space of the RobOff framework Section 2 4 The RobOff output space p 23 Conservation science systematic conservation planning spatial conservation prioritization and conservation resource allocation focus on cost effective resource allocation A substantial amount of literature under the keywords reserve selection and site selection concerns the design and expansion of reserve networks Most of these studies focus on a binary select or not decision optimization problem that is applied to implicitly static data describing the distributions biodiversity features and costs Software packages specifically developed to address such problems include Marxan Zonation ConsNet and C Plan Clearly the select or not decision problem is a comparatively simple variant of a more complicated decision problem the multi action allocation problem where the focus is on the allocation of alternative conservation actions We are interested in habitat maintenance where the objective is to maintain present biodiversity values conservation management where there are specific
115. ion values across time File of feature sustainability ratios across time File of environment sustainability ratios across time 70 Standard RobOff output A directory listing of a typical RobOff output directory looks like this drwxr xr x 2 fedemp fedemp 4096 Dec 12 13 12 drwxr xr x 3 fedemp fedemp 4096 Dec 12 13 12 rw r r 1 fedemp fedemp 156 Dee W257 13 12 0 readmestxts rw r r 1 fedemp fedemp 6805 Dec TATS kesumat y e Sv rw r r 1 fedemp fedemp Soc NINE ENIKO Ce ERE rw r r 1 kedemp fedemp 80546 Dec 12 13 12 3 conservation_values csv rw r r 1 fedemp fedemp 743 Dec 12 13 12 4 conservation_performance_ratios csv rw r r 1 fedemp fedemp 222 Dec 12 13 12 5 uncertainty_analysis csv rw r r 1 fedemp fedemp TR pec T2 Sisal 2 G Optimization CSV rw r r 1 fedemp fedemp LOG DEG ITA S d 27D imassadslo cat one es rw r r 1 fedemp fedemp GSUboOCENI QI OTIO aCtion COmparr son esy All the output files are in plain text format The outputs of RobOff are split into a number of simple files that are easy to process in scripts or load into spreadsheet software The following output files are always generated 0 readme txt General useful information about how the software was run such as the time and machine where the analysis was run RobOff version employed computation time etc It also lists what results and corresponding files are available This is an example readm
116. iversity feature 21 benefit function 60 editing 90 GUI 90 occurrence level 62 presence file 62 90 response 64 90 response file 64 90 129 score 23 66 115 simple 23 23 115 uncertain level 62 utility function 60 weight 60 see Weight Budget 3 22 65 92 allocation 65 92 analysis 48 77 77 file 65 92 GUI 102 output 77 77 range 102 resolution 54 sensitivity 102 C C Plan 19 Clipboard 96 104 Comma 10 Comma separated values 58 58 96 Command line 46 47 65 70 75 77 84 120 first contact 120 Comment 49 50 Comment line see Comment Comparison actions see Action Complementarity 22 23 Computation time 39 83 see Optimization Concave see Benefit function Connectivity 21 38 Conservation planning 40 Conservation resource allocation 19 25 28 41 48 76 Conservation unit 21 Conservation value 26 29 73 78 see Aggregation aggregation 26 per environment 79 per feature 78 uncertainty 112 ConsNet 19 38 Consplan 9 Constant see Benefit function see cost Convex see Benefit function Convex optimization 36 Core see multi core Cost 17 20 21 21 35 58 constant 21 68 92 cost area function 21 68 92 editing 92 effectiveness 35 efficiency 17 22 23 file 92 GUI 92 time dependent 21 68 81 89 92 variable 68 92 csv see Comma separated values D Dam Forest 124 Dam Forest River 127 Data availability 110 Decima
117. k for allocation of conservation resources to multiple actions and analysis of alternative land use options It has a broad applicability to protection management maintenance restoration and offsetting In the RobOff framework different conservation and or development actions have uncertain responses over time for different biodiversity features in different environments RobOff can evaluate a particular solution and find an optimal allocation of resources for a given setup In essence this software is a decision support tool with an emphasis on uncertainty and time preference that can solve planning problems with a potentially large set of alternative actions It aims at solving high dimensional problems of the order of thousands or tens of thousands of alternative actions and or features and environments RobOff is currently under active development Keep an eye on our website for updates http Awww helsinki fi bioscience consplan Acknowledgements Special thanks to John Leathwick from DOC New Zealand for providing the Dam Forest River offsetting example setups and for using testing and commenting on early versions of RobOff Great thanks are also due to Mariana Fuentes from James Cook University and Heini Kujala from University of Melbourne for their feedback and suggestions for improvements to both the software and manual Thanks as well to Adriano Mazziotta Mikko M nkk nen and Janne Kotiaho from University of Jyvask
118. l separator 10 Design protected area network 41 Directory see File path Discounting 40 see Time see also Economic discount Diversity genetic 29 Dynamic interactions 39 82 E Economic discount 69 89 exponential 55 hyperbolic 55 quasi hyperbolic 55 Ecosystem services flow 29 Stock 29 Encapsulated PostScript see format Envelope 30 64 Environment 21 110 action 90 biodiversity features presence file 62 conservation value 79 editing 90 file 58 90 GUI 90 per environment file 62 Environment weight see Weight 130 eps see format Examples see Dam Forest Forest River see tutorial Exercises see tutorial Exhaustive search 36 36 39 55 103 Expert opinion 3 38 Exponential see Benefit function see Time discounting F Feature see Biodiversity feature conservation value 78 Feature response see Response Feature weight see Weight File name 45 File path 46 52 full 46 relative 46 First contact Command line 120 GUI 120 Flow see Ecosystem services Folder see File path Format see csv eps 96 jpg 96 pdf 96 plain text 71 png 96 postscript 96 svg 96 Free software 83 G GCC 83 Generalized see Benefit function Genes 21 Genetic algorithm 36 36 55 103 GIS 5 38 GNU Linux see Operating system Graphical user interface 87 Greedy search 36 36 55 103 Grid search 36 GUI action 90 allocation 92 see Dam
119. lative change in their representation as a consequence of a given action Representation values are transformed into conservation values using a benefit function with feature specific weights and functions By applying specific functions different mappings from representation of biodiversity features to conservation value can be modeled In parallel representations of score features can be transformed to score values by analogous transformations and by an aggregation procedure which produces score features These are further aggregated with simple biodiversity features in the next stages effectively integrating scoring and complementarity approaches in a unified approach to conservation resource allocation Table 2 2 Mathematical symbols Symbol Description j Subscript for biodiversity features i Sub superscript for environments k Subscript for actions l 5 Set of features present across all the environments s E y Set of environments that make up the landscape Uil gf Benefit transformation for feature wj Weight for feature d Condition of the jth feature in environment i JE Set of actions defined for environment E l Ia si Area amount of area units where action d is performed Of j Occurrence level of feature fj in environment when action d is undertaken If feature f is not present 0 70V r Representation value of feature fj in environment E o F Info gap ex
120. ld be kept in mind including geographical and taxonomic biases On the bright side RobOff provides a complete set of tools to account for various forms Getting a RobOff analysis running of uncertainty More details about possible steps to define RobOff setups are given in Section 5 1 General remarks p 109 More details about the RobOff conceptual model can be found in Section 2 2 RobOff Setups p 20 1 4 2 Getting a RobOff analysis running The next points are a step by step guide to develop a basic RobOff setup and is intended to get you started Note however that several example setups are included with RobOff You might want to start by looking at those before developing your own setups Further information on common planning setups is included in Chapter 5 RobOff analysis setups for common planning needs p 109 1 Get a basic analysis running a Install RobOff and make sure it works correctly by trying one of the examples provided b Decide your degree of uncertainty see Section 2 5 1 Uncertainty Analysis p 25 and concept of time preference see Section 2 5 4 Time discounting p 33 c Change these parameters in the graphical interface or in the general settings file see Section 3 3 1 General settings file p 50 Check that you are able to generate results for instance conservation value through time after these changes d Try variants of this basic setup by changing the
121. le can be specified for every environment RobOff will look for this type of file by prepending this prefix to the 57 Environments file environment names used in the environments file see Section 3 3 8 Set of files score features p 66 for details on the score features files Default response_ data files extension String to be used as file extension or suffix for the RobOff input and output files This suffix will be appended to most input files Default value csv Note that this default value assumed throughout this manual is intended for better interaction with common spreadsheet software packages which typically load and save comma separated values files using the csv extension If it is changed to txt then for example the environments file name will be environments txt rather than the default environments csv This extension would make it easier to associate RobOff input files with plain text editors Note that regardless of the extension used RobOff always loads and saves files as plain text files 3 3 2 Environments file The environments considered in a RobOff setup are specified in a single file that contains one line per environment The following listing is an example of an environments file Environment types file containing a table list of environment types Habitat weight totalArea areaNoAction condition ListOfTriplets Actions perAreaUnitCost areaAvail environment 1 1 0 25
122. limits These last parameters have a strong influence on results and you will probably need to explore their impact see more on this below in Section 4 3 Results p 96 The economic discount model and rate are relevant only if time dependent costs are defined In this tab it is also possible to modify file names even though this is not normally needed Figure 4 2 Editing a setup general settings p 89 Figure 4 2 Editing a setup general settings RobOff GUI 1 0 0rc4 censored_name_second_level_offset home fedemp example setup dam enu z amp Setup Global parameters Name Robust offsetting Rescale responses Time discounting Vi Discount rate lo Uncertainty horizon Value Range From To Step Results Optimize second_level_offset Environments 3 Features le 03 Roboff GUI View Results Tools Help wo A M General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring Time planning horizon O Regular intervals Custom intervals lo 5 15 25 50 70 lt gt Ir Budget limits ao Total budget 200000 00 Preset budget 0 00 2 D 00 Economic discounting 2 00 E Rate 2 50 bc E Model Quasi hyperbolic O File names and prefixes Environments file nam nvironment Environment files feature presence pref features present
123. lity S2ee eee a ay Or Po UDPSS scole ec Info gap uncertainty horizon alpha 0 5 range from 0 to 2 Conserv value discounted weak across features Minimum Nominal Maximum No action 0 8415 0 8579 0 8737 Conservation performance ratios discounted Robust Nominal Opportunity Weak for environments 0 XS sil 1 000 1038 Strong for environments 059 GO 1 000 IOS Weak for features 029 63 1 1 000 1 038 Strong for features 0 9631 1 000 1 038 L22222222222222 2 End of output Roboff finished ok To obtain the output above this command was used roboff setup toy offsetting ro setup output example output 123 Dam Forest Figure 6 2 Summary of results for the minimal setup in the RobOff GUI RobOff GUI 1 0 0rc4 RobOff minimal toy offsetting home fedemp roboff example setups setup minimal toy offsetting Roboff GUI View Results Tools Help DA Bi fv eo o 4 Setup Results Summary Time Uncertainty Environments Features Env Features Actions Compare Actions Budget Conservation value Robust Nominal Opportunity No action 0 84147 0 85792 0 87366 Mandatory actions 0 84147 0 85792 0 87366 Mandatory preset actions 0 84147 0 85792 0 87366 Optimal actions 0 84147 0 85792 0 87366 Sustainability ratios for No action Robust Nominal Opportunity Weak environments 0 96315 1 1 0383 Strong environments 0 96315 1 1 0383 Weak features 0 96315
