Linking cost efficiency evaluation with population viability analysis to prioritize wetland bird conservation actions Esther Sebastián-González a,b,⇑, José Antonio Sánchez-Zapata a, Francisco Botella a, Jordi Figuerola b, Fernando Hiraldo b, Brendan A. Wintle c a Ecology Area, Department of Applied Biology, Miguel Hernández University, Ctra. Beniel Km 3.2, E-03312 Orihuela, Alicante, Spain Doñana Biological Station – CSIC, Américo Vespucio s/n, E-41092 Sevilla, Spain c School of Botany, University of Melbourne, E-3010 Victoria, Australia b Keywords: Conservation planning Efficiency Prioritization Risk assessment Threat status Uncertainty Cost-benefit a b s t r a c t Prioritizing management actions for wildlife conservation is a difficult task due to the large number of problems relative to available conservation resources and uncertainty about the benefits arising from numerous potential management actions. In this study we use a cost-efficiency protocol to evaluate and prioritize eight different management actions for waterbird community in wetlands throughout southeastern Spain. The protocol generated an action priority ranking based on the costs and predicted benefits of the actions in terms of waterbird carrying capacity. Action prioritization outcomes were also evaluated using population viability analysis models for two of the study species. Removal of dead bird carcasses to prevent disease outbreaks was identified as the most cost efficient action. Removing lead pellets from the sediment was the least efficient strategy. Our approach highlights the role of detailed risk assessment as a form of quality control on the simpler prioritization protocols. We recommend a twostep prioritization protocol based on (i) a rapid, usually simpler prioritization approach for the bulk of species or values being managed, and (ii) a more sophisticated risk assessment for a subset of the species of interest for which detailed risk assessments are tractable. This process strikes a balance between sophistication and practicality. . 1. Introduction Conserving and managing sites to protect species demands evaluation judgments because such actions often involve a choice among a set of alternatives. Competing management options might be prioritized in order to maximize species persistence. The budget for conservation is generally small relative to the number and magnitude of conservation problems and it is seldom possible to deal with all the identifiable problems. Therefore, managers have to decide how their budget can be spent most effectively (Naidoo et al., 2006; Wilson et al., 2006; Polasky, 2008). An increasing number of researchers attempt to quantify the possible gains in conservation efficiency by including information about economic costs of locations in reserve selection (e.g., Polasky et al., 2001; Williams et al., 2003; Juutinen et al., 2004; Strange ⇑ Corresponding author at: Ecology Area, Department of Applied Biology, Miguel Hernández University, Avda. Universidad s/n, E-03202 Elche, Alicante, Spain. Tel.: +34 965 22 21 23. E-mail addresses: esebastian@umh.es (E. Sebastián-González), toni@umh.es (J.A. Sánchez-Zapata), paco.botella@umh.es (F. Botella), jordi@ebd.csic.es (J. Figuerola), hiraldo@ebd.csic.es (F. Hiraldo), b.wintle@unimelb.edu.au (B.A. Wintle). et al., 2006). These studies emphasize the value of incorporating biodiversity benefits such as increased habitat area or increased species abundance, and costs in an integrated approach with the aim of giving priority to the most cost-efficient options and, therefore maximizing net expected benefits of conservation investments. Cost-efficiency analysis are based on different approaches such as the design of reserve networks (Jantke and Schneider, 2010), the prioritization of species for conservation (Marsh et al., 2007) or the selection of management actions (Rodonini and Boitani, 2007). However, the bulk of the literature around systematic conservation planning focuses on the situation involving binary investment or zoning choices (e.g. reserve/non-reserve, restore/not-restore). Simple and workable strategies for maximizing the net benefits (in biological terms) obtained from multiple competing management actions across multiple environmental values (e.g. species, vegetation types, wilderness values) are in the early stages of development (e.g. Polasky et al., 2008; Joseph et al., 2009). Both Polasky et al. (2008) and Joseph et al. (2009) provide a compelling case for utilizing cost-efficiency in conservation planning. However, in the case of Polasky et al. (2008) the economic modeling approach utilized is challenging to implement without modeling skills that are not necessarily accessible to all practitioners (e.g. Polasky et al., 2008). In the case of Joseph et al. (2009) the subjective estimates of biodiversity benefit (expert opinion about extinction probabilities) are not transparent, and if applied in most circumstances would not make the most of available data and ecological modeling tools. Here we attempt to strike a balance between ecological realism and practicality, by presenting a cost-efficiency action prioritization case study that exploits a simple cost-efficiency definition combined with available ecological data and models. Our approach utilizes available time-series data on species responses to habitat change and seasonal variation to underpin estimates of the biodiversity benefits arising from proposed management actions. Biodiversity benefit in this case study is measured in terms of increases to focal population abundances resulting from proposed management actions. We illustrate our approach by prioritizing non-spatial habitat restoration options for the extensive, speciesrich complex of wetlands in south-eastern Spain. Our approach is twofold. First we implement a cost-efficiency analysis to identify the habitat restoration and management investments that bring the greatest expected net benefit (in terms of increased abundance of 25 waterbird species) for our fixed budget. We then attempt to evaluate our prioritization