Conservation by Design is currently under construction. Please check back next year.

Appendix J: Strategy Selection Tools

Con­se­quence tables

In many cas­es, good solu­tions can be found with­out fan­cy soft­ware or analy­sis, sim­ply by look­ing care­ful­ly at the esti­mat­ed con­se­quences of the options being con­sid­ered. A con­se­quence table has the options being con­sid­ered list­ed down one side and the objec­tives list­ed across the top, and the esti­mat­ed con­se­quence of each option on each objec­tive pop­u­lat­ing the cells. Estab­lish­ing a con­se­quence table is also the per­fect plat­form from which to use the next two types of meth­ods we cov­er; mul­ti­cri­te­ria deci­sion analy­sis and return-on-invest­ment.

  • Remove options that are infe­ri­or for every dimen­sion. An option is infe­ri­or if there is anoth­er option that per­forms at least as well against all objec­tives and bet­ter than it for at least one objec­tive. If such a sit­u­a­tion exists, there is no ratio­nal rea­son for choos­ing the infe­ri­or option so it can be dis­re­gard­ed as an option.
  • Remove options that are large­ly infe­ri­or for every dimen­sion. Some options may be rough­ly equiv­a­lent for most objec­tives but sig­nif­i­cant­ly infe­ri­or in oth­ers. These also can be discarded.
  • If some­one dis­agrees with remov­ing an infe­ri­or option, con­sid­er whether this is because it met an addi­tion­al objec­tive that was not con­sid­ered in the con­se­quence table. If so, this might require going back and adding anoth­er objective.

Mul­ti­cri­te­ria Deci­sion Analy­sis (MCDA)

Con­se­quence tables. Mul­ti­cri­te­ria Deci­sion Analy­sis (MCDA) is a broad term that encom­pass­es a range of meth­ods for incor­po­rat­ing mul­ti­ple objec­tives into the eval­u­a­tion of alter­na­tive options. Equiv­a­lent terms include mul­ti­cri­te­ria deci­sion mak­ing (MCDM) and mul­ti­cri­te­ria analy­sis (MCA). Cri­te­ria should include all those con­sid­er­a­tions by which the per­for­mance of a strat­e­gy should be eval­u­at­ed (but lists of cri­te­ria are no replace­ment for clear iden­ti­fi­ca­tion of objec­tives). The basic ele­ments of a MCDA, are:

  1. A set of cri­te­ria against which the desir­abil­i­ty of strate­gies are to be judged.
  2. Assign weights to cri­te­ria. Weights reflect the impor­tance of cri­te­ria in deter­min­ing the out­come. Weights might be assigned indi­vid­u­al­ly or as con­sen­sus amongst a group.
  3. Com­bine an assess­ment of the per­for­mance of each strat­e­gy for each cri­te­ria with the weight for that criteria.
  4. Aggre­gate scores for each strat­e­gy across all cri­te­ria to give an over­all assess­ment of per­for­mance or util­i­ty.

Although there are many ways that cri­te­ria weights and per­for­mance scores can be com­bined to cal­cu­late util­i­ty, by far the most com­mon is known as a lin­ear addi­tive ben­e­fit func­tion or model:

where Uj is the over­all ben­e­fit or util­i­ty of strat­e­gy j, Wi is the weight giv­en to cri­te­ri­on i and Vij is the val­ue of option j for cri­te­ri­on i.  In sim­ple terms this means that for each strat­e­gy, the per­for­mance score for cri­te­ria 1 is mul­ti­plied by the weight giv­en to cri­te­ria 1; the same cal­cu­la­tion is made for each addi­tion­al cri­te­ria, and then these are all added togeth­er. The result is an over­all score for each strategy.

  • Nor­mal­iz­ing data. Cri­te­ria can be mea­sured on very dif­fer­ent scales. In order to cal­cu­late util­i­ty using a lin­ear addi­tive func­tion, it is nec­es­sary that the per­for­mance val­ues of strate­gies for dif­fer­ent cri­te­ria are mea­sured on the same scale; oth­er­wise those cri­te­ria mea­sured on scales with larg­er num­bers will be unin­ten­tion­al­ly weight­ed more heav­i­ly. The process of con­vert­ing data to the same scale is known as nor­mal­iz­ing.