124. lopment of response functions improves understanding about the effects of restoration or management actions Failure to complete a RobOff setup points out deficiencies in data models and understanding that is necessary for sound multi action allocation If a non spatial analysis turns out difficult to parameterize then doing a spatial multi action analysis will be next to impossible 17 The RobOff framework Table 2 1 Typical or suggested usages of RobOff Adapted and extended from Pouzols amp Moilanen 2013 Type of analysis Characteristics and tradeoffs analyzed Education Teaching about resource allocation in multi action problems How to balance tradeoffs between actions time preferences and uncertain responses of biodiversity features Insight and problem clarification Development of response functions Development of response functions of features for different actions improves our understanding of restoration and management alternatives Identification of data and or model deficiencies Failure to develop temporal non spatial analysis of conservation actions will reveal deficiencies in our ability to plan restoration or offsetting in a quantitatively well informed manner Management oriented analysis simple level Influence of Influence of the planning horizon study period or time time preference discounting model and rate on decisions Short term on decisions
125. ment protection maintenance restoration offset etc The RobOff software distinguishes three levels or classes of actions see Section 3 2 1 Sets of actions p 48 mandatory preset and optimal Mandatory and preset actions are defined a priori by the user whereas sets of optimal actions are generated automatically for some types of analyses The onset of an action is implicitly defined by the associated responses Alternative starting points of asame action should be represented as alternative actions with responses that can potentially differ in more than a delay See also Section 3 5 What RobOff does not do directly p 82 Costs of actions can be specified in three forms constant values cost area functions or time dependent functions Implemented amounts of actions The amounts and target environments of different actions are assumed to be either mandatorily implemented due to a 21 Complementarity and scoring priori decisions e g environmentally damaging actions or are optimized by the software e g compensation actions The budget constrains the total cost of the set of actions that can be implemented The uncertainty model and robustness requirements allow for accounting of uncertain responses of features to actions One can evaluate risk opportunity tradeoffs for some uncertain parameters Figure 2 2 RobOff conceptual diagram Adapted from Pouzols Burgman amp Moilanen 2012
126. mental some actions are This analysis does not require optimization if assuming a pre specified set of actions 112 Uncertainty Type lll Reliability of biodiversity offsets In this analysis one is interested in how reliably a set of compensatory actions offsets can compensate for a set of ecologically damaging actions Are the proposed offsets likely to be adequate Here the need for analysis follows from the fact that ecologically damaging actions frequently cause immediate and certain loss of ecological value whereas offsetting if it is implemented via restoration only produces uncertain gains with a time delay There are two variants of this analysis that correspond to the concepts of weak and strong sustainability Again this analysis does not require any optimization if assuming a pre specified set of actions Type IV Optimal division of resources between conservation actions Conservation actions are available given constraints to the budget Given the present state of biodiversity features actions and their uncertain effects through time which set of actions would deliver the highest conservation value accounting for uncertainty time costs and all other factors summarized in the sections describing the data and computational models Type V Optimal biodiversity offsetting set of environmentally damaging actions has been proposed With which actions in what environments does one best compensate for the d
127. ments file and that all the corresponding specific response files can be found 53 General settings file budget mandatory budget preset budget budget resolution budget allocation file optimization method Mandatory parameter that specifies the maximum budget available for conservation Expressed in arbitrary monetary units as long as the same units are consistently used in every file Maximum budget available for mandatory actions This is an optional parameter Default global budget Maximum budget available for preset actions This is an optional parameter Default global budget Resolution used for allocating the available budget in percentage When performing resource allocation optimization RobOff will always use multiples of this resolution value to allocate fractions of the available budget Default 5 Name of an optional file specifying the amounts allocated to different actions see Section 3 3 7 Budget allocation file p 65 for details Default value budget_allocation Optimization method The options currently supported are genetic algorithm exhaustive exhaustive local search greedy local search random The default method is genetic algorithm The genetic algorithm method is strongly recommended for highly dimensional problems exhaustive Will provide exact solutions but can be very time consuming and in practice will be feasible only for problems of
128. mization analysis is enabled in the command line The mandatory preset and best allocations are reported whenever available This type of file can be loaded in the optimization section of the GUI see Section 4 4 Optimization p 103 The following listing shows an example of an optimal allocation output file Optional output files Roboff optimization results optimal allocation RobOff saved this allocation for the setup Dam example second level offset Edit at your own risk Format environment action amount area equivalent money mandatory allocation MiddleForest Inundate 300 0 MiddleRiver Inundate 20 0 UpperRiver Obstruct 200 0 LowerRiver AlterFlow 10 0 preset allocation optimal allocation MiddleForest Restore 200 1000 UpperForest Restore 19800 99000 8 actions_comparison csv This file is generated if the comparison of actions analysis is enabled c or compare actions command line options This file shows for the range of values of the uncertainty horizon the preference between O and 1 between pairs of actions The first column corresponds to the different values of the uncertainty horizon The values provided in the other columns are the proportion of resources that should be allocated to the first action in every pair in order to maximize conservation value For every environment as many columns as possible pairs of actions are generated In the example sho
129. mple setup files will give you enough hints on what can be changed and how to change it The next listings are intended as a complete reference that can be consulted at any time when needed 50 General settings file RobOff setup file beta release name simple example Whether to perform robust estimations robust offsetting 1 formerly k a read uncertainty info allow overwriting 0 Uncertainty horizon for which robust results are calculated allocation of resources is optimized info gap alpha 1 Range of uncertainty horizons Three possible formats start end OR start step end OR alphal alpha2 start end is equivalent to start 0 lend step 0 1 info gap alpha range 0 0 25 2 Range of planning horizons Three possible formats start end OR start step end OR datel date2 date3 start end is equivalent to start 1 end step 1 planning horizons 0 30 Time discounting options These can be defined in an optional time discounting file Time discounting model quasi hyperbolic default OR hyperbolic OR exponential time discounting model quasi hyperbolic in Time discounting rate in time discounting rate 2 5 Number of environments that should be defined in the environments file environment types 5 Total number of respones that should be defined in response files feature responses 1 File names and prefixes All these are default values and do not need to be specified unless
130. n be found in the Discounting tab of the setup section see Section 4 2 Setup p 89 The influence of these parameters on the results of a RobOff analysis will depend on how immediate or delayed the effects of actions on biodiversity features are The effects of time discounting can be observed in most of the results generated by RobOff from the summary table to all the plots and time series of conservation value and sustainability ratios over time 5 5 Offsetting This section is intended to provide general guidance on how to analyze biodiversity offsets within the RobOff framework 5 5 1 How much compensation is enough In the context of biodiversity offsets one often wishes to find an answer to the question of how much compensation restoration maintenance etc is enough to compensate loss of biodiversity due to development A robust compensation strategy often requires undertaking conservation actions in areas much larger than the ones impacted by development activities In offsetting systems multipliers are commonly considered in order to account for disparities between negative consequences of development and benefits from compensation as well as risks inherent to compensation actions e g uncertain benefits of restoration For example in the GUI see Chapter 4 RobOff Graphical User Interface p 87 you can follow this sequence of steps assuming that there is a development action devel and at least a compensation act
131. n eye on our website for updates http www helsinki fi bioscience consplan 119 A first contact with the RobOff GUI 6 2 A first contact with the RobOff GUI The aim of this example is to learn how to visualize a setup and the results obtained and how to find optimzal allocations of resources If you prefer not to use the graphical interface of RobOff just skip this section See Section 6 3 A first contact with the RobOff command line interface p 120 for a first contact with the RobOff command line To start using the RobOff GUI follow these steps Find the first contact gui directory in the example setups included with the RobOff distribution and open the setup that can be found inside file first contact gui ro_setup This setup is already complete and consistent You do not need to know how to define it or change its core components Once you have loaded the setup you can inspect the environments features actions responses and other entities included in the different tabs of the setup section The General settings tab includes important parameters such as the uncertainty horizon planning horizon and available budget See Section 4 2 Setup p 89 for details on what are the components of the setup section of the GUI You can visualize a summary of results and plots of conservation value and sustainability ratios over time in the results section different types of analyses like sensitivity
132. n of the degree of uncertainty or info gap alpha value on the horizontal axis Gaps in the actions comparison plot denote that no resources are allocated to none of the actions Note that before comparing any pair of actions a specific analysis needs to be performed This requires finding the best allocation of resources for different degrees of uncertainties uncertainty horizon on the x axis This analysis process started by pushing the analyze button can be time consuming depending on the complexity of the setup The computation time increases linearly with the number of steps of the uncertainty range 101 Sensitivity of results to Budget variations Figure 4 1 8 Comparing actions Roboff GUI Edit View Tools Help m E wW T Setup Results Summary Time Uncertainty Environments Features Actions Compare Actions Budget Settings Use Sum of allocations Analyze Robustness Refresh Action Cost 1007 Action Cost 5C enwiro A enviro do not develo o amp 807 develo benig 300 J benig active 1000 Be 5 enviro enviro 604 do not do not 08 IDs v develo develo 0 8 benig benig 300 8 4 active active goo 8 9 amp Ep foo 1 D foo 1 H 4 do not do not o 8 develo develo o 204 penis benig 300 a ace active 800 a bar barl 0 I T T T 1 o 05 1 15 2 D Uncertainty horizon GI D Optimize E3 Looki