strategy by comparing the prioritized ranking of actions determined by the cost-efficiency analysis across all species with a more detailed analysis of the consequences of management options based on population viability analysis (sensu Akçakaya, 2000; Wintle et al., 2005a; Bekessy et al., 2009) for a subset of well-studied species. Combining population viability analysis (based on metapopulation modeling) with cost-efficiency prioritization approaches is novel in conservation planning (Newbold and Siikamaki, 2009; Wakamiya and Roy, 2009). Our protocol can be seen broadly as a prioritization based on cost-efficiency analysis across the full range of species of interest, followed by a quality control step that evaluates the likely outcomes for a subset of the species using population viability analysis. While the approach does not guarantee optimality in an economic sense, we believe that it is an appropriate trade-off between the practicality of simple cost-efficiency analysis (sensu Joseph et al., 2009) and the elegance and rigor of population risk assessment that cannot be practically applied to all of the species of interest. Our specific aims were to (i) describe a management action prioritization approach in which the biodiversity benefits of candidate actions are measured using commonly available ecological data and models, (ii) illustrate how population viability analysis modeling can be integrated with a cost-efficiency prioritization scheme, and (iii) produce a cost-efficiency prioritized ranking of possible actions for managing threatened wetland bird populations in Spain. 2. Methods 2.1. Case-study: waterbirds in a wetland network The study was conducted in the Vega Baja Valley in Southeast Spain (Fig. 1). It covers an area of 95,840 ha with a landscape dominated by citrus fruit trees, vegetables, palm trees and housing developments. In the 1980s, an inter-river water transfer was built to bring water for irrigation purposes. Since then, more than 2600 ponds have been constructed to store the water received, and the area has turned into a mosaic of crop fields and artificial wetlands (Sánchez-Zapata et al., 2005). Apart from the ponds there is also an important network of natural and semi-natural wetlands. Some of them enjoy regional environmental protection, as well as the international status of SPAs and RAMSAR sites. This complex includes Salines (Torrevieja-La Mata, Santa Pola, San Pedro del Pinatar), large water reservoirs (El Hondo, La Pedrera) and temporary ponds (Clot de Glvany). The wetlands complex holds one of the most important populations of breeding waterbirds in southern Europe, including some globally endangered species such as marbled teal (Marmaronetta angustirostris), whiteheaded duck (Oxyura leucocephala) or audouin’s gull (Larus audouini). During the twentieth century several factors, such as the creation of new irrigated lands and the urbanization of traditionally irrigated lands, have changed the use of natural resources, giving rise to a growing imbalance between water resources and irrigation demands and a loss of fertile soil and other environmental and cultural values of this traditionally agricultural area (Martínez-Fernández et al., 2000). Actually fresh water and land management is subject to many different public and private interests including biodiversity conservation, public use, agriculture, hunting, fishing, industrial salt production and housing. Furthermore, climate is semiarid and water resources are heavily exploited including local resources (basin and ground waters), water transferred from neighbor basins (Tajo-Segura rivers) and desalination, resulting into important socio-political conflicts. This background reinforces the need for providing sound scientific-based proposals for conservation under a cost-efficiency approach. 2.2. Cost-efficiency analysis Our stated objective was to achieve the maximum increase in the abundance of 25 wetland bird species across the region within the available budget. In order to achieve this, we ranked management actions according to cost-efficiency (greatest gain in abundance per unit cost (€)) and then chose as many of the most cost efficient set of actions as could be purchased within the available budget. Some actions aim to decrease mortality (e.g. removing pellets from a natural park), while other actions seek to increase carrying capacity (e.g. installing floating devices). To evaluate the impact of these measures, we selected both common and endangered species which were likely to be affected by the management actions. Candidate management actions were determined by eight experts. The selection of the management options depended on the availability of information about costs and possible benefits to the waterbirds. Eight candidate strategies were selected. We included both proposed and existing actions in the analysis. We incorporated some actions that have already been performed because they could be used in other similar wetlands or again in the same wetland, so it is important to evaluate their effectiveness. We considered all the management actions as additive. Actions that affected the carrying capacity included: (i) Installation of floating devices in the irrigation ponds to provide waterbirds with resting and breeding sites. The number of devices per pond depended on the size of the pond. We evaluated the installation of devices with vegetation in ponds without natural vegetation and vice verse. We estimated the benefit on the basis of the installation of ponds in 30% of the total number of irrigation ponds at the area. (ii) and (iii) Change in the construction design of the irrigation ponds. We have distinguished two types of ponds: LDP (constructed with Low Density Polyethylene) and HDP ponds (constructed with High Density Polyethylene). LDP ponds are lined by a layer of gravel to protect the plastic from solar radiation that can damage them, and this cover provides the pond with a more natural appearance. HDP materials allow the ponds to have higher slopes and to store more water. For