Monitoring Martes Populations in California: Survey Design and Power Analysis

Fishers (Martes pennanti) and American martens (M. americana) have been protected from trapping in California since the mid19OOs, yet in portions of each of their historic ranges their numbers are extremely low, perhaps due to the effects of timber harvest. We propose a method capable of detecting declines in the occurrence and distribution of fishers or martens using baited track-plate stations. The proposed sampling unit is a small grid of stations that has a high probability of detecting animals when they are present. These multistation units are sufficiently spaced to meet the assumption of independence for a binomial model. We propose a stratified random sampling design with strata sampled for proportions of occurrence at discrete points in time. Stratification is based on variation in occurrence by region and is estimated from preliminary survey data. A previously published bias adjustment is applied to the proportion of units with detections to adjust for possible failure to detect resident individuals at a sampling unit. A Monte Carlo simulation model was developed to determine the sample size necessary to detect 20 and 50% declines, with 80% power, in the proportion of sampling units with occurrence. We assume a 10-yr sampling interval. Sensitivity analysis, using a range of values for means and standard deviations of strata proportions, determined that power was much more sensitive to changes in mean than the standard deviation. When the best current estimates of the fisher strata proportions were input for 10 strata (five regional and two habitat) in California, 115 and 17 sampling units per stratum were necessary to detect 20 and 50% declines, respectively. For some circumstances this sampling effort was also sufficient to achieve strata estimates with 5% error and to detect statistical differences between individual stratum proportions. The steps in the process of implementing a monitoring program for Pacific fishers in California are outlined as an example of the planning and preparation necessary to monitor changes in the distribution of a rare forest carnivore. Key power. words: detection; forest carnivores; Martes; monitoring; occurrence; sign survey; statistical