133. n sections of the graphical interface setup results and optimize Alternatively you will be editing input files and running RobOff analyses from the command line 10 Quick start The three main sections of the RobOff GUI correspond to three stages between which you can switch at any moment see Figure 1 1 Example screen capture of the RobOff GUI summary of results p 11 and Figure 1 2 Example screen capture of the RobOff GUI editing responses in the setup section p 12 Setup section change settings and data model as needed e Result section visualize results for predefined or optimized allocations of resources Optimize section define optimization criteria and generate optimal allocations of resources Figure 1 1 Example screen capture of the RobOff GUI summary of results RobOff spatially implicit conservation planning Roboff GUI Edit View Tools Help eg Twv o e T Setup f iol Results Summary Time Uncertainty Environments Features Actions Compare Actions Budget Settings Conservation value Robust Nominal Opportunity No action 0 7417 0 7417 0 7417 Mandatory actions 0 7029 0 7029 0 7029 Optimal actions 0 6791 0 6791 0 6791 Performance ratios for Mandatory actions Robust Nominal Opportunity Weak features 0 697 0 954 1 51 Strong features 0 510 0 812 1 29 Weak environments 0 580 0 640 0 902 Strong environments 0 438 0 508 0 859
134. nal output files are optionally generated for the following analyses uncertainty analysis comparison of actions optimization of resource allocation and budget sensitivity analysis The standard file names and their contents are as follows 5 uncertainty_analysis csv This file summarizes the results of the uncertainty analysis u or uncertainty command line options The file contains a table of values as a function of the uncertainty horizon for the performance indices ratios of weak and strong sustainability across features and across environments Average and minimum conservation values across features and environments are included as well These results are reported for three robustness requirements nominal robust and opportunity This is the same information that can be visualized in the uncertainty analysis plot of the graphical interface An example output uncertainty analysis file looks like this 75 Optional output files Roboff uncertainty analysis results Uncertainty analysis for optimal allocation Format alpha sust_weak_features sust_strong_features sust_weak_environments sust_strong_environments consval_weak_features consval_strong_features consval_weak_environments consval_strong_environments Results for nominal robustness 0 7 On DOG On ISA 7 On TIAS 0 1ISA7 7 OU MISS On ZU OGIO 0 5 7 7 59 O625 ORO OQ RUIN 0s TALS 05 TIS27 E a AA gt WALA 0 GSES Os
135. nalyses defined in the RobOff framework and their implementation in publicly available software make solutions to a significant set of conservation resource allocation problems accessible for the first time to conservation scientists and managers The RobOff framework in a nutshell For a quick start see Section 1 5 Software installation and quick start p 9 The RobOff framework is described in depth in Chapter 2 Framework Methods and Algorithms p 17 The software and its graphical interface are described in Chapter 3 The RobOff Software and Command Line Interface p 45 and Chapter 4 RobOff Graphical User Interface p 87 respectively Further details on how to use RobOff for different planning needs are provided in Chapter 5 RobOff analysis setups for common planning needs p 109 and Chapter 6 Tutorial and examples p 119 1 2 The RobOff framework in a nutshell Aim and purpose To provide a conservation planning tool that is able to evaluate the outcome of conservation and development actions and to optimally allocate resources to alternative actions taking into account their uncertain effects over time on biodiversity features in different environments RobOff is intended for a wide variety of conservation activities such as protection management maintenance restoration and offsetting Analyses Analysis of uncertainty of conservation value Analysis of the time perspective Reliability of biodiversity o
136. ncl costs Features Planning horizon Specific responses incl uncertainty envelopes 110 Environments actions biodiversity features and responses In a more comprehensive and strict sequence the process of defining a RobOff setup can be split into four steps Figure 5 2 Flow of definition of RobOff setups p 111 The definition of a setup can be seen as a first stage in the use of the RobOff framework and software leading to a second stage decision analysis and optimization and a third stage interpretation of results For details on optimization see chapter methods then command line and then GUI For interpretation of results the most relevant sections of this manual are Section 2 4 The RobOff output space p 23 Section 4 3 Results p 96 for the GUI and Section 3 4 Standard RobOff output p 70 for the command line tool Figure 5 2 Flow of definition of RobOff setups Adapted and extended from Pouzols amp Moilanen 2013 1 1 Define environments actions amp features 1 2 Define responses Input data description of real world mainly related to ecological entities End points amp shape Uncertainty envelopes Costs of actions Area available for actions Area of environments Input data on values preferences amp human decision making Treatment of uncertainty preference Performance criter
137. nclude different conditions levels of threat availability of data differences in administrative regions and or management practices and requirements and the scale and scope of analysis Environments are the entities at the highest level of abstraction in RobOff setups As such the choice of environments will condition the definition of all the other types of entities actions features responses costs etc The second step consists of defining possible actions in every environment for which a cost has to be estimated A global budget or budget range related to these costs should be estimated as well Next the set of biodiversity features present in each environment has to be defined Finally specific responses of biodiversity features to every possible action in every environment must be defined Before defining these responses the planning horizon that is relevant and or viable has to be decided upon Specific responses of features to actions are uncertain time discrete responses which has two implications 1 responses should be defined for a consistent sequence of intervals time span or planning horizon and 2 each response is threefold comprising best estimate upper envelope and lower envelope of the response Figure 5 1 A possible simplified sequence of steps required to define a RobOff setup Habitats threats condition administrative issues scale of analysis data availability etc Environments Budget i
138. nerate a budget analysis plot in the Budget tab of the results section In the budget analysis plot the x coordinate monetary amount at which the performance ratio attains a value of 1 is the minimum amount that answers the question of how much compensation is enough Note that if the amount of resources available was not enough you will need to increase the maximum of the budget range for which the budget analysis is performed 5 6 Interactions The RobOff conceptual model does not account for process based planning and dynamic interactions between features through time However it is for instance possible to enter interactions as simple biodiversity features or score features which are sums or products of several interacting factors See Section 3 5 What RobOff does not do directly p 82 for more details on the limitations of RobOff regarding interactions and how to incorporate simple models of interactions in RobOff setups 115 116 Part VI Tutorial and examples Table of Contents 6 Tutorial and examples oor lace inp tage T cap sactneaeethedvablantauaystuee tants 119 mE pM Aea 119 6 2 A first contact with the RobOff GUI ssssssseeee mene 120 6 3 A first contact with the RobOff command line interface 120 6 4 A minimal example 0c ccececeeceeee cece ee R 121 6 4 1 Minimal set of input files 20 eect tr
139. nformation that can be edited in this section is contained in the time discounting file see Section 3 3 6 Time discounting file p 65 Score features see Figure 4 10 Editing score features in the RobOff GUI p 95 The information that can be edited in this section is contained in the files of score features see Section 3 3 8 Set of files score features p 66 92 Setup Figure 4 6 Editing allocations in the RobOff GUI Roboff GUI View Results Tools Help DABifve o x G 4 Setup Environment Amount allocated 246 MiddleForest Amount Left 99754 UpperForest MiddleRiver Allocate actions a 1 eure ee rri a aa LowerRiver 1 do nothing o o o 2 Obstruct o o 3 ObstructPl 2 246 Error allocating resources You are trying to allocate too much The available area in this environment is 300 Results Optimize Figure 4 7 Editing benefit functions in the RobOff GUI RobOff GUI 1 0 0rc3 Dam_example second level_offset home fedemp example setup dam river Roboff GUI View Results Tools Help DW Birve o Q9 4 Setup Feature name eature weigh Functiontype Par 1 Par2 PaL3 Pan4 Par 5 1 River 1 1 0 25 2 ForestTypeA 1 2 0 25 0 5 3 ForestTypeB i 1 0 25 4 New featur 1 1 0 25 s 1 1 0 25 6 4 1 0 25 z 1
140. ng for feature response file setups setup simple demo response r 6 3 txt Regenerating uncertainty graphs for uncertainty parameter 1 Regenerating uncertainty graphs for uncertainty parameter 0 8 Setup setups setup simple demo successfully loaded 2 F OpenMP support there are 4 threads in parallel regions RobOff spatially implicit conservation planning T wo 97 N Note The actions comparison plots will be empty if no resources are allocated to none of the selected actions for any degree of uncertainty which can be fairly common in setups with large numbers of actions and or limited budget In such cases most comparisons between pairs of actions will result in empty or partially empty plots It is thus advisable to check first which actions are allocated some resources before starting to compare pairs of actions at random 4 3 4 Sensitivity of results to Budget variations Similar to the analysis of sensitivity to uncertainty in the budget tab it is possible to analyze how results vary as a function of available budget This type of analysis can be useful to compare the return on investment for different budget levels The range of variation of the available budget can be set into the setup general settings Alternatively see Chapter 3 The RobOff Software and Command Line Interface files p 45 for details on how to define the range of budget variation in input 102 Optimizati
141. nimal example This section describes what are the components required to build a minimal RobOff setup The set of input files presented is not intended to model any actual planning case but rather to a illustrate what are the strictly necessary inputs and b serve as a starting point to develop your own setups A Note This setup can be found in the RobOff software distribution under the folder directory setup minimal toy offsetting 6 4 1 Minimal set of input files This example setup includes the following files toy offsetting ro_setup General settings file Defines options such as the planning horizon uncertainty horizon time discounting parameter names of files etc environments csv Environments file Define the list of environments relevant to this problem 121 Output obtained features_present_environment 1 csv Feature presence file Defines what features are present in an environment environment 1 in this case which is the only environment response_donothing csv Response of the only biodiversity feature considered to the action do nothing or business as usual scenario Two additional response files are included response_development csv and response_compensation csv feature_weight_functiontypes csv File of feature weight function types where the feature specific weights and benefit functions are defined budget_allocation csv Budget allocation file containing a list of amoun