these reasons, LDP ponds hold more abundant and richer waterbirds, vegetation and macroinvertebrates communities than HDP ponds (Sánchez-Zapata et al., 2005; Abellán et al., 2006; Sebastián-González et al., 2010). We have evaluated the conversion of 30% (ii) or 60% (iii) of the HDP to LDP ponds. Fig. 1. Study area. The figure represents the natural and artificial wetlands at the study area and their location. (iv) Construction of a permanent pond at the Clot de Galvany. This wetland often dries completely in summer and the construction of these temporary ponds has allowed some species to stay all year around at the wetland. It consists in a small wetland filled with water coming from a nearby sewage system. This action has already been carried out. (v) Installation of floating devices at the Salinas. Waterbirds breeding at the Salinas often suffer from a high chick predation. Thus, the installation of this type of devices offers predator-free breeding sites for some species. This type of devices has already been installed in the Salinas de San Pedro. (vi) Installation of artificial islands at the Salinas. Another strategy for avoiding high predation rates at the salinas is the installation of artificial islands constructed with stones and soil. These islands have been installed at the Salinas de La Mata-Torrevieja. Actions that influence survival include: (vii) Cleaning the Hondo and the Salinas de Santa Pola from lead pellets to avoid lead poisoning. (viii) Develop a prevention protocol to reduce biological contamination outbreaks at El Hondo (being botulism the most important one). The protocol consists in a weekly patrol to collect dead animals that are one of the main causes of the amplification of outbreaks. The cost of each action was calculated in Euros (€) and estimated for a 10-year period. The costs were based on the accomplished actions and estimations derived from small actions. The price for the floating devices at the ponds and salinas was calculated for the construction and the installation of the devices. The price of the devices at the Salinas was the real cost of the action performed in 2008 at the Salinas de San Pedro (Concejalia de Medio Ambiente de la Región de Murcia, pers. com.), while the price for the devices at the ponds was estimated using the cost of five experimental devices constructed by a NGO in 2007 (Sánchez, pers.com.). The cost was calculated for the renewal of the devices every 5 years, which was the average life of the experimental devices. For the calculation of the price of the transformation of the ponds construction design we took in account that the edges of LDP ponds slope more gently and, therefore require a larger surface area to store the same amount of water. We included both the price of the extra water loss by evaporation and the money the owner does not earn because they cultivate less area. LDP ponds are cheaper to construct but have to be renewed with a higher frequency. We considered both the higher price of HDP ponds and the higher renewal frequency of LDP ponds in the calculation of the costs. Moreover, as both the artificial permanent pond at the Clot and the artificial islands at the Salinas are actions already accomplished and the price is the total cost of the action (Aranda, pers. com.). For calculating the price of the lead cleaning we used the price of two executed cleaning actions at the study area. We used the same cleaning system used in the executed actions. As the cost increases with the amount of pellets present, we stratified the costs of cleaning in all the affected area using previous information about pellet concentration in the affected wetlands (Terrones Contreras, 2006). Finally, the costs of the protocol to control for biological outbreaks include the costs of the work force (two people, 5 months a year, 20 h a week) using stipulated salaries at the sector. We first estimated the population size of each species in every wetland. As we did not have information about the total population size at the irrigation ponds we constructed species-specific Generalized Linear Models (GLMs) in R (R Development Core Team, http://www.r-project.org) to estimate the abundance. We followed the steps given in Wintle et al. (2005b) to create individual specieshabitat models. We used information about waterbird surveys in 7 consecutive years (2003–2009) and in around 200 ponds (Sebastián-González et al., 2010). We included information about the size and location of the pond, the presence/absence of different types of vegetation and the construction design. We finally estimated the total population for each of the species in the 2600 ponds existing at the study area (see Appendix A, Supplementary material for a more detailed explanation and Appendix B for the results of the model). To estimate the population size at species level for the rest of the wetlands we used an average of the population given by the official surveys during the same years. To calculate the benefits of the installation of floating devices at the ponds and the lead cleaning we used extrapolations of previous experiences at small scale. We used the predictions given by the habitat models to estimate the benefit of the change in the construction design in all the ponds at the study area. The term benefit is here defined as the increase in expected species abundance (as predicted by the statistical model). For the construction of a permanent pond at the Clot and the artificial island at the salinas we used real data from previous experiences, while for evaluating the effect of the biological outbreaks we used the average mortality rates in the last three biological contamination outbreaks at the Hondo to estimate the number of pairs that would survive if there were not biological contamination outbreaks episodes. In the case of contamination outbreaks, we assume that the mortality associated with individual outbreaks is additive. We calculated the benefit (increased abundance relative to doing nothing) after 10 years in order to make outcomes comparable, despite