There are dif­fer­ent pro­ce­dures for nor­mal­iz­ing data but one that is con­sid­ered robust for use in MCDA6 is giv­en by the formula:

where Vij is the per­for­mance of strat­e­gy j for cri­te­ri­on i, and min[Vi] and max[Vi] are the min­i­mum and max­i­mum pos­si­ble val­ues for cri­te­ri­on i. The prime mark after V on the left hand side of the equa­tion is stan­dard nota­tion that indi­cates the para­me­ter is a trans­for­ma­tion of anoth­er parameter.

  • Weight­ing cri­te­ria. Cri­te­ria are very unlike­ly to all be of equal impor­tance in deter­min­ing which strat­e­gy is best. Weight­ing is intend­ed to help ensure that the cal­cu­la­tion of util­i­ty actu­al­ly reflects the per­ceived impor­tance of the dif­fer­ent cri­te­ria. The weight­ing giv­en to cri­te­ria can sig­nif­i­cant­ly change the out­come of an MCDA, and there­fore it is real­ly the sys­tem­at­ic and trans­par­ent weight­ing of cri­te­ria that is the sub­stance of an MCDA. An impor­tant and under-appre­ci­at­ed point about weights is that they are only mean­ing­ful with ref­er­ence to the observed range of out­comes for each cri­te­ri­on. In oth­er words, the impor­tance of a cri­te­ri­on in influ­enc­ing a deci­sion depends both on its inher­ent impor­tance to the objec­tives, and on how well or poor­ly the strate­gies under con­sid­er­a­tion per­form for that cri­te­ria. For exam­ple, a stake­hold­er group might val­ue recre­ation oppor­tu­ni­ties over bio­di­ver­si­ty con­ser­va­tion, but if the options strate­gies being con­sid­ered dif­fer lit­tle and are all sat­is­fac­to­ry in their con­se­quences on recre­ation oppor­tu­ni­ties, then it does not make sense to assign the recre­ation cri­te­ri­on a rel­a­tive­ly greater weight. Giv­ing greater weights to cri­te­ria that do not vary much has the effect of mak­ing strate­gies seem more sim­i­lar in util­i­ty val­ue, which is not infor­ma­tive for deci­sion mak­ing. This means that it is pre­ma­ture to weight cri­te­ria before know­ing the expect­ed per­for­mance of strate­gies against these cri­te­ria (hence the impor­tance of some form of con­se­quence table).
  • Use a defen­si­ble approach to devel­op cri­te­ria weights.Weights of cri­te­ria in an MCDA embed the val­ue judge­ments of those doing the weight­ing. As such, weights should not be seen as objec­tive assess­ments of the rela­tion­ships between cri­te­ria, and might ben­e­fit from includ­ing sev­er­al per­spec­tives pri­or to final selec­tion of weights. It also means that it is impor­tant to assign weights through a process that makes these val­ue judg­ments clear. Good approach­es to estab­lish­ing weights include Swing Weight­ing and the Ana­lyt­ic Hier­ar­chy Process.

Return on Invest­ment (ROI)

In the field of con­ser­va­tion, return-on-invest­ment (ROI) has come to be a rather gen­er­al term for pri­or­i­ti­za­tion approach­es that explic­it­ly con­sid­er the cost of the strate­gies being con­sid­ered. Con­ser­va­tion ROI analy­sis belongs to a gen­er­al class of eco­nom­ic analy­sis known as cost-effec­tive­ness analy­sis. Econ­o­mists con­sid­er some­thing to be a cost-effec­tive­ness analy­sis rather than a cost-ben­e­fit analy­sis when the out­come or return side of the equa­tion is not mon­e­tized (for exam­ple, expressed as a dol­lar val­ue). The basic notion behind ROI or cost-effec­tive­ness is that the expect­ed return or out­come from a con­ser­va­tion action should be bal­anced against the cost of achiev­ing that out­come. Although they are not always recog­nised as such, the Con­ser­van­cy’s Ecore­gion­al Assess­ments were often a sort of ROI analy­sis, par­tic­u­lar­ly when they are imple­ment­ed using opti­miza­tion soft­ware such as Marxan.