142. nstant functions are typically used for score features see Section 3 3 8 Set of files score features p 66 Section 2 3 Complementarity and scoring p 22 and Section 2 5 2 Aggregating conservation value p 26 In general it is possible to define arbitrary shapes with piecewise constant and piecewise linear functions This feature should be used with care as the use of piecewise functions with a large number of intervals can slow down RobOff computations significantly With a number of intervals of approximately ten or less computations will be as fast as for the simple functions In any case for simplicity and speed of computation we suggest using only the simplest possible types of functions unless otherwise required 60 Feature weights utility functions file Figure 3 3 Example benefit functions A Power diminishing returns B Step target C Concave ol D 1 sigmoid E Piecewise constant score F Piecewise constant score 1 2F 1 0 S 08 S 0 6 0 4 0 2 0 0 i i G exponential decay H GBF Piecewise linear 12r 1 0F 1 S08 J S 0 6 1 0 45 J 0 2 1 Uh 95m 10 15 2000 05 10 15 2000 05 10 15 20 Occurrence Occurrence Occurrence Different types of functions require different parameters all the functions require at least one This is an overview of the parameters needed in every case see the example
143. nting Biodiversity Note that the figure just shows a conceptually useful classification that is subject to exceptions and different interpretations For example the set of actions defined for a particular problem will normally depend on different pieces of ecological information and conversely the choice of view on substitutability can depend on what features and environments are under consideration Also for simplicity some factors are not shown in the figure such as costs In the case of costs one needs to consider the full set of costs of actions possible in every environment which is described more exactly as a tow level tree of costs as one more dimension for the output space Different results can be obtained for every different set of costs which can be efficiently visualized in the RobOff GUI In this view RobOff produces multiple outputs that can be considered alternative marginal projections of a high dimensional solution space into a lower dimensional space RobOff provides the outputs and tools necessary to analyze and visualize the effects of actions along different dimensions This is possible by processing the results obtained in output files see Section 3 4 Standard RobOff output p 70 or in a more interactively manner in the results section of the RobOff GUI see Section 4 3 Results p 96 Some output dimensions are 24 Aggregation of conservation value categorical and typically require choi
144. nts instead of results per feature Roboff results environment specific conservation value Format time cons val environmentl cons val environment2 cons val last environment Names of environments MiddleForest UpperForest MiddleRiver UpperRiver LowerRiver Results for No action Results for nominal robustness conservation value per environments Op 0 EA e 0 ilar Ike dE dL IO R DRC S 0 GSAS TL Ib at 12 sustainability ratios per feature csv This file is analogous to the file 4 sustainability ratios csv but contains sustainabiliy ratios for individual features in the sequence indicated by the list of names provided in the file header It is is generated only if the command line option f or per feature out is used Here is the first lines of an example file 79 Optional output files Roboff results sustainability ratios per feature All values are discounted Robustness requirement robust Info gap alpha 1 Names of features ForestTypeA ForestTypeB River SH oce c HEHEHE sb Results for No action sustainability ratio per features 0 3L 1 7 3E Oi the Ag Ak alo TE cbe 907 1E 1L ab GN Lo LE 4590 1L akg ab sustainability ratio per features time discounted OV ple E IE 29r tbo al gal ATS EN BOUWE SH aks aly al TOO 3L thy Ah Results for mandatory actions
145. oach which includes aggregation into score features In stage 2 the representation values of score and simple features are transformed into conservation values following a benefit function approach In stage 3 a global conservation value is calculated by aggregating individual values according to implemented actions and applying an uncertainty model which resolves the uncertainties propagated in stages 1 and 2 Finally after the time discount stage scalar values of conservation and performance are obtained 26 Aggregating conservation value Figure 2 4 Flow of aggregation of occurence levels and conservation value in RobOff Score transformations Representations gt of score components 1 Scoring Score aggregation simple features Fi re repr ntations Score features eature representatio benefit functions 2 Benefit transformation Uncertain conservation values of features across time Uncertainty model 3 Aggregation across environments Implemented and features actions Global conservation value across time Robust Nominal Opportunity 4 Aggregation across time Time discountin 3 and time discounting model Conservation value and performance ratio 27 Aggregating conservation value The mathematical symbols used in this manual are listed in Table 2 2 Mathematical symbols p 28 Occurrence levels of features are standardized responses that model the re
146. objectives for features restoration where the aim is to return past conservation values that have been lost and biodiversity offsetting where the aim is to compensate for ecological damage caused by human activity by managing or rehabilitating alternative sites The multi action problem is significantly more complicated and data hungry because actions have a range of costs and responses for features in different areas making analysis and optimization hard Ideally spatial interactions between conservation actions should be taken into account further complicating the problem The software Marxan with Zones provides for the allocation of sites to zones that have different conservation treatments i e local representation for a feature depends on the zone the site is assigned to Generic integer programming approaches have been applied to find optimal habitat restoration strategies Uncertainty about what happens to biodiversity features species habitats etc through time when landscapes and environmental conditions change and various actions are applied in different locations complicates resource allocation questions Time is particularly relevant in the context of offsetting where compensation measures are assigned to ecologically damaging economic activities Ecological damage is frequently immediate and certain whereas compensation via restoration will appear with a time delay and is not certain RobOff is intended to complement the m
147. of a file of features present in an environment For each habitat type a file with features present and actions that can be applied to these features features present estimate resp no action list of quartets actionf response end value uncert weight E Hi deis we devel action sec iki 0 active restor magmnt E 37 P3 1 0 E2 QA t 19405 devel astin Ta ir 10 active restor magmnt r 223 1 2 1 0 sil In benig neglect ees Loor LU p ip benign neglect r 707 0 8 170 N Note The parameters start_value per feature and end_value per response to action are not used by default and are not needed for simple uses of RobOff To enable it use the option enable rescaling of responses needs to be enabled in the general settings file Section 3 3 1 General settings file p 50 These parameters and the rescaling option are provided for convenience and are meant to be used with generic response shapes that can be rescaled between a minimum and maximum Under the assumption that responses are increasing or decreasing curves over time there are two possibilities 1 start_value is used as minimum and end_value as maximum monotonic increasing response or 2 vice versa monotonic decreasing response The effect is that the first and last points of responses are shifted to start_value and end_value respectively and the responses are rescaled or multiplied by an amplitude factor correspondingly Note that this o
148. of money into actions in environments is divided into a multidimensional grid at a resolution specified by the user All feasible solutions as defined by this grid are evaluated Exhaustive search is reliable easy to understand deterministic and straightforward to implement but suffers from the problem that computation time increases exponentially with the number of dimensions alternative actions The fourth optimization method expands the grid based exhaustive search by adding a local search process that is executed starting from each feasible grid point The last optimization method implemented in RobOff is stochastic global search using a genetic algorithm which is applicable to very large and complex problems but which cannot guarantee optimality of solutions For details on how to use these methods see Section 3 3 1 General settings file p 50 and Section 4 4 Optimization p 103 36 Optimizing resource allocation Table 2 4 Scalability and optimality of the optimization methods supported by RobOff general properties Adapted and expanded from Pouzols amp Moilanen 2013 Method Approximate maximum Optimality problem size Random Virtually unlimited None Greedy Virtually unlimited Hundreds Potentially very suboptimal or thousands of action Good first and fast environment pairs are no approximation in some cases issue Grid based Up to 8 10 dimensions action Optimal for the budget exh
149. on Figure 4 19 Visualizing results as a function of the available budget a d d Roboff GUI View Results Tools Help eg rv eo So 4 Setup Results Summary Time Uncertainty Environments Features Actions Compare Actions Budget Minimum 0 00 Maximum 182000000 Step 20 Analyze Weak environments El Dashed Add Clear 0 335 5 0 33 g 0 3254 and E P 5 z 032 4 v 0315 Su 0 314 r T T T 1 o 5e 07 1e 08 1 5e 08 2e 08 Available budget Optimize 4 4 Optimization In this section of the RobOff GUI you can find the optimal allocation of resources according to different criteria and constraints Once you select the necessary options and run the optimization process the optimal allocations found will be visualized on the right via a tree of actions to which resources have been allocated The main options that need to be selected before starting an optimization are Optimization method greedy search exhaustive search genetic algorithm grid search with local operator etc For details on the different optimization approaches see Section 2 6 Optimizing resource allocation p 35 Metric to optimize variant of sustainability index weak or strong performance ratio of sustainability across features or environments Robustness criterion robust nominal opportunity Additional options budget resolution time bounds etc 103 Preferences
150. ouzols F M and A Moilanen 2013 RobOff software for analysis of alternative land use options and conservation actions Methods in Ecology and Evolution in press DOI 10 1111 2041 210X 12040 Pouzols F M Burgman M and A Moilanen 2012 Methods for allocation of habitat management maintenance restoration and offsetting when conservation actions have uncertain consequences Biological Conservation 153 41 50 Moilanen A van Teeffelen A Ben Haim Y and S Ferrier 2009 How much compensation is enough A framework for incorporating uncertainty and time discounting when calculating offset ratios for impacted habitat Restoration Ecology 17 470 478 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 Moilanen A 2007 Landscape Zonation benefit functions and target based planning unifying reserve selection strategies Biological Conservation 134 571 579 Laitila J and Moilanen A 2012 Use of many low level conservation targets reduces high level conservation performance Ecological Modelling 247 40 47 Time preference and time discounting Green L and J Myerson 2004 A Discounting Framework for Choice with Delayed and Probabilistic Rewards Psychological Bulletin 130 769 792 Loewenstein G and J
