Identifying the minimal demographic unit for monitoring pond-breeding amphibians

Amphibians are declining globally, and efforts are underway to establish long-term programs to monitor population trends. Because researchers are constrained to sampling a small subset of populations that Occur across a landscape, monitoring units should be selected that maximize the statistical independence of population responses to environmental stressors. Breeding sites of pond-breeding amphibians are often Clustered locally, and habitat switching between neighboring ponds could significantly compromise both demographic and statistical independence. This raises a fundamental issue of whether the basic monitoring unit for trend analysis should be an individual pond or a cluster of local ponds. To help resolve the degree of demographic independence within a cluster of local ponds, we conducted a spatial-scale analysis of breeding-population synchrony of the wood frog (Rana sylvatica) and spotted salamander (Ambystoma maculatum) in four national parks in Kentucky, North Carolina, and Tennessee, USA. We used the number of egg masses deposited annually from 1993 through 2002 as an index of relative breeding-population size and used correlation coefficients of annual percentage change in population size as an index of synchrony. Within parks, synchrony for both species was independent of the distance between ponds and was lower between paired ponds within a local cluster compared with paired comparisons of geographically isolated groups. These patterns indicate a lack of demographic independence within clusters and tire consistent with field observations of wholesale shifting of local pond populations to adjoining ponds. Habitat shifting occurred primarily in response to beaver activity that created new ponds or allowed fish to invade fish-free habitats. Ambystoma maculatum shifted habitats less frequently and exhibited higher levels of within-Cluster population synchrony than R. sylvatica. Turnovers occurred for local pond populations for both species and at higher rates for R. sylvatica. However, no local pond cluster experienced a turnover event during the study. Our data suggest that pond populations that are only a few hundred meters apart are not demographically independent and are best treated as subpopulations of the same monitoring unit. We discuss the implications of these findings for amphibian-monitoring programs that are designed to assess population trends.