Search ARMI Database

The invasive American bullfrog (Lithobates catesbeianus) and a variety of non-native sport fish commonly co-occur in lowland lentic habitats of the western United States. Both invasive taxa are implicated in declines of native amphibians in this region, but few long-term studies of communities exist. Further, field studies of invasive-native interactions are complicated by confounding habitat modifications and observation errors. We surveyed amphibians and measured habitat characteristics for 12 years across 38 wetland sites within the Willamette Valley, Oregon, USA. We assessed the influence of invasive species, habitat, and their interactions on the distributions of five native amphibian species using a multispecies dynamic occupancy model that accounted for false-negative and false-positive detections. In general, habitat characteristics ? such as within-pond vegetation cover, surrounding forest, and drought severity ? were important for local persistence of native species when bullfrogs co-occurred. We also found evidence of a cumulative negative effect of bullfrogs and non-native fish (families Centrarchidae and Ictaluridae) on northern red-legged frog (Rana aurora) local persistence that was mediated by the dominance of invasive reed canarygrass (Phalaris arundinacea). Non-native fish and bullfrogs had variable effects on native amphibian species, but neither invasive taxa appears to be causing declines in occupied sites within our study area. Moreover, species relationships with habitat differed when invaders were present, indicating that certain habitats may increase persistence of native amphibians in the invaded landscape.

Infectious diseases are causing catastrophic losses to global biodiversity. Iridoviruses in the genus Ranavirus are among the leading causes of amphibian disease-related mortality. Polymorphisms in major histocompatibility complex (MHC) genes are significantly associated with variation in amphibian pathogen susceptibility. MHC genes encode two classes of polymorphic cell-surface molecules that can recognize and bind to diverse pathogen peptides. While MHC class I genes are the classic mediators of viral acquired immunity, larval amphibians do not express them. Consequently, MHC class II gene diversity may be an important predictor of Ranavirus susceptibility in larval amphibians, the life stage most susceptible to Ranavirus. We surveyed natural populations of larval wood frogs (Rana sylvatica), which are highly susceptible to Ranavirus, across 17 ponds and two years in Maryland, USA. We sequenced the peptide-binding region of an expressed MHC class II? locus and assessed allelic and genetic diversity. We converted alleles to functional supertypes and determined if supertypes influenced host responses to Ranavirus. Among 384 sampled individuals, 26% were infected with Ranavirus. We recovered 20 unique MHC class II? alleles that fell into two deeply diverged clades and seven supertypes. MHC genotypes were associated with Ranavirus infection intensity, but not prevalence. Specifically, heterozygotes and individuals with genotype ST1/ST7 had significantly lower Ranavirus infection intensity compared to homozygotes and all other genotypes. We conclude that MHC class II? functional genetic variation is an important component of Ranavirus susceptibility. Identifying immunogenetic signatures linked to variation in disease susceptibility can inform mitigation strategies for combatting global amphibian declines.

Understanding how natural and anthropogenic processes affect population dynamics of species with patchy distributions is critical to predicting their responses to environmental changes. Despite considerable evidence that demographic rates and dispersal patterns vary temporally in response to an array of biotic and abiotic processes, few applications of metapopulation theory have sought to explore factors that explain spatio-temporal variation in extinction or colonization rates. To facilitate exploring these factors, we extended a spatially explicit model of metapopulation dynamics to create a framework that requires only binary presence-absence data, makes few assumptions about the dispersal process, and accounts for imperfect detection. We apply this framework to 22 years of biannual survey data for lowland leopard frogs, Lithobates yavapaiensis, an amphibian that inhabits arid stream systems in the southwestern U.S. and northern Mexico. Our results highlight the importance of accounting for factors that govern temporal variation in transition probabilities, as both extinction and colonization rates varied with hydrologic conditions. Specifically, local extinctions were more frequent during drought periods, particularly at sites without reliable surface water. Colonization rates increased when larval and dispersal periods were wetter than normal, which increased the probability that potential emigrants metamorphosed and reached neighboring sites. Extirpation of frogs from one watershed during a period of severe drought demonstrated the influence of site-level features, as frogs persisted only in areas where most sites held water consistently and where the amount of sediment deposited from high-elevation wildfires was low. Application of our model provided novel insights into how climate-related processes affected the distribution and population dynamics of an arid-land amphibian. The approach we describe has application to a wide array of species that inhabit patchy environments, can improve our understanding of factors that govern metapopulation dynamics, and can inform strategies for conservation of imperiled species.