151. pact in the final optimal allocation of forcibly including certain actions which may reflect for example political preferences a priori expert opinion etc Optimal This set will be automatically defined by RobOff by maximizing conservation value given the sets of mandatory and preset actions and the available budget These sets of allocations are specified in the budget allocation file see Section 3 3 Input files and settings p 49 When looking for the best set of actions to offset the impact of development development actions would typically be allocated in the set of mandatory actions More generally the sets of mandatory and preset 48 Input files and settings actions are useful when planning is done incrementally such as in the case of incremental reserve design 3 3 Input files and settings A complete and correct RobOff setup needs to be defined in a set of files before running RobOff from the command line Some of these files are always required while some others are optional A diagram that summarizes the different types of files is shown in Figure 3 2 Set of input files and their equivalent GUI dialogs p 49 These files list the entities and attributes that make up a RobOff setup such as environments features and responses The figure also includes the equivalent GUI dialogs see Chapter 4 RobOff Graphical User Interface p 87 for a description of the RobOff GUI where the same entities can be
152. pected occurrence level and representation of 2 feature f in environment discrete time sequence Uk p ul Info gap upper envelope for occurrence level and representation 28 Aggregating conservation value Symbol Description i Info gap lower envelope for occurrence level and iod representation Vj Conservation value of feature for a given set of actions and amounts implemented N weak Weak conservation performance index Two variants across features and across environments Three robustness requirements robust nominal and opportunity Nstrong Strong conservation performance index Two variants and three robustness requirements see N weak e Cost of action k in environemnt i Table Table 2 3 Key terms for understanding conservation value aggregation in RobOff p 29 lists some key terms needed to understand conservation value aggregation in RobOff The terms come from various research areas and traditions and have frequently been used in ambiguous ways Table 2 3 Key terms for understanding conservation value aggregation in RobOff Term Meaning in RobOff Occurrence level or value Level of occurrence of a biodiversity feature May refer for example to observed abundance richness probability of occurrence habitat suitability or values at different levels such as genetic diversity or species richness In the context of ecosystem services the level of occurrence
153. ption is provided for convenience and it has the same effect as pre processing response files and providing different rescaled response files to RobOff For responses that are non monotonic the interpretation of the rescaling is case dependent f Note The parameter uncert_weight per specific response is not currently used This type of file is related to the option per environment files prefix of the general settings file 3 3 5 Set of files responses of biodiversity features 63 Set of files responses of biodiversity features The specific responses of biodiversity features to actions are formatted in text files with three columns of numbers nominal lower envelope and upper envelope The columns especify estimate lower envelope and upper envelope of uncertain occurrence levels for a feature specific response see Chapter 2 Framework Methods and Algorithms p 17 for details on how RobOff handles uncertain responses A file of this type must be provided for every different response that appears in the files of type biodiversity features present in an environment see Section 3 3 4 Set of files biodiversity features p 62 If you forget any of them RobOff will detect it at the setup error checking stage The response files must be named according to the following convention concatenate the response file prefix with the name of the action and the extension csv by default For example if the respon
154. r bounc Upper bound a 4 r7 CC env OMT act business as usual 10 r4_CC_env_OMT act_set_aside r1_CC_env_OMT_act_clear_cut_ set_as 9 9 Om OF r16 HG env OMT act business as usual 30 r13_HG_env_OMT_act_set_aside 40 0 0803 0 00915 0 1514 du T10 HG env OMT act clear cut i set o SE r25 LTT env OMT act business as usual 722 LTT env OMT act set aside poi 0 242 0 0298 0 45425 aa r34 FS env OMT act business as usual 80 10 2698 0 0628 0 47685 r31 FS_env OMT act set_aside azil 28 F5_en OMT act ear cut set a 99 BAA r43_TTWO_env_OMT act_business as us 100 0 2934 0 0725 0 5143 J 140 TTWO env OMT act set aside d exl r37_TTWO_env_OMT act clear_cut_ _s 1 Len lv 120 0 2688 0 05685 0 4807 P 130 od j F pM Cu 8 01 O 50 100 150 200 250 300 Results Optimize 8 Feature response file 53 home fedemp postdoc hy mrg papers roboff clearcuts example finLand metso clearcuts setup response r8 CC env MT zl Feature response file 54 home fedemp postdoc hy mrg papers roboff clearcuts example finland metso clearcuts setup response r9 CC env VT a Found 6 actions subject to optimization in 3 environments D o xi l El 91 Setup All the elements described in Chapter 3 The HobOff Software and Command Line Interface p 45 can be edited via tables and various widgets For example it is possible to define simple and score fea
155. r send a letter to Creative Commons 444 Castro Street Suite 900 Mountain View California 94041 USA In short you are free to use distribute and reproduce this manual in any medium under the Attribution you must attribute the work and Share Alike you may distribute derivative works only under the same license conditions Please see http creativecommons org licenses by sa 3 0 legalcode for the full license legal code The RobOff software is Copyright C 2011 2013 Biodiversity Informatics Conservation Group University of Helsinki Development of RobOff was significantly supported by the European Research Council as part of the project Global Environmental Decision Analysis GEDA ERC StG Grant 260393 We also thank the Finnish Ministry of Environment and the Academy of Finland Centre of Excellence Program 2012 2017 for further support Key references about RobOff Pouzols F M and A Moilanen 2013 RobOff software for analysis of alternative land use options and conservation actions Methods in Ecology and Evolution in press DOI 10 1111 2041 210X 12040 Pouzols F M Burgman M and A Moilanen 2012 Methods for allocation of habitat management maintenance restoration and offsetting when conservation actions have uncertain consequences Biological Conservation 153 41 50 RobOff Software for Robust Offsetting Habitat Restoration Maintenance and Management RobOff is a conservation planning framewor
156. relevant to the extent to which the input data is an appropriate and representative model Spatial planning RobOff is not intended for explicit spatial planning It can however be used for extracting area targets See Section 2 7 Dealing with connectivity p 38 for some comments on how to incorporate connectivity into RobOff models and Section 5 2 Analysis types p 112 for how to use RobOff to extract area targets Process based planning and dynamic interactions RobOff is not intended for spatial dynamic planning It does not build any dynamic interactions of feedback loops into its computations In particular there is no direct support for process based planning and dynamic interactions See Section 2 8 Assumptions and limitations p 39 for a more methodological discussion of simplifications and assumptions made in the RobOff framework DX Tip Q Interactions can be represented in RobOff setups as biodiversity features whether simple or scores Note that scores are calculated as sums or products of simple features This can be an effective yet simple way of modeling interactions 82 Scheduling Scheduling RobOff does not implement any scheduling mechanism here we understand scheduling as dynamic planning through space and time Extensions might be added in the future to support simple approaches to dynamic allocation 3 6 Implementation details about RobOff RobOff was developed in the C language Ther
157. res and scores are entered into complementarity based computations thus combining both approaches We use complementarity in the sense that actions and their consequences need to be evaluated jointly aiming at a well balanced and cost effective outcome An important difference between the two approaches is that score values are typically ranked whereas complementarity implies iterative computation and typically or ideally would require some form of optimization Also scoring can be applied without full knowledge of occurrence of biodiversity features across the full landscape and a Score can be computed based on local information which is an important practical advantage for scores Scoring is used by many conservation agencies possibly due to its conceptual simplicity ease of implementation and low data demands in the sense that information is only required from the sites of interest not from the entire landscape 2 4 The RobOff output space The inputs to RobOff are descriptions of multiple entities together with their relationships This includes information about ecological entities such as biodiversity features and environments together with the responses of features to actions and a second class of inputs that are qualitatively different related to values socio political factors and human decision making such as costs substitutability criteria and availability of actions While RobOff can produce simple performance indices
158. ria such as weak or strong sustainability This can be interpreted as the most cost effective solution allocation of resources among alternative actions for a given budget The optimization process must also account for the area availability constraints and different costs of actions and it is subject to the set of mandatory actions Also an optimal budget allocation can be determined for the robustness or opportunity immunity functions robustness or opportunity versions of the performance ratios that will be defined below In their general form both the robustness and opportunity optimization cases are highly nonlinear problems and exact methods for linear and integer programming are not appropriate Effectively various sources of non linearities are involved in the aggregation of conservation value including cost functions responses of features benefit functions etc Different alternative and complementary optimization methods can be used in the RobOff framework For problems where the number of actions is lower than about 35 Optimizing resource allocation 10 exhaustive grid search methods with local search operators can be employed in order to find a global optimum However this approach does not scale well for a larger number of actions For a restricted subset of problems it would be possible to use convex optimization methods that are exact and scalable However this can only be applied when all the benefit functions as
159. rom the optimization process and an optimal allocation of resources i e a list of amounts allocated to actions Optional A pairwise comparison between actions that lists the optimal distribution of resources between them for a varying degree of uncertainty Optional 12 Major features See Section 3 4 Standard RobOff output p 70 for a detailed description of the output files that RobOff produces 1 6 Major features The RobOff framework contains many features that distinguish it from other biodiversity conservation tools Some major features include Methods for dealing with uncertain development of biodiversity features aiming at robust decisions Time discounting taking time preference into consideration Integrated use of complementarity and scoring approaches Environment habitat type and features priorities via weighting Automated comparison of alternative actions for varying uncertainty conditions Possibility to account for both positive and negative consequences of uncertainty opportunity and robustness analyses 13 14 Part ll Framework Methods and Algorithms Table of Contents 2 Framework Methods and Algorithms sssesssssseeene 17 2 1 The RobOff framewotK eiie teo ohne a dx peau cene Ga 17 2 2 RODO Setups iiao a E E E E E 20 2 3 Complementarity and scoring sssssssssee eene 22 2 4 The RobOff output space
