Bending the curve: Simple but massive conservation action leads to landscape-scale recovery of amphibians

Abstract

Success stories are rare in conservation science, hindered also by the research-implementation gap, where scientific insights rarely inform practice and practical implementation is rarely evaluated scientifically. Amphibian population declines, driven by multiple stressors, are emblematic of the freshwater biodiversity crisis. Habitat creation is a straightforward conservation action that has been shown to locally benefit amphibians, as well as other taxa, but does it benefit entire amphibian communities at large spatial scales? Here, we evaluate a landscape-scale pond-construction program by fitting dynamic occupancy models to 20 y of monitoring data for 12 pond-breeding amphibian species in the Swiss state Aargau, a densely populated area of the Swiss lowlands with intensive land use. After decades of population declines, the number of occupied ponds increased statewide for 10 out of 12 species, while one species remained stable and one species further declined between 1999 and 2019. Despite regional differences, in 77% of all 43 regional metapopulations, the colonization and subsequent occupation of new ponds stabilized (14%) or increased (63%) metapopulation size. Likely mechanisms include increased habitat availability, restoration of habitat dynamics, and increased connectivity between ponds. Colonization probabilities reflected species-specific preferences for characteristics of ponds and their surroundings, which provides evidence-based information for future pond construction targeting specific species. The relatively simple but landscape-scale and persistent conservation action of constructing hundreds of new ponds halted declines and stabilized or increased the state-wide population size of all but one species, despite ongoing pressures from other stressors in a human-dominated landscape.

Keywords: amphibian decline; conservation management; evidence-based conservation; freshwater biodiversity; recovery.

Fig. 1.

Trajectories of estimated metapopulation size (number of occupied sites; mean with 95% CI) in five regions from 1999 to 2019. Not all species occur in all regions. Filled upward triangles indicate increases in metapopulation size (with certainty >90%), open circles no change, and open downward triangles decreases between 1999 and 2019. EN, VU, NT, LC, and NE in parentheses indicate the national Red List status following IUCN criteria and categories.

Fig. 2.

Patterns of change in old and new ponds for 43 regional metapopulations with (A) increasing (n = 28 metapopulations), (B) stable (n = 9), and (C) overall declining (n = 6) metapopulation size from 1999 to 2019 (>90% probability). Left panels show the number of metapopulations that follow each of eight possible patterns. Bold font indicates patterns where the colonization of new ponds influenced the overall trend (33 or 77% of the 43 metapopulations). Right panels show one example per pattern of a corresponding trajectory of estimated total regional metapopulation size (black), split into the number of old (blue) and new (red) sites occupied. Symbols indicate increasing (filled upward triangles), stable (open circles), or decreasing (open downward triangles) Nos. of occupied ponds from 1999 to 2019 (with >90% probability). Error bars show 95% CI.

Fig. 3.

Probabilities of (A) colonization and (B) persistence in new ponds (red, with 95% CI) compared with mean probabilities in old ponds (blue, point estimates as baseline without associated uncertainty) across all regions. Note the log scale y axis for colonization probability. Estimated probabilities in old and new ponds in individual regions are shown in SI Appendix, Fig. S6.

Fig. 4.

Estimates (logit scale) of the effects of environmental variables on colonization probability (mean, 95% CI). Environmental variables were pond surface area, fluctuations of the water table, the percent area of forest within 100 m of the pond, elevation, the area of large (width ≥6 m) roads within 1 km of the pond, connectivity, and the age of newly constructed ponds. Absolute values of parameter estimates indicate effect size, as all covariates were standardized to mean zero and unit variance prior to analysis.