Apex research : Shenandoah National Park
Apex Sites are specific National Parks or National Wildlife Refuges where intensive amphibian research and monitoring is being conducted. Research is aimed at answering questions about changes in amphibian populations, diseases, and malformations or gathering more detailed information about the dynamics of particular amphibian populations.
Stream salamanders in Shenandoah National Park: Movement and survival of stream salamander populations
Research in population biology is concerned with factors affecting the change in a population over time, including births, deaths, immigration and emigration. Despite the potential importance of immigration and emigration, empirical data on movement patterns are lacking in many systems. Hence, there is a large body of theory on the causes and consequences of dispersal that remains to be tested in real biological systems. Until recently, research on stream salamanders has not considered the role of dispersal in determining the distribution and abundance of species on the landscape. Recent work suggests that these organisms may be able to move among habitats separated by large distances, despite their diminutive size and physiological reliance on moist microhabitats. Stream ecosystems have been dramatically influenced by land-use change and habitat loss that has resulted in the loss of entire tributaries or the fragmentation of stream reaches within a watershed. We know that the hierarchical nature of stream networks can influence population dynamics and patterns of connectivity in stream-associated organisms and this may have a disproportionate influence on extinction risk for species or life stages that are restricted to movement along riparian corridors.
The goals of this research are to understand 1) rates of interpopulation dispersal in a community of stream-associated salamanders, 2) whether dispersal is related to the spatial arrangement of habitat reaches, population density and demography, and 3) whether the dynamics of a local populations are influenced by dispersal from adjacent stream segments in the river network.
Our study includes an experimental removal and translocation of a population of stream salamanders, combined with repeat visits to capture individually marked animals. Under this design, we will be able to (1) evaluate survival of resident and translocated animals and (2) quantify the probability of dispersal for E. bislineata, D. fuscus, D. monticola and G. porphyriticus to test whether patterns of movement relate to individual characteristics or characteristics of the stream habitat (i.e., a direct measure of dispersal). This work is being conducted in Shenandoah National Park, an ideal place to study movement of stream salamanders in a natural system. Information from this study can be easily translated to other National Parks in the Northeastern United States, and will be useful in managing stream systems within natural areas in the face of increasing urbanization.
In the first year of the study (2007), we captured and marked over 3500 unique individuals. We will use information from 2008 to estimate survival and dispersal probabilities.
Monitoring the Shenandoah salamander in Shenandoah National Park
Plethodon shenandoah is a small, terrestrial woodland salamander, with a limited distribution solely within Shenandoah National Park. Originally listed as federally endangered in August 1989, the Shenandoah salamander was thought to be declining due, exclusively, to natural interspecific competition with the red-backed salamander (Plethodon cinereus), though other human-related factors have been mentioned as possible contributing factors (e.g. forest defoliation, trail and road use, and fire management practice) (USFWS 1994). Under the Plethodon shenandoah Recovery Plan (USFWS 1994), a principal challenge posed to both USFWS and Shenandoah National Park (SNP) is the protection, monitoring and maintenance of remaining P. shenandoah populations. Our study has been funded to establish the current status of existing P. shenandoah population(s), explore factors that may limit its current distribution, and develop a long-term monitoring program to explore the relative impacts of natural and human-caused factors.
Several objectives need to be met to develop an effective long-term monitoring program including exploring sampling design tradeoffs such as the size and shape of sample plots, timing and conditions for optimal survey effort, and the appropriate number of repeated surveys necessary to achieve unbiased and precise estimates of the proportion of habitat occupied by the endangered salamander. Efforts for the 2007 pilot sampling season focused on locations within P. shenandoah historically known range on Hawksbill and Stony Man mountains. Pinnacles was not included in the 2007 sampling because the historic P. shenandoah distribution on this mountain was small (less than 10 ha) and not well documented (USFWS 1994). We selected a random set of 4 points that were at least 200 meters apart (to insure independence) within the previously known stratum on each mountain. The Appalachian Trail bisects both mountains and since trail use and maintenance may influence P. shenandoah populations by serving as a barrier to movement, or as a conduit to the proliferation of the competitor P. cinereus, we placed paired sample sites at each point: one 15 meters above the AT trail and one 15 meters below the trail (8 sites each on Stony Man and Hawksbill mountains). These 16 sites served as our pilot study 'sites'. Two transects (8 and 32 m long x 2 m wide) and 2 plots (4 m x 4 m and 8 m x 8 m) were established at each site. It's possible that the shape of the sampled area (long transects vs. square plots) may affect the probability of detection, presumable because transects may contain more heterogeneous habitat compared to square plots. Sites were surveyed at least 3 times during the active spring, summer, and fall sampling seasons under varying conditions: day and night-time searches, during wet or dry conditions. During each survey of a site, the number of individuals of each of the 2 target species (P. shenandoah and P. cinereus) were recorded, measured, weighed and sexed, if possible. Several habitat and environmental covariates were also measured during each site visit.