160. rval The deviation is not necessarily symmetrical In a general info gap formulation for a representation level ri the envelope bound model is Ula ri ul D 00 9 o D P ah P 20 020 Note that both the upper and lower bounds scale with and can be as high or low as desired The computational model of RobOff has been defined in a way that reduces the amount of information required from the user while still accounting for important factors such as time and uncertainty Nevertheless the model is sufficiently flexible sothat domain specific subtleties can be effectively incorporated It is for instance possible to enter the same actual feature several times as different factors in the model possibly in different environments and either as a simple or score feature This flexibility can be exploited to incorporate interactions into RobOff models as described in Section 5 6 Interactions p 115 If only nominal values are considered i e there is no uncertainty at all then all occurrence values are nominal values rs P o o We will first consider this case for simplicity The calculations involve the info gap model and its uncertainty horizon time preference environments features actions and areas where these are performed The simplified pseudo code for the RobOff computations of the conservation value across features is as follows n For each feature Test in the landscape je Ne For each environment Eee
161. rvation complementarity and scoring Moilanen A 2008 Generalized Complementarity and Mapping of the Concepts of Systematic Conservation Planning Conservation Biology 22 6 1655 1658 Burgess N D Hales J D Rickets T H and E Dinerstein E 2006 Factoring species non species values and threats into biodiversity prioritisation across the ecoregions of Africa and its islands Biological Conservation 127 383 401 41 42 Part Ill The RobOff Software and Command Line Interface Table of Contents 3 The RobOff Software and Command Line Interface esses 45 3 1 Introduction and important general information about files 45 3 2 Running RobOff from the command line esse 46 3 2 1 Sets Of ACTIONS skiing iina n anaE AAE EE 48 3 3 Input files and settings 2 0 0 eee cece eee ee tree eee tree RANKEAR AREER 49 3 3 1 General settings file cece cece ee cece cece ae nna aa aaa 50 3 3 2 Environments file sess 58 3 3 3 Feature weights utility functions file eesesessseseesss 59 3 3 4 Set of files biodiversity features sssssee 62 3 3 5 Set of files responses of biodiversity features suse 63 3 3 6 Time discounting file essaiera E 65 3 3 7 Budget allocation file seresa skake 65 3 3 8 Set of files score features cccc
162. s tete t dte ete tete ea pu ed sx edu Ro a dud 115 Chapter 5 RobOff analysis setups for common planning needs 5 1 General remarks RobOff is designed to help answer questions concerning the type of actions that should be undertaken and how much of each type Normally this is just one quantitative aspect of the conservation planning process In a more comprehensive view the RobOff flow would be embedded or combined with general frameworks for conservation planning See for example Wilson KA et al 2007 Conserving Biodiversity Efficiently What to Do Where and When PLoS Biol 5 9 e223 Roughly speaking a RobOff setup consists of two parts see Figure 2 1 Basic components of a RobOff setup p 20 for a conceptual diagram The first part includes all the proper inputs and is more related to observations of ecosystems e g environments biodiversity features actions that can be undertaken and responses to actions The second part comprises inputs that can be seen as parameters and are more related to interpretation and human decisions e g degree of uncertainty robustness requirement discounting costs and budget in some cases etc Normally the second part is more variable as it essentially consists of parameters for calculations and optimizations peformed in RobOff For details on how to define a RobOff setup including all the aforementioned inputs and parameters in the RobOff graphical interface see Ch
163. s and benefit functions need to be defined consistently with the choice of scale For details see Section 2 5 Aggregation of conservation value p 25 The default value is in principle a good choice in most cases This option does not prevent the use of any value in the response files but the occurrence values specified in the responses will be treated differently The default range 0 to 1 corresponds to the typical case in which responses are in a certain range of values and the benefit functions are defined accordingly The maximum of the range does not need to be 1 but 1 can be used as a reference for good pristine satisfactory or similar occurrence levels What is important is that the same range or comparable ranges should be used for the range of variation of the responses of different features in different environments and the range of values for which benefit functions are defined for example definitions of benefit functions see Section 3 3 3 Feature weights utility functions file p 59 By default RobOff will calculate aggregated occurrence levels representation and retention averaged across Spatial units features and environments Other alternatives for the responses scales option are proportional gain and absolute If proportional gain is used benefit functions will be calculated on the ratios between response values over time divided by the initial response value Note that the same effect can be obtaine
164. s nothing but three fields environment action and amount area extent This type of file is related to the options budget preset budget and mandatory budget of the general settings file see Section 3 3 1 General settings file p 50 3 3 8 Set of files score features Score features are defined as combinations of simple features see Figure 2 4 Flow of aggregation of occurence levels and conservation value in RobOff p 27 for details on how conservation value is aggregated for simple and score features in the RobOff computational model In order to define how score features are aggregated from simple features it is necessary to Include the score components as features in the biodiversity features present file Provide an additional file score features file per each environment where there is at least one score feature The per environment score features files must be named according to the following convention concatenate the score features files prefix with the name of the environment and the extension csv by default For example ifthe score features files prefixis score features and there are two environments named forest and lake then the corresponding score features file names are score features forest csv and score features lake csv If a score features file is not provided for an environment RobOff assumes that all the features present in that environment should be treated as simple f
165. same plot the weak sustainability index over time for different individual features The table of results summary corresponds to the results summary file in standard RobOff outputs see Section 3 4 Standard RobOff output p 70 N Note This section explains how to visualize results interactively To save results into text csv files for later analysis you can use the option Results Save into text files from the main menu of the RobOff GUI or the equivalent save results from the tool bar Both can be normally found at the top of the main window This will generate a set of files in the specified output directory For details on the format of these files see Section 3 4 Standard RobOff output p 70 96 Visualizing results across different dimensions Figure 4 11 Visualizing results summary RobOff spatially implicit conservation planning Roboff GUI Edit View Tools Help fB TY 6 9 T Setup Results Summary Time Uncertainty Environments Features Actions Compare Actions Budget Settings Conservation value Robust Nominal Opportunity No action 0 7417 0 7417 0 7417 Mandatory actions 0 7029 0 7029 0 7029 Optimal actions 0 6791 0 6791 0 6791 Performance ratios for Mandatory actions Robust Nominal Opportunity Weak features 0 697 0 954 151 Strong features 0 510 0 812 1 29 Weak environments 0 580 0 640 0 902 Strong environments 0 438 0 508 0 859 Optimize Looking for feature
166. se files prefix option is set to response and there are three responses named business as usual act1 and act2 then three files must be provided response business as usual csv response act1 csv and response act2 csv Here is an example of response file RobOff response fille Response of feature 1 to action 1 Columns estimate lower envelope upper envelope 2007 M5 100 0 300 HOO 0 200 9 750 200 05000 600 OOF amo 1909 9 iu 9 100 0000 0 0 0 f o Sqeoqe eae es RO O O O O O O O O O st sk s o o o N o o response This type of file is related to the option response files prefix ofthe general settings file N Note Notice that in the examples above the names of the response files start with response_ because that is the default prefix for the names of response files This prefix is not part of the names of actions and is only used for file names The prefix can be any valid path The same applies to the default extension or suffix As an example the file response_business as usual csv would typically contain the response values best estimate and uncertainty envelopes that correspond to the response called business as usual For more details see the option response files prefix Note that the prefix is not part of the names of responses This means that the prefix should not be used in the fields of the environments file Section 3 3 2 Environments file p 58 or the files of
167. shows the responses of the Dam Forest River example in the setup section of the GUI for more details about defining setups and visualizing results with the GUI see Chapter 4 RobOff Graphical User Interface p 87 Figure 6 3 Feature responses of the Dam Forest River example in the RobOff GUI RobOff GUI 1 0 0rc4 Dam_example second _level_offset home fedemp roboff example setups setup dam forest river offset Roboff GUI View Results Tools Help Demg 7Zvo t 4 Setup i97 General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring Response Estimate wer bou Jpper boun r1 CC env MiddleForest act business as usual o 0 7 0 695 r2 CC env MiddleForest act Inundate loos r4_CC_env_UpperForest_act_business as usual 20 0 82 0 8051 0 8349 r5_CC_env_UpperForest_act_Restore 30 0 865 0 845 0 88485 O9 0 705 r6_CC_env_MiddleRiver_act_business as usual r7_CC_env_MiddleRiver_act_Inundate r8_CC_env_UpperRiver_act_business as usual 50 0 925 0 895 0 95475 0 85 r9_CC_env_UpperRiver_act_Obstruct 60 0 94 0 9053 0 9747 r10_CC_env_UpperRiver_act_ObstructPlusTT r11_CC_env_LowerRiver_act_business as usual r12_CC_env_LowerRiver_act_AlterFlow so Bape 6 9896 90 0 945 0 905 0 99455 0 75 40 09 0 8752 0 9248 70 0 945 0 905 0 98465 0 8 100 0 945 0 905 0 9995 0 74 0 65 i b r T T T T T T T T T 1 Results oo ptimize
168. sts of different types of entities with well defined interactions The main entities are Environments which in the simplest case are equivalent to habitat types Actions including development and compensation conservation actions Biodiversity features Specific responses of features to actions in different environments In spatial planning tools data requirements are mainly layers of distributions of features This usually involves the use of GIS tools As opposed to this the essential data required to use RobOff are the responses over time of biodiversity features to actions Responses are entered as three time discrete values best estimate upper envelope and lower envelope This comes down to text files with three numeric columns or tables with three columns in the graphical interface Responses describe the evolution of the occurrence levels of biodiversity features for a given action in a certain environment Consequently one must first consider what the relevant biodiversity features and environments are and what actions are possible or compulsory 1 3 2 Outputs RobOff implements a computational model that aims at evaluating the outcome of sets of conservation and or development actions under uncertain conditions Different views on time preference and robustness to uncertainty such as robust or opportunity values are supported RobOff can also find optimal allocations of conservation resources to alternative actions for
169. t 1 River O r8 CC env UpperRiver act business as usual MiddleRiver 3 r1_CC_env_MiddleForest_act_business as usual fi LowerRiver r2_CC_env_MiddleForest_act_Inundate 13 CC env MiddleForest act Restore r4 CC env UpperForest act business as usual r5 CC env UpperForest act Restore r6_CC_env_MiddleRiver_act_business as usual Response to actions r7 CC_env_MiddleRiver_act_Inundate Action r8 CC env UpperRiver act business as usual 1 Obstruct 9 CC env UpperRiver act O 2 ObstructPlusTT 710 CC env UpperRiver act ObstructPlusTT 5 r11_CC_env_LowerRiver_act_business as usual r12_CC_env_LowerRiver_act_AlterFlow 4 Results Optimize 8 Optimizing budget allocation to 3 actions in 5 environments Genetic algorithm search finished Trials 150500 Budget spent in best solution 99920 Performance 0 866432 Optimization finished at 12 07 39 on Thu Nov 8 2012 took 8 479 seconds Log Figure 4 5 Editing uncertain responses in the GUI RobOff robust offsets calculator Roboff GUI View Results Tools Help Dgemz7zve o o 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs l Discounting Scoring 0 6 f Response Estimate owe