the fact that some benefits derive from increasing carrying capacity and some focus on reducing mortality. We calculated for each action i, the total benefit (Bi) as the sum of the number of pairs in 10 years over the execution of the action for all the species, X Bi ¼ bij ; where bij is the benefit of the action i in terms of the expected abundance of species j. For each action j and species i we included a correction factor that represents the likelihood of success (Lij) that ranged between 1 (actions with the maximum probability of success) and 0 (actions without any probability of success). This factor accounted for the variability in the data and for the feasibility of the action, and can be viewed as a measure of uncertainty (see Appendix C in Supplementary material for the list of the correction factors used). The benefit of each action on each species was multiplied by this factor to provide an expected benefit accounting for the probabilistic uncertainty about whether or not actions will be successful. Therefore, the total expected benefit arising from each action was estimated as Bi ¼ X ðbij x Lij Þ: The value of the correction factor varied depending on the experts’ confidence on the probability that each action could be implemented successfully. This factor was high for already accomplished actions because the benefit is based on real data about the benefit for the species. It was also high when we could calculate a confidence interval and use the bounds of the interval for the evaluation of alternative scenarios. When the data availability was low, the factor was also low because the risk of failure is higher. We ranked the actions on the basis of a cost-efficiency criterion. We calculated the efficiency E of each action j as Ej ¼ C j =Bj where Cj is the cost of the action over 10 years. The actions were ranked from the lowest to the highest Ej. We evaluated alternative scenarios for management using two approaches. When available from the data, we used the upper (Lmax) and lower (Lmin) limits of the estimates. We used this approach for the action considering changes in the construction design of the ponds, and for the control of biological outbreaks. The second approach used the likelihood of success for each project. We estimated the maximum benefit by considering the correction factor to be maximal (Lij = 1). For the minimum benefit we reduced the factor to the half. This approach was used for the rest of the actions evaluated. In order to reflect the relatively high social importance of highly threatened species compared with relatively common species, we weighted each species by its threat status (Madroño et al., 2004). Multiplicative weighting applied in the analyses were: (1) if the species is cataloged as non-threatened, 1.5 if it is cataloged as almost endangered, (2) if it is vulnerable, (3) for endangered species and (4) for critically endangered ones. 2.3. Population viability analysis We modeled the population viability of a subset of the species under each of the management actions, to provide a separate line of evidence about the relative performance of management options and to ensure that certain options were not highly detrimental to the viability of the species that could plausibly be modeled in this way. Due to the availability of data and expertise, we were able to model the 100-year viability of the black-winged stilt (Himantopus himantopus) and the little grebe (Tachybaptus ruficollis) in southeastern Spain using a spatially-explicit, stage-structured, stochastic model based on habitat suitability maps and on demographic data (see Appendix C in Supplementary material for a further explanation of the model). The demographic component of the model included density dependence, stage and sex structure, demographic and environmental stochasticity. The population model was used to estimate the expected minimum population size (EMP: McCarthy and Thompson, 2001) for the two modeled species over the 100 years of the simulation under each investment option. EMP was used as the basis of comparison between actions (sensu Wintle et al., 2005a) and to evaluate the congruence between the population viability analysis and the simpler evaluation approach based on the correlative habitat models. We used the population viability analysis program RAMAS/GIS, which is designed to link landscape data from a geographic information system with a metapopulation model (Akçakaya, 1995; Akçakaya et al., 1995). 3. Results 3.1. Action ranking Our protocol provided information at species level about the effect of the implementation of each management action. The cost of the actions was very variable, but the range of potential effects was also high (Table 1). Some actions produced small increases in the populations and affected few species (e.g. 15 new pairs and one species for the construction of floating islands) while others had widespread effects (e.g. 846 pairs and 20 species for the installation of floating devices at the ponds). In general, the control of biological contamination outbreaks was the measure that affected a higher number of individuals and species. The cleaning of the El Hondo Park from lead pellets and the two actions based on the irrigation ponds also had high benefits in terms of net increase in the number of pairs after 10 years. The ranked order of the cost-efficiency evaluation changed depending on the approach used (Table 2). For most of the analyzed scenarios, removing pellets from the Hondo was the least efficient strategy. Investment in the irrigation ponds also seemed to be relatively inefficient. The control of biological contamination outbreaks was the most efficient action, while the two management actions affecting the Salinas were also relatively efficient. The ranked order of the actions varied when using Lmin. Neither the rank obtained using Lmax nor the one obtained when weighting the results by threat status of the involved species, changed the overall ranking of actions. 