The cost-effec­tive­ness, CE, of tak­ing action i in place j can be giv­en by the gen­er­al equation:

where Bij is the ben­e­fit of tak­ing action i in place j, Cij is the cost of tak­ing that action, and Prij is the prob­a­bil­i­ty that if tak­en, the action will deliv­er the expect­ed ben­e­fit. This last term is not strict­ly nec­es­sary but is good prac­tice, easy to do, and increas­ing­ly expect­ed. Again, it is worth empha­siz­ing that nei­ther the ben­e­fit nor the cost por­tions of this equa­tion need to be explic­it­ly stat­ed in finan­cial terms.

  • Know­ing bad ROI is more impor­tant than know­ing good ROI. In our expe­ri­ence, there is gen­er­al­ly enough uncer­tain­ty around the costs of dif­fer­ent strate­gies that minor dif­fer­ences in the rel­a­tive ROI of alter­na­tives should be inter­pret­ed with cau­tion, espe­cial­ly as cost-effec­tive­ness is not the only con­sid­er­a­tion in select­ing an action. In some ways the most rel­e­vant infor­ma­tion con­tained in an ROI analy­sis is at the bot­tom-end, that is, those options which appear to deliv­er a poor ROI. The the selec­tion of a strat­e­gy rel­a­tive­ly poor cost-effec­tive­ness would require care­ful­ly justification.
  • Link ROI to strat­e­gy and oppor­tu­ni­ty map­ping. ROI analy­sis can also be used to select strate­gies in a way that con­sid­ers both action and loca­tion simul­ta­ne­ous­ly, essen­tial­ly answer­ing the ques­tion, “what should we do where?” The strat­e­gy and oppor­tu­ni­ty map­ping described ear­li­er in this Guid­ance is a strong plat­form for ROI analysis.

Trade-off analy­sis

Trade-off analy­sis is a gen­er­al approach to eval­u­at­ing strate­gies when there are mul­ti­ple objec­tives. Trade-offs are nat­u­ral­ly pre­sent­ed as con­se­quen­tial rela­tion­ships between things we care about, and they exist when achieve­ment of one objec­tive comes at the expense of the achieve­ment of anoth­er objec­tive. For instance, when plan­ning the allo­ca­tion of a land­scape to dif­fer­ent activ­i­ties there is like­ly to be a trade-off between food pro­duc­tion (the amount of food able to be pro­duced) and bio­di­ver­si­ty con­ser­va­tion (the num­ber of species con­served). The more land we use to grow food, the few­er the species like­ly to sur­vive, and sim­i­lar­ly, the more land we ded­i­cate to con­ser­va­tion, the less food we can pro­duce from that land­scape (assum­ing the same lev­el of pro­duc­tiv­i­ty). By iden­ti­fy­ing dif­fer­ent com­bi­na­tions and extent of land-use place­ment, and plot­ting the expect­ed food pro­duc­tion and bio­di­ver­si­ty con­served for each, we are able to see the con­se­quence of improv­ing one of these objec­tives in terms of loss for the oth­er .In these cas­es the best solu­tion only be decid­ed through explor­ing actu­al trade-offs in pre­dict­ed out­comes. For instance, in the food ver­sus bio­di­ver­si­ty exam­ple above, the weight local stake­hold­ers give to bio­di­ver­si­ty objec­tives rel­a­tive to food pro­duc­tion objec­tives will depend on the con­se­quence of bio­di­ver­si­ty con­ser­va­tion for food pro­duc­tion: if the trade-off is small they might weight bio­di­ver­si­ty con­ser­va­tion strong­ly, but if con­ser­va­tion requires giv­ing up sub­stan­tial food pro­duc­tion, they are like­ly to weight bio­di­ver­si­ty much lower.

For prag­mat­ic rea­sons, trade-off analy­ses typ­i­cal­ly empha­size trade-off between two objec­tives. For­tu­nate­ly there are many con­ser­va­tion plan­ning prob­lems that while not two dimen­sion­al, can be use­ful­ly sum­marised in terms of two dimen­sions. All ana­lyt­i­cal tools ulti­mate­ly involve sim­pli­fi­ca­tion of high­ly com­plex social-eco­log­i­cal sys­tems and we believe that illus­trat­ing trade-offs can be a use­ful con­ser­va­tion plan­ning tool, espe­cial­ly for prob­lems involv­ing two major con­sid­er­a­tions, say food and car­bon, or car­bon and bio­di­ver­si­ty. Illus­trat­ing trade-offs between a cou­ple of dimen­sions can be a strong advo­ca­cy tool through pro­mot­ing con­struc­tive delib­er­a­tion about unavoid­able trade-offs, and pro­vide a pow­er­ful basis for strat­e­gy selection.

