Spatial priorities for freshwater biodiversity conservation in light of catchment protection and connectivity in Europe

proxy for the cost of protection [44] to minimize total selected area for efficiency. The maxi- mum total cost of the solutions was specified as 17% of the total area of Europe, corresponding to Target 11 of the global Aichi Biodiversity Targets. We used three Marxan outputs: (i) the network that best met the targets at the lowest cost, (ii) the proportion of species’ ranges cov- ered by the best network, and (iii) the number of times a catchment was selected as part of the optimal network in 1000 runs (selection frequency, a proxy for the irreplaceability of the catch- ments). We note that “irreplaceability” is used here as a general measure of conservation prior- ity and not in the sense of “ecological irreplaceability” [45]. We obtained spatial data on Natura 2000 protected areas in member states of the European Union (EU) from the European Environmental Agency [46] and on areas protected by national laws from the World Database on Protected Areas (IUCN categories I to IV) [47]. We then calculated the combined proportion of protected areas in each catchment as the sum of all areas protected (either by Natura 2000 or national laws, with the overlap between the two considered only once) divided by catchment area. We used the R statistical environment [48] for all calculations and data preparations and ArcGIS 10.0 for Windows [49] for visualizing the results on maps. All calculations were conducted in the Lambert Azimuthal Equal Area projection (coordinate system GCS_ETRS_1989), which is appropriate for area calculations

and is recommended for statistical analysis and display in Europe [50]. To answer Questions 1–3, we ran three scenarios. In Scenario 1, catchment protection was not considered. In Scenario 2, we studied whether the inclusion of well-protected catchments improves the efficiency of prioritization, defined as the number of threatened species repre- sented in the solution given a fixed surface area (17% of the planning area). We implemented this by locking in well-protected catchments a priori in the prioritization. We defined a catch- ment as well-protected if the proportion of protected areas exceeded 70% of the total catch- ment area. This threshold was chosen to reflect that the macroinvertebrate and fish fauna of pristine and degraded rivers do not differ substantially as long as not more than 30% of the catchment area is transformed to agriculture [51]. The total area of well-protected catchments (n = 1011) was 206,878 km2, corresponding to 2% of the total area of geographical Europe. Finally, in Scenario 3, we aimed to identify catchments with high biodiversity and low protec- tion (more than 30% not protected), which may be degraded, and which may thus need further conservation interventions (more protected areas, management, restoration, species protec- tion) by excluding (locking out) well-protected catchments from the prioritization. To evaluate the effect of catchment connectivity on conservation priorities, we ran each sce- nario without connectivity and then again with connectivity. In the first case, we used the Bound-

ary Length Modifier (BLM) parameter of Marxan to control for habitat fragmentation (Ball et al. 2009). Based on the results of preparatory Marxan runs (S1-S3 Figs in S1 File), we selected an appropriately large value (BLM = 10) to avoid solutions that are too fragmented and to enhance clumping of priority catchments [52]. In the second case, we implemented the idea described by [27] and implemented connectivity in Marxan runs by replacing the BLM file with a river connec- tivity file that listed catchments and their immediate downstream catchment in a second field. We used the HydroBASINS database to create this file, which contains the outflowing catchments for each river catchment. Because upstream catchments are expected to have a larger influence on catchments downstream than vice versa, we only used upstream connectivity. In Marxan, the parameter Connectivity Penalty (CP) controls the connectivity of the optimal network relative to the area selected for protection. To determine an optimal value for CP, we implemented prepara- tory Marxan runs with eight different values of CP (S1 File). The results suggested that a CP = 10 is a reasonable compromise between costs and connectivity (S4 Fig in S1 File). We developed three ways to assess the correspondence between conservation priority and current level of protection (Question 4) by assuming that if protection is ideal, there should be PLOS ONE Spatial priorities for freshwater biodiversity in Europe PLOS ONE | https://doi.org/10.1371/journal.pone.0267801 May 17, 2022 4 / 15