3.2. Population viability analysis The expected minimum population size (EMP) for the two study species varied depending on the management strategy used (Table 3). All the strategies resulted in an increase in EMP. The scenario with the highest EMP was produced by the control of biological contamination outbreaks for both species. As the population at the irrigation ponds is large, the two actions affecting these artificial wetlands also produced an important increase in the population of the species. The sensitivity analysis performed showed that the model for the black-winged stilt was affected by the inclusion of demographic stochasticity, while the model for the grebe suffered from the strongest variations when removing the correlation of the stochastic events in the vital parameters (see Appendix D in Supplementary material). Changes in the rest of the parameters did not produce large variations in the EMP of the model, considering a variation large when predicted population size is double. There were also no changes in the ranking of Table 2 Cost-efficiency (E) calculated as the cost of the action divided by the expected net benefit for each of the management actions (in terms of the change in number of breeding pairs compared with the status quo for all species), under four criteria: average estimated benefit, maximum benefit (L max), minimum benefit (Lmin) and weighted by species’ threatened degree (T.D.). Final ranking of the action for the given prioritization strategy is shown in brackets. The negative sign appears because the variability in the population of some species at the ponds is high and the lowest confidence interval shows a reduction in the population. Management action Average 30% floating devices 30% ponds LDP 60% ponds LDP Pond clot Floating islands Artificial island Lead cleaning Biological outbreaks 3311 (5) 12,833 (7) 9829 (6) 1556 (4) 592 (3) 351 (2) 22,831 (8) 77 (1) L max 845 (5) 4534 (6) 5128 (7) 622 (4) 300 (3) 255 (2) 13,334 (8) 43 (1) Lmin T.D. 6622 (5) —15,453 (8) 118,051 (7) 3111 (3) 592 (2) 703 (3) 45,662 (6) 338 (1) 1933 (5) 12,691 (7) 9712 (6) 1521 (4) 592 (3) 207 (2) 19,341 (8) 23 (1) management actions between the PVA and the full prioritization protocol (Table 3). 4. Discussion 4.1. Prioritization protocol The analysis of management actions’ effectiveness can be an important tool in planning conservation strategies. However, a rigorous evaluation of the outcomes of the cost-efficiency analysis is necessary to ensure that the result is not simply an artifact of particular settings and assumptions made during the analysis. Several factors affected the final prioritized rank order obtained using our approach. Variation in the estimated likelihood of success and the weights assigned to species influenced the rank of the actions, making our results similar to previous studies (Joseph et al., 2009; van Teeffelen et al., 2008). This highlights the importance of well defined management objectives and careful elicitation of social and political preferences (e.g. how much more important it is to people to protect threatened species rather than just increase overall waterbird abundance; Wilherea et al., 2008; Wikberg et al., 2009). Action prioritization is likely to be sensitive to error in the estimation of action costs (Bryan, 2010). The cost of some of the actions was estimated with previous pilot studies and the final value depends on the accuracy of the estimations. For example, the budget for cleaning the Hondo may vary if a different technique is used for cleaning. We also did not include the costs of maintaining the actions in the prioritization protocol because this information is difficult to obtain. These costs should be considered to ensure the success of the management in time. Furthermore, all the costs and benefits were estimated for a 10-years study period to allow tractable estimation of costs. Nonetheless, the rank for Table 1 Costs of each of the management actions over 10 years and the expected number of pairs arising from that action after 10 years (summed across all species). Values for the number of pairs, species and threatened species are those estimated to be affected by the proposed management actions. Maintenance cost reflects the economic cost of maintaining the action over 10 years after implementation. Number of pairs was calculated as a consequence of a higher carrying capacity or as a reduction in the mortality for all the studied species. The baseline scenario represents the total number of pairs, species and threatened species currently present at the study area (without taking any action). Management action Baseline scenario 30% floating devices 30% ponds LDP 60% ponds LDP Pond clot Floating islands Artificial island Lead cleaning Biological outbreaks Cost (€) 2.800.000 3.500.000 7.000.000 44.892 8.999 66.150 20.000.000 100.000 Maintenance cost Number of pairs Number of species Number of threatened species Low None None Medium Low Low None Included cost 4397 845 273 712 29 15 188 876 1.303 25 20 11 11 16 1 3 12 17 10 7 3 3 1 0 1 3 6 Table 3 Management action prioritization rank for the black-winged stilt and the little grebe under the population viability analysis (Rank 1) and under our protocol (Rank 2). The expected minimum population (EMP) size is given for 100 years from now, as is the estimated benefit in terms of expected increase in abundance above the ‘do nothing scenario’ predicted by the statistical models used in the prioritization for all species (‘benefit’). Note that the rank is only based on expected benefits and that cost information has not been included for the comparisons. The installation of floating and artificial islands could not be evaluated using PVAs because none of the modeled species used these facilities. Little grebe Black-winged stilt Management EMPa Rank 1 Benefit Rank 2 EMPa Rank 1 Benefit Rank 2 30% floating devices 30% ponds LDP 60% ponds LDP Pond clot Lead cleaning Biological outbreaks 861 859 870 865 861 881 4 6 2 3 4 1 91.6 44.3 100.5 3.1 64.5 141.3 3 5 2 6 4 1 2012 2024 2028 1999 2006 2053 4 3 2 6 5 1 41.3 97.7 220.6 2.5 143.8 756.4 5 4 2 6 3 1 some of the