Evaluation of seven aquatic sampling
methods for amphibians

Margaret S. Gunzburger

Florida Integrated Science Center, USGS, Gainesville, Florida
mgunzburger@usgs.gov

Presented at the Joint Meeting of Ichthyologists and Herpetologists
the week of July 10, 2006 in New Orleans, Louisiana.

Background

Concern about the status of amphibian populations has led to a refinement in the development and implementation of amphibian monitoring programs. Evaluating the effectiveness of sampling methods in terms of the number of species and individuals collected, as well as the relative efficiency in terms of labor and cost, is critical to planning inventory and monitoring programs. A side-by-side comparison of multiple methods at repeated sites is the best way to evaluate sampling methods.

Objectives

The overall goal of this study was to determine what combination of methods should be used to provide the best estimates of aquatic amphibian community richness. This study focused on lentic aquatic habitats in the Coastal Plain of the southeastern United States, but the results may be relevant to other areas with similar habitat types. Specific objectives of this study were: 

• To determine which amphibian species were collected by each method.
   
• To evaluate the precision of individual counts for each method.
   
• To compare the number and identity of fish and invertebrate species sampled with each method.

Methods

Ten wetland sites at the Ordway-Swisher Biological Station in Putnam Co., FL were sampled from 2-11 May 2005 with each of seven amphibian sampling methods. Four methods allowed individual counts (D-frame dipnet, metal dipnet, crayfish trap, box trap). Ten samples per site were collected for each of these methods and total count of amphibians, fish, and invertebrates was recorded. Three additional methods (Froglogger - automated audio recording device, aural, and visual sampling) allowed collection of detected/not-detected data for amphibians. 

Results

Amphibian Species Richness

A total of 11 species of anurans and 2 species of caudates were detected during this study. Amphibian species richness varied among the seven sampling methods. There was a positive correlation between the number of sites at which a species was detected and the number of methods by which that species was detected. There was no relationship between the number of anuran species detected as adults (by froglogger, visual, aural sampling) and the number of anuran species detected as larvae (crayfish trap, dipnet, box trap) at a site.

Amphibian Abundance

The total number of amphibian individuals collected at a site was positively correlated with the number of amphibian species detected at that site by all methods. Precision for the four count sampling methods was relatively low and similar for the four methods. The count of amphibian individuals was higher with the metal dipnet than D-frame dipnet, and count in both dipnet types was positively correlated with the count in the box trap.

Amphibian Species Richness

Occupancy (naïve and estimated Ψ) and detection probabilities (p) for five species of amphibians at nine localities using three different sampling methods (10 repeat samples for each method are considered separate visits in the analysis). A dash indicates the species was not detected with that method.  Two site covariates included in the analysis were pH and presence of fish. 

Accumulation curves for amphibian species, fish species, and invertebrate taxa for the four count sampling methods. 

Number of 10 total sites at which each of 13 amphibian species were detected using 7 aquatic sampling methods. 

Conclusions

• There was significant variation among the seven sampling methods tested in terms of species richness and counts of amphibians. In general, the methods which required the most time (froglogger and box trap) also yielded the greatest species richness. The variation in effectiveness of these seven methods is due in part to variation in behavior and life-history of the different amphibian species. Caudates were consistently detected with fewer methods than anurans because they do not vocalize and thus are impossible to detect aurally or with frogloggers, and are unlikely to be sampled during visual encounter surveys.
   
• Detection of anuran species varied by life stage and species, and appeared to be related to breeding season and larval period length. Detection of anuran larvae was generally correlated with tadpole size and activity level, with larger, more active tadpoles (ranids) captured more often using passive crayfish traps, and smaller tadpoles (hylids) more often detected using active sampling methods.
   
• Occupancy and detection varied considerably for several species for three different sampling methods. Occupancy estimates may not be comparable across studies with different sampling methods.
   
• A combination of frogloggers, crayfish traps, and dipnetting is a cost and time efficient sampling protocol for amphibian inventory and monitoring. This study has demonstrated the importance of recording incidental, and virtually cost-free, observations, including visual encounter and aural sampling, because all but two of the eleven species of anurans found in this study were detected using these techniques combined.