  • Mak­ing sure strate­gies are as effi­cient as pos­si­ble. One of the func­tions for which econ­o­mists use trade-off analy­sis is to check that strate­gies are as effi­cient as pos­si­ble, or in their lan­guage, Pare­to opti­mal. A strat­e­gy is Pare­to opti­mal when no improve­ment in one objec­tive can be made with­out simul­ta­ne­ous­ly dimin­ish­ing the achieve­ment of at least one oth­er objec­tive. Strate­gies that are not Pare­to opti­mal are inef­fi­cient because they involve unnec­es­sar­i­ly sac­ri­fic­ing achieve­ment of one or more objec­tive, where­as strate­gies along the effi­cien­cy fron­tier rep­re­sent trade-offs that can­not be avoided.

Effi­cien­cy fron­tiers are easy to con­struct in the­o­ry but often dif­fi­cult in prac­tice. The stan­dard approach is to take one objec­tive, and across its entire range of poten­tial val­ues, find the best you can do on anoth­er objec­tive. This is typ­i­cal­ly accom­plished through opti­miza­tion. Doing this opti­miza­tion can be a seri­ous com­pu­ta­tion­al task in its own right. A use­ful and more eas­i­ly accom­plished start­ing point is to sim­ply plot the con­se­quence for the two or three objec­tives of each of the alter­na­tives being con­sid­ered. Or alter­na­tive­ly devel­op a hypo­thet­i­cal set of alter­na­tives that favour one or the oth­er objec­tive to vary­ing degrees. For exam­ple, The Nature Con­ser­van­cy and the Uni­ver­si­ty of Ten­nessee col­lab­o­rat­ed to devel­op a tool that helps pro­pose a series of alter­na­tive infra­struc­ture pat­terns for shale gas devel­op­ment and then site these along an effi­cien­cy fron­tier of project cost and envi­ron­men­tal impact (Fig­ure 15).

Fig­ure 15: Envi­ron­men­tal Impact vs. Addi­tion­al Project Cost

  • Mov­ing from illus­tra­tion to strat­e­gy selec­tion Arguably a more dif­fi­cult task than illus­trat­ing trade-offs between strate­gies is actu­al­ly decid­ing on the best option. In the­o­ry (and if we were tru­ly pas­sive objec­tive observers, which of course we are not), all solu­tions locat­ed on the effi­cien­cy fron­tier are equal­ly good. In prac­tice this is where the influ­ence of oth­er objec­tives comes into play (either con­scious­ly or uncon­scious­ly). For exam­ple, in addi­tion to the land­scape scale trade-off between food pro­duc­tion and bio­di­ver­si­ty, a land allo­ca­tion deci­sion might be influ­enced by the eco­nom­ic inter­ests of par­tic­u­lar indi­vid­u­als. There are, how­ev­er, ways to look at trade-offs to iden­ti­fy points which might be desir­able. Trade-off curves often exhib­it points where the rate of decrease in one objec­tive increas­es rapid­ly as the oth­er objec­tive increase. These are referred to as points of inflec­tion (Fig­ure 16). Such points are not always obvi­ous and there is no guar­an­tee that these rep­re­sent a desir­able solu­tion. How­ev­er, they are often valu­able to iden­ti­fy because they rep­re­sent the point at which loss­es for each objec­tive are min­i­mized. With increas­ing dis­tance from an inflec­tion point, small improve­ments in one objec­tive rep­re­sent sig­nif­i­cant loss­es in anoth­er objec­tive, and deci­sion mak­ers have to increas­ing­ly favour one objec­tive over another.