actions may change using a different time period. For example, a park free of lead would continue to increase the survival rate of the species using the wetland until the lead pellets become unavailable when the sediment covers them. In general, there are so many factors affecting the final cost of an action that managers often need to generalize to make the calculations feasible. Even if in this study we did not evaluate the effect of changes in the cost of the actions, costs are crucial in determining the final ranking of the actions and more complex methods are available for this part of the evaluation (Naidoo et al., 2006). Moreover, in such a framework it could happen that the budget available for conservation is bigger or smaller than the cost of the prioritized management action. When the budget is small, then other projects with smaller costs could be executed, following the prioritized order. If the budget is high, we can perform several actions, as the effect of each action has been estimated as additive with this method. Besides, the possible complementary effect of the projects has not been evaluated with this framework but should be considered for the action selection (i.e. Leathwick et al., 2010). The benefits obtained from the management actions might be synergic. Therefore, as we are assuming that the actions are additive, we are probably underestimating the benefits of the projects. Moreover, some projects, such as the installation of floating devices or the change in the construction design of the ponds may be expanded or diminished depending on the budget if necessary. Other factor affecting the benefit that is difficult to evaluate but that should be considered is the suitability of an action for a specific habitat and place. As it happened with the islands at the Salinas, that were a response to previous predation problems. Our protocol also assumes that all the species are substitutable. It evaluates actions by means of the benefit given by any species. Again, including a parameter concerning species complementarity would deal with this problem, but it is beyond the scope of this study. Managers interested in cost-efficiency evaluations should also be aware of the importance of the quality of the data used in the assessment (Cabeza and Moilanen, 2001; Hegland et al., 2010). In many cases information about the effectiveness of management actions is not very well understood. The inclusion of likelihood of success (L) goes part way to catering for this form of uncertainty. However, the quality of decisions made using cost-efficiency analysis still rests squarely on the quality of the ecological and economic data and knowledge underpinning them. If data are sparse and knowledge is poor about a particular species, ecosystem or management strategy, then the use of cost-efficiency analysis should be couched in a decision theory that deals with uncertainty. At the very least, analysts should explore the consequences of misinformation using sensitivity analysis or robustness analysis (Ben-Haim, 2006). Our methods here do not reduce the need for, or importance of good ecological (Arponen et al., 2010) and economic information. They do, however, provide a means to ensure logical consistency in decisions made on the basis of cost-efficiency, given the available information. Our analysis also provides insights into the relative efficiency of investing in various habitat types. In general, investing in rich areas with big populations may produce higher benefits and there is also a clear bias in our results for the actions that affect the richest wetlands. For example, the most efficient action was the control of biological contamination outbreaks, predominantly in the Hondo, the most important wetland for waterbirds at the area (Martí and del Moral, 2003, 2004). Moreover, in most of the scenarios, the less efficient actions were those proposed for the irrigation ponds, which are less suitable habitat for a variety of reasons, including low protection from predation. The secondary evaluation of the actions using PVAs reinforced the feasibility of the protocol in this case. PVA provides managers with predictions about the actual variable of management interest; the probability that species will persist in the study region over a specified period, or in the case of EMP, the lowest abundance expected for the species of interest over the simulation period (Akçakaya, 2000; Wintle et al., 2005a; Bekessy et al., 2009). In contrast, prioritization made on the basis of predictions about carrying capacity alone, rely on a set of implicit assumptions about the relationship between carrying capacity and the long-term viability of the species of interest. Therefore, PVAs are more ‘proximal’ to the management objective of interest and management evaluations based on this approach may be more likely to achieve their objectives. Nonetheless, the realism of PVA hinges on the availability of relevant biological information which may be difficult to obtain. There is a need for long-term studies that can provide reliable information about the survival rates of the species so that PVA can be used more widely and reliably in conservation action prioritization. Moreover, PVA necessarily operates on species information. If we need to evaluate actions affecting communities, we would have to perform PVAs for each of the species at the assemblage and understand something about the interactions between species within a community, which may not be possible in most cases. Unfortunately, community and ecosystem models are not yet applicable in most conservation planning and prioritization frameworks as they tend not to be quantitative. 