170. t it cannot find certain files check first that the relative or full paths to your input files are correct Also use special or not so common characters in file names with extreme care see section Section 3 3 Input files and settings p 49 Computer memory capacity In principle RobOff can use as much memory as available in 64 bits systems If at some point RobOff runs too slowly it is a good idea to start by checking how much RAM memory is actually available If your system memory is exhausted by RobOff or other processes RobOff may start using virtual memory hard drive This will result in overly poor performance and you should definitely free RAM memory or use a machine with more memory available Parallel execution RobOff makes intensive use of multi core systems If you experience significant performance degradation check that the number of cores in use by RobOff processes possibly running in parallel is not higher than the number of cores available in your machine Use the m or max cores command line option conveniently especially if you are running several RobOff processes in parallel Differences in the length of responses Should not be an issue as long as all the responses have a number of samples equal or greater than the length of the time or planning horizon However mixing responses of different length can be a very error prone practice Decimal separator and commas Do not use commas as decimal separators
171. t rate Hyperbolic s J O Constant Discount weights amp Time dependent Time Weight s cm TN I 1 1 o 2 5 L 1 25 0 035 2 1 Is A 2 25 p 3 25 0 03 4 4 1 B F 4 25 i z ag 0025 meee 6 1 7 i 6 25 p 7 25 0 02 4 8 1 P i 8 25 L 9 25 0 015 4 10 1 I m 1 10 25 0 01 12 1 11 25 r T T T T T 1 I o 5 10 15 20 25 30 E 12 25 Rate Results Optimize Robust offsetting configuration 94 Setup Figure 4 10 Editing score features in the RobOff GUI RobOff GUI 1 0 0rc3 censored_name_second_level_offset home fedemp example setup dam Roboff GUI View Results Tools Help Demg 7vo o s o9 TE 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring Environment Score components Score features MiddleForestA MiddleForestB UpperForest Name ponent weic Aggregation type Name Features ForestTypeA Results 9 Optimize 95 Results 4 3 Results This section of the GUI shows global results as well as results through time and across environments features and actions The first leftmost tab in the results section of the RobOff GUI shows a table with a summary of results The remaining tabs allow inspecting results across different dimensions time environments features actions etc In addition there are tabs for specific results such as comparisons of actions These ar
172. take into account explicit information about the spatial distribution of biodiversity features and habitats or environments This is a necessary compromise to reduce data demands and to make it possible to solve allocation problems within reasonable computational resources See Section 2 7 Dealing with connectivity p 38 for effective ways of dealing with connectivity In addition to space other factors have been intentionally omitted in the RobOff framework It does not account for process based planning and dynamic interactions between features through time While a conceptual limitation this is perhaps less relevant in practice as the data demands of including dynamic interactions would in most cases be prohibitive In such a case one would be approaching a parametric dynamic model of landscape and metacommunity dynamics something that can albeit with considerable difficulty be implemented using generic dynamic modelling software Note however that it is possible to consider simple forms of interactions in RobOff by defining simple or score features that correspond to interactions see Section 5 6 Interactions p 115 for details Implementation limitations In principle there are no hard limitations to the computations described in previous sections See Section 3 7 Data limitations and system requirements p 83 for more specific implementation details of the RobOff software Memory limitations RobOff imposes no har
173. tions which may include optimization of resource allocation The results of these calculations are written into output files In actuality the RobOff command line can be used together with the GUI You can actually create setups via the GUI and save them for later processing using the RobOff command line Likewise results generated using the command line can later be opened and visualized using the GUI However in this chapter we concentrate on the workflow parts see Figure 3 1 Running RobOff stages from inputs to outputs p 46 that are directly performed using the RobOff command line This is the output obtained if you type roboff or roboff v as a command RobOff software for allocation of conservation effort with multiple actions Version 1 0 0rc4 Soviet o EP ONES SINS Biodiversity Conservation Informatics Group Center of Excellence in Metapopulation Biology University of Helsinki http www helsinki fi bioscience consplan Usage roboff OPTIONS Options hacer Print help and exit Nip D ie Siakiouat Print version and exit V verbose Verbose output s FILENAME DIR setup Set setup file directory OSDIR Olt pulls Set output directory name ES O pnm es isle c Find an optimal allocation of resources Su cuncertaudnty Perform uncertainty analysis of cons value k compare actions Perform comparison of actions b budget analysis Perform budget analysis t FILENAME time disc Use time discountin
174. to environments operational blocks of area units to be dealt with as a whole for planning purposes The degree of uncertainty in the responses of features should be given special attention It is also key to decide how the importance of different biodiversity features and environments are weighted in relative terms 4 Develop the data to represent the ecological model and social or human factors The relevant data can be formatted into text files and or tables in a graphical interface It should be noted that the points above are better accomplished by or in cooperation with experts and different stakeholders In practice these stages are part of an iterative process in which the aims ecological model and social or human factors must be revisited in the light of new information In particular data availability can be a fundamental limiting factor In fact before fixing the broad aims and structure of the ecological model it should be ascertained what data is available or can be generated with reasonable effort In RobOff setups emphasis is placed on two aspects time and uncertainty It is therefore most important to be aware of what time scales are relevant and known what is the evolution of ecological factors over time and how to quantify uncertainties in responses As a final cautionary note the relevance and usefulness of results depend strongly on how much data is available Usual concerns in conservation planning exercises shou
175. to enable messages about general RobOff operation such as consumption of resources memory and processing time progress of computations etc Number of cores to use in parallel calculations By default the maximum number of available cores will be used This is especially relevant when optimizing resource allocations Remember to reduce this number if you want to perform other tasks in the same computer possibly several RobOff optimizations in parallel Background color in plots This parameter is used when saving plots or copying them into the clipboard Its default value is white with alpha transparency for those formats that support alpha channel transparency such as png 104 Preferences Figure 4 21 Editing RobOff GUI preferences Preferences Global Log window Plots Y Show general information Error reporting level Errors only Warnings and errors Notices warnings and errors e Pa 105 106 Part V RobOff analysis setups for common planning needs Table of Contents 5 RobOff analysis setups for common planning needs eeeeeeeeeeeeeeeeneeeeeeeeeees 109 5 1 General remarks eer dt ete Ded hte cena a a SR Enea o vx oceans e DR EE eel ss 109 S2 AMAIYSIS TYPOS Re m 112 5 3 Uricertalnty iiec pobres veut cedex v gera kr teca a acta tke cda v oda uaria la 113 NI RC LEES 114 Menem Tm 114 5 5 1 How much compensation is enough 5 0 Interaction
176. to uncertainty pairwise comparison of actions or sensitivity to available budget can be performed in the different tabs of the results section See Section 4 3 Results p 96 for details on what results can be visualized in the results section To obtain an optimal allocation of resources go to the opt imize section of the GUI You can use the default settings or select a different performance criterion robustness requirement level and optimization method Try different options and compare the solutions that you obtain on the optimal allocation tree where you can visualize the amounts allocated to each action 6 3 A first contact with the RobOff command line interface The aim of this example is to learn how to edit the files that make up a RobOff setup as well as the output files generated by RobOff To learn how to find optimal allocations of resources by using the command line interface of RobOff If you prefer not to use the command line interface of RobOff just skip this section See Section 6 2 A first contact with the RobOff GUI p 120 for a first contact with the RobOff graphical interface Follow these steps Find the first contact command directory in the RobOff distribution If your current working directory is the root of the RobOff installation the general 120 A minimal example settings file main file for this setup is example setups first contact command first Contact command ro setup
177. ts allocated to different actions Setups can also be defined and modified interactively in the RobOff GUI If you load this setup into the GUI you can see the contents of this setup in the different tables included in the tabs of the setup section For example the environments and actions defined are shown in two tables in the Environments actions tab of the setup section of the GUI see Figure 6 1 Defining environments and actions in the setup section of the RobOff GUI p 122 Figure 6 1 Defining environments and actions in the setup section of the RobOff GUI RobOff 1 0 0rc1 RobOff minimal toy offsetting home fedemp roboff setups setup minimal toy offsetting Roboff GUI View Results Tools Help DABifve e o o 4 Setup General settings Environments Actions Features Responses Allocations Functions Costs Discounting Scoring Environments Actions Name A Weight Total area Condition Name Per unit cost Area available 2 developmentAction o 2 3 compensationAction 30 2 Results C Optimize 6 4 2 Output obtained 122 Output obtained By running a command line the summary shown below is obtained A similar summary table is shown in the Summary tab of the results section of the GUI Figure 6 2 Summary of results for the minimal setup in the RobOff GUI p 124 You can visualize different results in the multiple tabs of the results section See Section 3 4 Standard RobOff output
178. tures as well as the time discounting model and parameters These are the tabs where additional setup details can be defined Allocations see Figure 4 6 Editing allocations in the RobOff GUI p 93 The information that can be edited in this section is contained in the budget allocationen file see Section 3 3 7 Budget allocation file p 65 Note that this tab is about predefined allocations for how to obtain optimal allocations see Section 4 4 Optimization p 103 Benefit functions see Figure 4 7 Editing benefit functions in the RobOff GUI p 93 where the weights and utility functions for every biodiversity feature present in any environment can be edited In this tab every row corresponds to a different biodiversity feature The information that can be edited in this section is contained in the file of Feature weights utility functions see Section 3 3 3 Feature weights utility functions file p 59 Costs see Figure 4 8 Editing costs in the RobOff GUI p 94 The information that can be edited in this section is contained in the environments file for constant costs see Section 3 3 2 Environments file p 58 or in the cost files see Section 3 3 9 Set of files costs of actions p 68 for variable costs either as an area cost curve or as a time dependent curve Time discounting model and parameters see Figure 4 9 Editing time discounting model and parameters in the RobOff GUI p 94 The i