Fig­ure 16: Bio­di­ver­si­ty vs. Food Production

  • Is it appro­pri­ate to frame some­thing as a trade-off? Just because a trade-off can be illus­trat­ed as part of an analy­sis does not mean these things can actu­al­ly be trad­ed off. For exam­ple, some stake­hold­ers might feel that things such as cul­tur­al her­itage or the right to make a liv­ing off the land can­not be trad­ed-off at all. Anoth­er way to think about this is that stake­hold­er’s assess­ment of con­se­quence for a par­tic­u­lar objec­tive may have a clear thresh­old rather than being a con­tin­u­ous func­tion, for exam­ple, cul­tur­al her­itage can either be pro­tect­ed or not, there is no par­tial deliv­ery of this objec­tive. Wilder­ness advo­cates often hold sim­i­lar atti­tudes. Great care must be tak­en to avoid trade-off analy­sis that appears cal­lous because in many cas­es it is not some gen­er­al­ized com­mod­i­ty but actu­al­ly peo­ple who will be affect­ed by the trade-off. Sim­i­lar­ly, from a polit­i­cal and com­mu­ni­ca­tion point of view, for­mal trade-off analy­sis can present a chal­lenge because it involves acknowl­edg­ing the pos­si­bil­i­ty that an action or pol­i­cy may have neg­a­tive con­se­quences for val­ues that peo­ple care about. It is not appro­pri­ate to frame some­thing in this man­ner when the trade-off stands to neg­a­tive­ly impact a vul­ner­a­ble population.

Sce­nario analysis

Imag­ine that an effec­tive glob­al car­bon mar­ket is in place and all nations have agreed to strict emis­sions caps that keep the price of CO2 off­sets at $20 a ton or above. This guar­an­tees a vast and sus­tain­able flow of rev­enue to land­hold­ers, com­mu­ni­ties and gov­ern­ments who pro­tect forests and peat bogs, vir­tu­al­ly elim­i­nat­ing defor­esta­tion glob­al­ly. This is a sce­nario, albeit an opti­mistic one.  Sce­nar­ios such as this are not intend­ed to be pre­dic­tions, nor are they a choice that some­one in charge of a con­ser­va­tion project could make, but rather they are learn­ing tools. What would it mean for our poten­tial strate­gies if the sce­nario above became real­i­ty? Analysing sce­nar­ios can help us under­stand how dif­fer­ent strate­gies might fare in the inevitably uncer­tain future. Sce­nario analy­sis can there­fore be a use­ful strat­e­gy selec­tion tool where uncer­tain­ty is high.

All strate­gies have uncer­tain­ty asso­ci­at­ed with their out­comes because we can­not per­fect­ly pre­dict the future; sce­nar­ios describe some of this uncer­tain­ty through inten­tion­al manip­u­la­tions of imag­ined futures. The basic premise of sce­nario analy­sis is to devel­op a small set of pos­si­ble future sce­nar­ios that describe how some of the main uncer­tain­ties — such as demo­graph­ic trends, poli­cies, mar­kets, bud­gets, degree of  cli­mate change or stake­hold­er sup­port — might behave. Some of the most glob­al­ly recog­nis­able sce­nar­ios are the dif­fer­ent emis­sions sce­nar­ios devel­oped by the Inter-gov­ern­men­tal Pan­el on Cli­mate Change (IPCC) to explore uncer­tain­ties in nation­al poli­cies and social behav­iour and the impact on cli­mate change. The set of strate­gies being con­sid­ered are then imag­ined to be occur­ring on the back-drop of these dif­fer­ent sce­nar­ios with the aim of learn­ing about how they would be expect­ed to per­form. For exam­ple, we might assess the per­for­mance of coastal adap­ta­tion strate­gies under each of the dif­fer­ent IPCC sce­nar­ios or using sea lev­el rise mod­els with dif­fer­ent assump­tions. Some­times sce­nar­ios might yield para­me­ters that can be used in a for­mal pre­dic­tive mod­el (like the extent of sea lev­el rise expect­ed under each of the IPCC sce­nar­ios), but sce­nario analy­sis can equal­ly be accom­plished sim­ply by describ­ing how we think a strat­e­gy might per­form under a set of future con­di­tions. The aim of sce­nario analy­sis is to see how robust a strat­e­gy is to a range of pos­si­ble futures, and there­fore how much con­fi­dence we should have select­ing it. This infor­ma­tion can also be used to refine strate­gies to make them more robust.