4.2. Management implications The waterbird community in the south-east of Spain benefited differentially under the range of proposed management investments: Some actions affected a high number of species while others focused on only a few of them. The cost-efficiency protocol provides clarity around the selection of management actions by providing a ranking that directly reflects the aim of managers; to maximize water bird abundance within the available budget. If the aim was slightly different, for example, to maximize the number of species maintained above a minimum population size, then the analysis would be slightly different (and more complex). In our analysis, the most efficient action was the control of biological contamination outbreaks and therefore this action should be prominent in the minds of regional managers. The main outbreak in the system is botulism, even though other agents like toxic algae blooms may produce massive mortalities among waterbirds, further favoring botulism outbreaks (Lopez-Rodas et al., 2008). However, removing corpses of dead animals would reduce the spread of the bacteria or the infection of animals preying on fly larvae with high toxin loads and reduce the probability of any biological outbreak (Rocke and Friend, 1999); even though a better water quality would also be important. The constructions of artificial and floating islands in the Salinas were also ranked in a good position. These actions were partially successful because they were carried out after detecting high chick mortality rates for some of the species already breeding in the wetlands. Both actions were very efficient and can be a good example to be followed in similar situations. Similarly, the success of the artificial pond construction at the Clot de Galvany was related to the water availability problems at the study area as a result of overexploitation. The objective of this action was not to increase bird numbers ‘‘per se’’ by creating new artificial conditions or areas, but to restore the former environmental conditions destroyed by human activities. The less efficient actions where those affecting irrigation ponds. The ponds are used as breeding, foraging, resting and wintering sites by some waterbird species (Sánchez-Zapata et al., 2005; Sebastián-González et al., 2010). Most of the species are widespread at the study area, but some of them are of conservation concern (Madroño et al., 2004). Besides, the analyzed actions affecting these artificial wetlands had in general high economic costs. Nevertheless, for the action affecting the construction design of the ponds, the estimation of the costs is not easy because LDP ponds are not constructed anymore. Even if LDP ponds are not constructed, their design could be easily copied and used, therefore, the best proposal would be to motivate from the regional government the construction of ponds with softer slopes that benefit the establishment of the vegetation and their use by waterbirds, by offering economic benefits to the farmers. 5. Conclusions Developing a method for prioritizing management options that is, both ecologically and economically, rigorous and practical to implement is a major challenge. The appropriate trade-off between rigor and practicality depends on the availability of time and expertise for the problem at hand. Many management agencies may not have the expertise and time to implement the most rigorous and sophisticated methods. Our protocol retains the elegant simplicity of existing cost-efficiency analysis, while adding slightly more rigorous ecological modeling than has currently be employed in similarly large action prioritization studies. We believe that the opportunity to include population demographic information and predictions in prioritization processes is worthwhile when the data and expertise exist to support it. The challenge of making rigorous ecological models a routine part of conservation management planning and prioritization remains one of the most important challenges facing conservation biologists. Acknowledgements We want to thank the researchers and students at the School of Botany (University of Melbourne) for many interesting comments. M. Bode and T. Regan helped with the PVA. S. Gilard helped with the preparation of the maps for the PVA. The following people, organizations and natural parks provided important information and comments: M. Ferrández, G. Ballesteros, J. Sánchez, J.L. Echevarrias, S. Polo, ANSE, A. Bonet, J. Jiménez, P.M. Mújica, the Conselleria de Medio Ambiente, Agua, Urbanismo y Vivienda, the directors and technicians of the Salinas de La Mata-Torrevieja, Salinas de San Pedro del Pinatar and Clot de Galvany. E. S.-G. benefited from a FPU grant from the Spanish Ministry of Education. The Servicio de Biodiversidad de la Generalitat Valenciana also funded this research. BW was supported by AEDA via a grant from the Commonwealth Department. Two anonymous reviewers made constructive critiques and comments improving the quality of this paper. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.biocon.2011.06.015. References Abellán, P., Sánchez-Fernández, D., Millán, A., Botella, F., Sánchez-Zapata, J.A., Giménez, A., 2006. Irrigation ponds as macroinvertebrate habitat in a semi-arid agricultural landscape (SE Spain). J. Arid Environ. 67, 255–269. Akçakaya, H.R., 1995. RAMAS/GIS: linking landscape data with population viability analysis (ver 2.0t). In: Anderson, E.R. (Ed.), Applied Biomathematics, Setauket, New York. Akçakaya, H.R., 2000. Viability analyses with habitat-based metapopulation models. Popul. Ecol. 42, 45–53. Akçakaya, H.R., McCarthy, M.A., Pearce, J., 1995. Linking landscape data with population viability analysis: management options for the helmeted honeyeater. Biol. Conserv. 73, 169–176. Arponen, A., Cabeza, M., Eklund, J., Kujala, H., Lehtomäki, J., 2010. Costs of integrating economics and conservation planning. Conserv. Biol. 24, 1198– 1204. Bekessy, S.A., Wintle, B.A., Gordon, A., Fox, J.C., Chisholm, R., Brown, B., Regan, T., Mooney, N., Read, S.M., Burgman, M.A., 2009. Modelling human impacts on the Tasmanian Wedge-tailed Eagle (Aquila audax fleayi). Biol. Conserv. 142, 2438– 2448. Ben-Haim, Y., 2006. Info-gap Decision Theory, second ed. Elsevier Ltd. Bryan, B.A., 2010. Development and application of a model for robust, cost-effective investment in natural capital and ecosystem services. Biol. Conserv. 