179. ty see Sustainability file 65 92 Web site 9 GUI 92 Weight 4 hyperbolic 34 53 environment 4 8 32 58 quasi hyperbolic 34 53 feature 8 32 59 92 Transparency see alpha transparency species 4 Troubleshooting 84 time 114 Tutorial 119 Windows see Operating system example 120 120 121 124 127 Work flow 6 109 WWW 9 U Uncertain Z consequence 40 Zonation 19 25 38 61 see Benefit function response 40 64 90 Uncertainty 21 112 alpha 100 analysis 48 52 73 75 100 112 112 114 budget 25 degree 52 GUI 100 horizon 25 73 100 Model 22 135 136 Please check our website for the latest version and news www helsinki fi bioscience consplan Federico M Pouzols and Atte Moilanen The following partners have supported the development and implementation of RobOff European Z 5 2 Research t Council M iac ACADEMY METSAHALLITUS OF FINLAND UNIVERSITY OF HELSINKI Contact information Department of Biosciences Biodiversity Conservation P O Box 65 Viikinkaari 1 Informatics Group Fi00014 University of Helsinki Finnish Centre of Excellence in Finland Metapopulation Biology University of Helsinki ISBN 978 952 10 8720 2 paperback ISBN 978 952 10 8721 9 PDF
180. vation value of the total retention in the environments considered V g t and denotes the conservation values obtained when no human intervention takes place i e business as usual scenario or in other words when only the do nothing action is performed everywhere throughout the landscape Note that the aggregation process described here involves a number of non linearities responses of features benefit functions etc 2 5 3 Weak and strong sustainability We can define a performance ratio of sustainability relating what is obtained for a particular allocation of resources against what would be obtained if no action is undertaken business as usual scenario Four options can be distinguished depending on whether sustainability is considered in a weak or strong sense and whether it is calculated across features or environments Strong sustainability implies that kind must be replaced with kind whereas weak sustainability implies that some particular type of loss can be compensated by gains of a different kind Measures of strong sustainability like the performance ratio defined here emphasize the non substitutability or non interchangeability of biodiversity features 32 Time discounting Weak sustainability across features nf i o yo j Js n V 2 V J pfeat N weak Strong sustainability across features V 0 f J N strong me V 9 Weak sustainability across environments If
181. ve for the 1 summary csv output file The rows of these matrices correspond to the different time intervals considered in the planning horizon indicated in the first column whereas the columns correspond to the different values of the uncertainty horizon range The following listing shows the beginning of a file of this type 73 Standard RobOff output Roboff results conservation value Results for nominal robustness Uncertainty horizon alphas ORO Om 201010 3 010 1004 0 0 O0E500MONOOONOSOOOSSOOPONOOOTES0 Time Avg conservation value across features 0H PO Ee aE S 0 SaaS OPI 0 5 Syabeesy 10 5 Shales 0 Yul CNS 10 easy OPSORTOPORSOFULIB LO sss A R Eo P 0 5 dal NS 10400709136 0r SSO SL HS Sal sSiGy 0 4 Salsa Gal BGip O PIOS GP 0 5 Sab SiS 5 aL SKS U SSN 0 Sal 3G As OO OSE N04 Sul 2 0 6 SLAG 0 Sal 2s Wy Sales Ole SULA ae A E ATE A a A 0 4 Salish Os SiLAshy 4 aL Zs 305 00 O Sulake 7S ook oE a Da les PA ee a Ma a E 0 5 Qiks Ox Sakae Oy a Ly e a Ae Pao ae AO P SO SI ak AN OO 0 Sal aL ONSOSIEO PE SO TITRE ONIS 0 Sk ALO 10 5 Gal Ly 0 S10 10 D RaP 10 Sak OSU Ea SN Oly 0 Sal ako 50 0070 9101 Qro Oso 5 0 SLO 0 Sab aly 0 Sal Oils 0 Saba 0 CHLO aL 0910r sS aL OMS ESOS 60 5000 SOD270 SOS O M092 WINS 2 0 90927 CsS092 0 2094 059082 0 9092 0 3092 0 9092 70 00 0 9082 0 9082 0 9082 0 9082 0 9082 0 9082 0 9
182. weights of biodiversity features see Section 3 3 3 Feature weights utility functions file p 59 and environments see Section 3 3 2 Environments file p 58 In principle these first analysis do not require any optimization 2 Identify a base analysis After getting some basic RobOff examples running you need to make a number of decisions regarding the most important entities both in the ecological model and the human and social factors a Decide the set of environments related to the relevant habitat types b Define the set of actions for every environment and how they are linked to different responses of features You should also consider their costs and how much area is available for every action c Define the weights and benefit functions for features and environments 3 Base analysis and sensitivity analyses At this point you have a baseline setup from which you will normally generate several variants a Run the base analysis and have a look at different results whether in the graphical interface or output text files You can inspect nominal opportunity and robust conservation values sustainability values etc You can do this Software installation and quick start for the business as usual scenario i e no conservation resources allocated at all b Set a value for the available budget and find an optimal allocation of resources This can be done for the robust option for example You can also
183. well as the constraints are convex i e a straight line segment between two points in the function always lies above the function For instance one can have a non convex constraint if the per area unit cost of an action depends on the total area in a non convex manner In a general case stochastic optimization methods such as evolutionary optimization approaches and genetic algorithms in particular are an efficient means for finding satisfactory solutions in reasonable time Five different optimization methods are presently implemented in the current version of RobOff Some general properties of these methods currently are summarized in Table 2 4 Scalability and optimality of the optimization methods supported by RobOff p 37 which gives an idea of the applicability of the methods to different planning problems and Table 2 5 Speed and ease of use of the optimization methods supported by RobOff p 37 which is informative about the practical usability of the software tool The first method is based on allocations of random amounts of resources to actions picked at random It is meant as a reference method that provides baseline results to compare against more elaborated methods The simplest method is a greedy search algorithm It is meant as a fast way to obtain a first approximation for large problems and can be effective and nearly optimal in some cases In the third method grid based exhaustive search the search space division
184. wn below there are 3 alternative actions in 2 environments 3 pairs of possible comparison within each environment Roboff results comparison of actions for different degrees of uncertainty Format uncertainty alpha area to 1st action money to 1st action area to 2nd action money to 2nd action area to last action money to last action These are the environment action names MiddleForest do nothing MiddleForest Inundate MiddleForest Restore UpperForest do nothing UpperForest Restore 0 0 0 300 0 200 1000 10200 0 19800 99000 0 0 300 0 200 1000 10200 0 19800 99000 0 300 0 200 1000 10200 0 19800 99000 0 300 0 200 1000 10200 0 19800 99000 0 300 0 200 1000 10200 0 19800 99000 300 0 200 1000 10200 0 19800 99000 9 budget analysis csv This file is generated if the budget analysis is enabled b or budget analysis command line options The budget analysis can 77 Optional output files be useful to compare the return on investment for varying budget levels This file shows for the budget range the following performance indices weak and strong sustainability ratios across features and environments The first column corresponds to the different budget levels The values provided in the other columns are the different variants of sustainability ratios and conservation value weak or strong and across features or environments This analysis requires specific options in the general settings see Section 3
185. ws tables and plots Tracking of the optimization process Browsable and searchable help manual The RobOff GUI is relatively self explanatory Probably the best way to learn how to use itis by experimenting and looking at examples It should be possible to start using the RobOff GUI almost straight out of the box as long as one has a working understanding of the RobOff conceptual model i e how to define a RobOff setup actions environments features responses etc The next sections describe the basic structure of the RobOff GUI and its main features 4 1 Main window From the main window you can open an existing setup create a new setup save the current setup edit a setup check the validity consistency visualize results for different allocations or resources and find optimal allocations for different criteria These options are also available from the main toolbar at the top of the graphical interface There is also an online help system Most of its content is shared with this manual The graphical interface consists of three main blocks or sections that can be opened one at a time setup results and optimization These three sections are part of a toolbox widget n the setup section you can configure general settings and edit all the entities that are part of a RobOff setup see Chapter 2 Framework Methods and Algorithms p 17 environments biodiversity features per area unit responses of features to actions
186. y actions sustainability ratio per environments OPO T952 qti a Oe gaT TORBAT AV On IISA 1 hy Op WSS 0 ots 19 14 costs_over_time csv This file is generated only if at least one time varying cost is defined see Section 3 3 9 Set of files costs of actions p 68 For the possible allocations whenever available it lists the costs over time of allocations to actions only for those actions with time dependent costs With this information it is possible to calculate benefit cost ratios or similar measures over time Here is an example excerpt of costs over time output file 81 What RobOff does not do directly Format of lines time cost_actionl cost_last_action only actions with time dependent costs included Names of environment actions MiddleForest Restore UpperForest Restore UpperRiver ObstructPlusTT Costs for No action Costs for mandatory actions 0 50000 20000 1000 20 20000 8000 220 40 5000 5000 240 60 0 0 260 80 070 290 100 790 300 3 5 What RobOff does not do directly Fixing inadequate data RobOff cannot generate correct or fix your data All that can be done is to detect and report inconsistencies in the input files RobOff will surely produce conservation values sustainability ratios optimal allocations of resources and other results no matter what data is provided as input However these results will be informative and
187. yla for providing example data and comments We are also grateful to all the participants of the first RobOff workshop held in Melbourne in July August 2012 for their comments and suggestions Thanks as well to Brendan Wintle Terry Walshe Mark Burgman and other colleagues from the Department of Botany of the University of Melbourne and the Australian Centre of Excellence for Risk Analysis ACERA for supporting and hosting the RobOff workshop Development of RobOff was significantly supported by the European Research Council as part of the project Global Environmental Decision Analysis GEDA ERC StG Grant 260393 We also thank the Finnish Ministry of Environment and the Academy of Finland Centre of Excellence Program 2012 2017 for further support Table of Contents k Intro CUCU ON agi bss ERE RERO RR isd E 1 1 Inicere Nein 3 1 1 Aims and purpose pd aea EEEa e 3 1 2 The RobOff framework in a nutshell ssssssssss 4 1 3 RobOff inputs and outputs sssseseseeee 5 1 94 Inputs oi Eee erre ebd deo ette Leda mes 5 1 9 2 OUIDULS kern eie eden ende e eene ee cuente 5 1 4 A typical RobOff work flow ssseeeen 6 1 4 1 Specification of aims ecological model and data 6 1 4 2 Getting a RobOff analysis running sss 8 1 5 Software installation and quick start ssssssssss 9 1 6 Major features 22 0 0 ec cec

Download Pdf Manuals

Related Search

Related Contents

NComputing IP Utility (NIU) User Guide  PORT UGUÊS - Instructions Manuals  SEGADORA DE ECONOMIA, ROTATORIA, A EMPUJE  Triarch 31086-15 User's Manual  Princess 3in1 Steam, Bread & Pizza Maker  Zanussi IH6048B Use & Care Manual  2015年度案内 Tシャツ購入  Bedienungsanleitung  Impex Marcy MD-PWR-20 User's Manual  

Copyright © All rights reserved. Failed to retrieve file