As a straight­for­ward exam­ple of eval­u­at­ing strate­gies for con­ser­va­tion of urban ecosys­tems in and around Stock­holm, Ulla Mort­berg and col­leagues (2012) explored the con­se­quence of two pos­si­ble urban growth sce­nar­ios. The first sce­nario they termed “Com­pact” in which urban growth poli­cies empha­sized ener­gy effi­cien­cy and min­i­miz­ing trans­port dis­tances and costs. The sec­ond sce­nario they termed “Urban Nature” in which urban growth poli­cies empha­sized pro­tec­tion of urban green spaces and planned but dis­trib­uted devel­op­ment. Mort­berg and col­leagues spec­u­lat­ed that the first sce­nario would like­ly result in the loss of urban bio­di­ver­si­ty and ecosys­tem val­ues, which would sub­se­quent­ly dimin­ish the engage­ment and sup­port of urban com­mu­ni­ties with con­ser­va­tion. Some of the con­se­quences may be off­set, how­ev­er, by reduced loss of bio­di­ver­si­ty in the peri-urban area because of a reduc­tion in urban sprawl. The sec­ond sce­nario, they sug­gest­ed, main­tained urban bio­di­ver­si­ty and ecosys­tems, but required greater ongo­ing bud­gets for trans­port infra­struc­ture and ener­gy avail­abil­i­ty because of a growth in low den­si­ty hous­ing, some of which was like­ly to come at the expense of peri-urban areas. Even using these two sce­nar­ios high­light­ed the con­se­quence for urban bio­di­ver­si­ty projects under dif­fer­ent pol­i­cy envi­ron­ments, and made obvi­ous the need to seek greater clar­i­ty around the bio­di­ver­si­ty and ecosys­tem ser­vice val­ue of urban ecosystems.

Once the per­for­mance of strate­gies has been eval­u­at­ed under dif­fer­ent sce­nar­ios, this infor­ma­tion needs to be analysed in order to help select the best strat­e­gy. One of the most use­ful approach­es is to com­bine sce­nario analy­sis with mul­ti­cri­te­ria deci­sions analysis.

  • When is sce­nario analy­sis appro­pri­ate? Sce­nar­ios are best suit­ed to explor­ing sit­u­a­tions where uncer­tain­ty is high and con­trol­la­bil­i­ty is low. For exam­ple, cli­mate change and glob­al gov­er­nance are large­ly beyond the con­trol of con­ser­va­tion deci­sion mak­ers, even in a large region. In these sit­u­a­tions, sce­nar­ios can help to illu­mi­nate the con­se­quences of these glob­al dri­vers of change and to for­mu­late robust local responses.
  • How do you design sce­nar­ios? A gen­er­al­ized approach to sce­nario devel­op­ment would include the fol­low­ing process:
  • Iden­ti­fy the three to five most impor­tant dri­vers of change,
  • For each of these dri­vers, iden­ti­fy pos­si­ble future trends with a small num­ber of cat­e­gories (for exam­ple, remain the same, small increase, big increase), or bifur­cat­ing deci­sions (for exam­ple, a pol­i­cy is imple­ment­ed or not),
  • Cre­ate a frame­work by group­ing these trends along two to three axes of uncertainty,
  • Devel­op a set of coher­ent sto­ry­lines (a nar­ra­tive about what may hap­pen in the future) that draw on the pos­si­ble trends and cov­er as much of the space in the frame­work as possible.

There is a good deal of debate in the sce­nario analy­sis com­mu­ni­ty about how ‘plau­si­ble’ each sce­nario and its sto­ry­line need to be.  Part of the point of sce­nario analy­sis is explic­it­ly not to focus on what is like­ly; extreme, low-prob­a­bil­i­ty sce­nar­ios can still be very use­ful in sce­nario analy­sis. It is crit­i­cal, how­ev­er, that sce­nar­ios are coher­ent; a sce­nario must encap­su­late a coher­ent sto­ry about the future world. For instance, a sce­nario in which sub­ur­ban areas, farm­land and intact habi­tat all increased is unlike­ly to be a coher­ent sto­ry as devel­op­ment of sub­ur­ban areas will gen­er­al­ly occur at the expense of farm­land or intact habitat.


Mört­berg, U., Haas, J., Zetter­berg, A., Franklin, J. P., Jon­s­son, D., and B. Deal. 20012. Urban ecosys­tems and sus­tain­able urban development–analysing and assess­ing inter­act­ing sys­tems in the Stock­holm region. Urban Ecosys­tems, 2012:1–20.