Using these 2007 pilot data, we determined the optimal sample size to: (1) investigate potential interspecific competition between P. shenandoah and P. cinereus) (considered the main natural threat to the remaining P. shenandoah populations: see USFWS 1994 and citations within), and (2) establish the extent of the current distribution of P. shenandoah within the known and unknown population talus habitat locations on all 3 mountains. To investigate potential interactions between P. shenandoah and P. cinereus, the optimal sample size should be small (4 m x 4 m plots sampled 3 times per season), so that changes in P. shenandoah and P. cinereus) distribution within small temporal and spatial windows can be explored. To establish the extend of the current distribution of P. shenandoah, larger plots or transects will be placed at randomly chosen points from a grid across all talus slope habitat on each mountain. These points will be chosen from both the previously known and unknown population habitat strata, and will be used to determine the current distribution of P. shenandoah and P. cinereus) across available talus habitats. Based on analysis from the pilot data, larger plots and transects have a much higher detection probability given that P. shenandoah is present (estimated detection probabilities ~ 0.65-0.80, depending on the season, with higher detection probabilities in the spring and summer). Sampling effort in 2008-2009 will focus on these 2 objectives.
Mid-level monitoring: National Capital Region
Monitoring amphibians at the mid-level is coordinated at a moderate number of parks or refuges across the region where amphibian habitats are sampled and inferences drawn about the occurrences of select species within the area. The state variable of interest is the probability of occupancy.
Developing a monitoring program for the National Capital Region
The National Capital Region Network has identified amphibians as a priority taxonomic group for its Inventory and Monitoring program. The goals of this program are to document at least 90% of the amphibian species in its parks, and to determine whether the integrity and status of amphibian populations are changing over time. As of 2004, all of the parks had completed amphibian inventories (Shawn Carter, National Park Service, personal communication). Amphibian monitoring was initiated in 2005, and is currently concentrated in Chesapeake and Ohio Canal National Historic Park and Rock Creek Park, with stream sampling also occurring in Prince William Forest Park.
The objectives of the monitoring program are to develop an efficient long-term sampling design to: (1) describe the current distribution of amphibians and explore factors that may influence occupancy probabilities or distributional patterns, (2) determine if amphibian distributions are changing annually, and if so, explore whether occupancy changes are relate to habitat quality and (3) provide information to aid in the generating and testing hypotheses that differentiate among possible causes of long-term changes in the proportion of area occupied among species, habitats, and park areas.
In the first three years (2005-2007) of wetland amphibian surveys there was only a slight change in the overall probability of occupancy, but there was considerable local turnover for most species, indicating that the same sites are not occupied each year. All of the environmental and habitat covariates influenced initial occupancy estimates for one or more species, especially wetland hydroperiod. Hydroperiod was an important variable in initial occupancy estimates for seven of the eight species analyzed and in all cases, the influence was positive. This finding suggests that wetland hydroperiod is likely a limiting factor in determining the distribution of many of the wetland breeding amphibians found at Chesapeake and Ohio Canal National Historical Park. Urbanization variables (conductivity) influenced the initial occupancy probability of several species, but these variables did not seem to influence site occupancy dynamics.
There was considerable uncertainty in terms of variables influencing occupancy probabilities and rate parameters for the 3 stream salamander species analyzed: Desmognathus fuscus, Eurycea bislineata, and Pseudotriton ruber. In general, initial occupancy estimates were higher for stream transects in CHOH compared to ROCR, except for E. bislineata which had high occupancy probabilities in both parks. D. fuscus and P. ruber had higher occupancy probabilities at transects near the stream headwaters, but our a priori hypothesis that proximity of the stream origin to the park boundary or road would result in lower occupancy probabilities was not well supported for any of the 3 species. We expected models specifying no turnover (i.e., ? = ? = 0) to be among the top models for all species, but this was not the case. While stream communities changed little among the 3 years study, models where colonization probabilities included the 'network' covariate were among many of the top models for P. ruber, suggesting the potential for higher colonization probability in stream networks with a confluent first order branch. As with our lentic sampling, more years of data, encompassing both wet and dry years, are needed to further elucidate the potential relationship between spatial covariates and site occupancy and related rate parameters for stream salamander species.