143, 1737– 1750. Cabeza, M., Moilanen, A., 2001. Design of reserve networks and the persistence of biodiversity. Trends Ecol. Evol. 16, 242–248. Hegland, S.J., Dunne, J., Nielsen, A., Memmott, J., 2010. How to monitor ecological communities cost-efficiently: the example of plant – pollinator networks. Biol. Conserv. 149, 2092–2101. Jantke, K., Schneider, U.A., 2010. Multiple-species conservation planning for European wetlands with different degrees of coordination. Biol. Conserv. 143, 1812–1821. Joseph, L.N., Maloney, R.F., Possingham, H.P., 2009. Optimal allocation of resources among threatened species: a project prioritization protocol. Conserv. Biol. 23, 328–338. Juutinen, A., Mäntymaa, E., Mönkkönen, M., Salmi, J., 2004. A cost-efficient approach to selecting forest stands for conserving species: a case study from northern Fennoscandia. Forest Sci. 50, 527–539. Leathwick, J.R., Moilanen, A., Ferrier, S., Julian, K., 2010. Complementarity-based conservation prioritization using a community classification and its application to riverine ecosystems. Biol. Conserv. 143, 984–991. Lopez-Rodas, V., Maneiro, E., Lanzarot, M.P., Perdigones, N., Costas, E., 2008. Mass wildlife mortality due to cyanobacteria in the Doñana National Park, Spain. Vet. Rec. 162, 317–318. Madroño, A., Gonzalez, C., Atienza, J.C., 2004. Libro Rojo de la Aves de España. Dirección general para la biodiversidad. SEO-Birdlife, Madrid. Marsh, H., Dennis, A., Hines, H., Kutt, A., McDonald, K., Weber, E., Williams, S., Winter, J., 2007. Optimizing allocation of management resources for wildlife. Conserv. Biol. 21, 387–399. Martí, R., del Moral, J.C. (Eds.), 2003. La invernada de aves acuáticas en España. Dirección General de Conservación de la Naturaleza-SEO/BirdLife. Ed. Organismo Autónomo Parques Nacionales, Ministerio de Medio Ambiente, Madrid. Martí, R., del Moral, J.C. (Eds.), 2004. Atlas de las aves reproductoras de España. Sociedad Española de Ornitología. Organismo Autónomo de Parques Nacionales, Madrid. Martínez Fernández, J., Esteve Selma, M.A., Calvo Sendín, J.F., 2000. Environmental and socioeconomical interactions in the evolution of traditional irrigated lands: a dynamic system model. Human Ecol. 28, 279–299. McCarthy, M.A., Thompson, C., 2001. Expected minimum population size as a measure of threat. Anim. Conserv. 4, 351–355. Naidoo, R., Balmford, A., Ferraro, P.J., Polansky, S., Ricketts, T.H., Rouget, M., 2006. Integrating economic costs into conservation planning. Trends Ecol. Evol. 21, 681–687. Newbold, S.C., Siikamaki, J., 2009. Prioritizing conservation activities using reserve site selection methods and population viability analysis. Ecol. Conserv. 19, 1774–1790. Polasky, S., 2008. Why conservation planning needs socioeconomic data. Proc. Nat. Acad. Sci. USA 105, 6505–6506. Polasky, S., Camm, J.D., Garber-Yonts, B., 2001. Selecting biological reserves costeffectively: an application to terrestrial vertebrate conservation in Oregon. Land Econ. 77, 68–78. Polasky, S., Nelson, E., Camm, J., Csuti, B., Fackler, P., Lonsdorf, E., Montgomery, C., White, D., Arthur, J., Garber-Yonts, B., Haight, R., Kagan, J., Starfield, A., Tobalske, C., 2008. Where to put things? Spatial land management to sustain biodiversity and economic returns. Biol. Conserv. 141, 1505–1524. R Development Core and Team, 2005. R Development Core Team, R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna. Rocke, T.E., Friend, M., 1999. Avian botulism. In: Friend, M., Franson, J.C., (Ed.), Field Manual of Wildlife Diseases: Birds. US Geological Survey, Reston, Virginia, pp. 271–281. Rodonini, C., Boitani, L., 2007. Systematic conservation planning and the cost of tackling conservation conflicts with large carnivores in Italy. Conserv. Biol. 6, 1455–1462. Sánchez-Zapata, J.A., Anadón, J.D., Carrete, M., Giménez, A., Navarro, J., Villacorta, C., Botella, F., 2005. Breeding waterbirds in relation to artificial pond attributes: implications for the design of irrigation facilities. Biodivers. Conserv. 14, 1627– 1639. Sebastián-González, E., Sánchez-Zapata, J.A., Botella, F., 2010. Agricultural ponds as alternative habitat for waterbirds: spatial and temporal patterns of abundance and management strategies. Eur. J. Wildlife Res. 56, 11–20. Strange, N., Rahbek, C., Jepsen, J.K., Lund, M.P., 2006. Using farmland prices to evaluate cost-efficiency of national versus regional reserve selection in Denmark. Biol. Conserv. 128, 455–466. Terrones Contreras, B., 2006. Restauración ecológica de los humedales alicantinos. Instituto de cultura Juan Gil-Albert. Alicante, Spain. Van Teeffelen, A.J.A., Van Cabeza, M., Pöyry, J., Raatikainen, K., Kuussaari, M., 2008. Maximizing conservation benefit for grassland species with contrasting management requirements. J. Appl. Ecol. 45, 1401–1409. Wakamiya, S.M., Roy, C.L., 2009. Use of monitoring data and population viability analysis to inform reintroduction decisions: peregrine falcons in the Midwestern United States. Biol. Conserv. 142, 1767–1776. Wikberg, S., Perhans, K., Kindstrand, C., Djupstro, L.B., Boman, M., Mattsson, L., Martin, L., Weslien, J., Gustafsson, L., 2009. Cost-effectiveness of conservation strategies implemented in boreal forests: the area selection process. Biol. Conserv. 142, 614–624. Wilherea, G.F., Goering, M., Wang, H., 2008. Average optimacity: an index to guide site prioritization for biodiversity conservation. Biol. Conserv. 41, 770–781. Williams, S.E., Bolitho, E.E., Fox, S., 2003. Climate change in Australia’s tropical rainforests: an impending environmental catastrophe. Proc. Roy. Soc. Lond. B Biol. 270, 1187–1192. Wilson, K.A., McBride, M.F., Bode, M., Possingham, H.P., 2006. Prioritizing global conservation efforts. Nature 440, 337–340. Wintle, B.A., Bekessy, S.A., Venier, L.A., Pearce, J.L., Chisholm, R.A., 2005a. Utility of dynamic-landscape metapopulation models for sustainable forest management. Conserv. Biol. 19, 1930–1943. Wintle, B.A., Elith, R.J., Potts, J., 2005b. Fauna habitat modelling and mapping; a review and case study in the Lower Hunter Central Coast region of NSW. Austral Ecol. 30, 719–738.