A comparison of monitoring methods for inferring demographics of a pond-breeding amphibian population over 14 years

Abstract

Long-term monitoring for conservation typically requires unbiased estimates or indices of abundance, thereby allowing inference of population trends. Biased estimates can arise when individuals vary in capture probability, especially when overall capture probability is low. We used 3 methods to compare estimates and indexes of the abundance and sex ratio of spotted salamanders (Ambystoma maculatum) breeding in a 4-ha lake over 14 years to counts from a drift fence around the perimeter of the lake. The 3 sampling methods were 1) deployment of aquatic funnel traps around the lake and marking all individuals captured; 2) counting egg masses on the lake; and 3) surveying a coverboard array around the perimeter. To estimate male and female abundance and survival from the aquatic trapping data, we fitted a state-space formulation of the Jolly-Seber model with data augmentation to the encounter histories. We compared model formulations with and without individual heterogeneity in capture probability. Annual breeding turnout at the drift fence varied from 4,476 to 5,282 individuals for the 3 years surveyed. Abundance estimates from aquatic trapping data were substantially underestimated (lower by 67–75%) relative to drift fence counts when individual heterogeneity in capture probability was not addressed, whereas including individual heterogeneity resulted in sex-specific abundance estimates that were consistent with drift fence surveys. Raw sex ratios from aquatic trapping were male-biased (ranging from 0.67 to 0.81 proportion male, or 2.0:1 to 4.2:1 as a male:female ratio), whereas counts from the drift fence survey and estimates from the Jolly-Seber model with individual heterogeneity indicated female-biased sex ratios (ranging from 0.35 to 0.49 proportion male, or a ratio of 0.54:1 to 0.95:1). Egg mass counts were weakly correlated with abundance estimates, and cover board surveys detected less than 1% of the population. Our study underscores the utility of drift fence surveys as a reference for estimating the abundance of breeding individuals and the importance of considering individual heterogeneity in capture probability when estimating abundance of pond-breeding amphibians from capture data with imperfect detection. Our study also cautions that potential indices of abundance, such as egg-mass counts and coverboard counts, should be validated against reliable abundance estimates.