Finally, to continue to enhance and optimize the NCR monitoring effort, we explored the trade-offs between different survey designs with our updated parameter estimates, under specified monitoring objectives. Using data from the two years of wetland surveys and a new program GENPRES (Bailey et al. 2007) we explored the ability of several different survey designs (varying the number of sites and frequency of visits) to detect change in amphibian populations within this biological system. The results of these simulations will help managers decide if they are satisfied with the current study's ability to detect change in amphibian occupancy and associated vital rates.
Bailey, L.L., J.E. Hines, J.D. Nichols, and D.I. MacKenzie. 2007. Sampling design trade-offs in occupancy studies with imperfect detection: examples and software. Ecological Applications17: 281-290.
Mid-level monitoring: Patuxent Research Refuge
Locating vernal pools and estimating the amount of habitat available for vernal pool-breeding amphibians
The loss of small, seasonal wetlands is a major concern for a variety of state, local, and federal organizations in the northeastern US. Identifying and estimating the number of vernal pools within a given region is critical to developing long-term conservation and management strategies for these unique habitats and their faunal communities.
Vernal pools have not been reliably mapped throughout most of the eastern US (Calhoun et al. 2003, Burne and Lathrop 2008). Current methods of locating and documenting vernal pools typically rely on map or photograph interpretation combined with field-verification of these putative habitats (Brooks et al. 1998, Burne 2001, Calhoun et al. 2003, Lathrop et al. 2005). Other means of estimating and mapping vernal pools appear promising, but have yet to be rigorously tested or applied across multiple landscapes (e.g., modeling physiographic features, Grant 2005) or using other remote-sensing technologies (Weier 2004). Ideally, any program whose goals are to conserve, monitor, and protect vernal pool habitat would utilize a sampling method that: 1) identifies and locates a subset of vernal pools (i.e., a sample) from which an unbiased estimate of the total amount of vernal pool habitat can be obtained, and 2) serves as a representative sample from which inference can be made about the status of the associated target organism(s) (e.g., species such as wood frogs and mole salamanders).
Sampling was conducted on one State Park and 10 National Wildlife Refuges across eight states in the Mid-Atlantic and Northeast regions of the US as part of a joint project between the US Geological Survey's (USGS) Amphibian Research and Monitoring Initiative (ARMI) and the US Fish and Wildlife Service's Region 5 National Wildlife Refuges (NWR).
We use three probabilistic sampling methods (simple random sampling, adaptive cluster sampling, and the dual frame method) to estimate the number of vernal pools on protected, forested lands. Adaptive cluster sampling is a set of probabilistic methods designed for sampling spatial clustered populations (Thompson 1990, Thompson and Seber 1996). Vernal pools often exhibit a clustered distribution in the landscape (Brooks et al. 1998, Grant 2005), and thus using an adaptive cluster sampling framework may yield more precise estimates of vernal pool abundance compared to traditional SRS methods. The dual-frame method utilizes two independent sample frames: an 'area' frame and a 'list' frame. The area frame is defined by the geographic boundary of the region of interest and includes all possible sample units. The list frame is often incomplete and consists of a subset of the area frame for which the locations of the sample units or objects are known. As such, the dual-frame method offers a way of utilizing a list of potential vernal pool locations obtained from aerial photograph interpretation, while acknowledging that the list does not constitute a census of vernal pools. Overall, these methods yielded similar values of vernal pool abundance for each study area, and suggest that photographic interpretation alone may grossly underestimate the number of vernal pools in forested habitats.
Now that we know the distribution of vernal pools on select federal lands across the northeastern United States, our monitoring program will investigate large-scale patterns in the distribution and occupancy of vernal pools by wood frog and spotted salamanders. We will relate species presence to surrounding land use, road density, proximity to or density of other potential breeding sites, water quality variables, hydroperiod, and climatic conditions.