We developed a multi-region community occupancy model to analyze 13 years (2005–2017) of amphibian monitoring data within the National Capital Region, a network of U.S. National Parks. Our analysis reveals how scale can mediate interpretation of results from scientific studies, which might help explain conflicting narratives concerning the impacts of fragmentation in the literature. Our hierarchical framework can help managers and policymakers elucidate the relevant spatial scale(s) to target conservation efforts.

1. Obtaining inferences on disease dynamics (e.g., host population size, pathogen prevalence, transmission rate, host survival probability) typically requires marking and tracking individuals over time. While multistate mark?recapture models can produce high-quality inference, these techniques are difficult to employ at large spatial and long temporal scales or in small remnant host populations decimated by virulent pathogens, where low recapture rates may preclude the use of mark?recapture techniques.

2. Recently developed N-mixture models offer a statistical framework for estimating wildlife disease dynamics from count data. N-mixture models are a type of state-space model in which observation error is attributed to failing to detect some individuals when they are present (i.e., false negatives). The analysis approach uses repeated surveys of sites over a period of population closure to estimate detection probability.

3. We review the challenges of modeling disease dynamics and describe how N-mixture models can be used to estimate common metrics, including pathogen prevalence, transmission, and recovery rates while accounting for imperfect host and pathogen detection. We also offer a perspective on future research directions at the intersection of quantitative and disease ecology, including the estimation of false positives in pathogen presence, spatially-explicit disease-structured N-mixture models, and the integration of other data types with count data to inform disease dynamics.

4. Managers rely on accurate and precise estimates of disease dynamics to develop strategies to mitigate pathogen impacts on host populations. At a time when pathogens pose one of the greatest threats to biodiversity, statistical methods that lead to robust inferences on host populations are critically needed for rapid, rather than incremental, assessments of the impacts of emerging infectious diseases.

Roadside amphibian citizen science (CS) programmes bring together volunteers focused on collecting scientific data while working to mitigate population declines by reducing road mortality of pond-breeding amphibians. Despite the international popularity of these movement-based, roadside conservation efforts (i.e. “big nights,” “bucket brigades” and “toad patrols”), direct benefits to conservation have rarely been quantified or evaluated. As a case study, we used a population simulation approach to evaluate how volunteer intensity, frequency and distribution influence three conservation outcomes (minimum population size, population growth rate and years to extinction) of the spotted salamander (Ambystoma maculatum), often a focal pond-breeding amphibian of CS and conservation programmes in the United States.

Temporary wetlands have value to both ecological and social systems. Interactions between local climate and the surrounding landscape result in patterns of hydrology that are unique to temporary wetlands. These seasonal and annual fluctuations in wetland inundation contribute to community composition and richness. Thus, predicting wetland community responses to environmental change is tied to the ability to predict wetland hydroregime. Detailed monitoring of wetland hydroregime is resource-intensive, limiting the scope and scale of forecasting. As an alternative, we determine which freely available measures of water availability best predict one component of wetland hydroregime, habitat suitability (i.e., the predictability of water in a wetland) within and among geographic regions. We used data from three North American regions to determine the climate index that best explained year-to-year variation in habitat suitability during a key phenological period—amphibian breeding. We demonstrate that simple, short-term climate indices based solely on precipitation data best predict habitat suitability in vernal pools in the northeast, montane wetlands in the west and coastal plain wetlands in the southeast. These relationships can help understand how changes in short-term precipitation patterns as a result of climate change may influence the overall hydroregime, and resulting biodiversity, of temporary wetlands across disparate biomes.

Batrachochytrium salamandrivorans (Bsal), a pathogenic chytrid fungus, is nonnative to the United States and poses a disease threat to vulnerable amphibian hosts. The Bsal fungus may lead to increases in Threatened, Endangered, and Sensitive status listings at local, state, and federal levels, resulting in financial costs associated with implementing the Endangered Species Act . The U.S. is a global biodiversity hotspot for salamanders, an order of amphibians that is particularly vulnerable to developing a disease called chytridiomycosis when exposed to Bsal. Published Bsal risk assessments for North America have suggested that salamanders within the Appalachian region of the U.S. are at a high risk. In May 2017, a workshop was facilitated by the Department of the Interior?s (DOI) Strategic Sciences Group (SSG). A discussion-based incident-response exercise focused on a hypothetical Bsal disease outbreak in Appalachia was led by U.S. Geological Survey (USGS) staff members. Participants included representatives of the Eastern Band of the Cherokee Indians, the U.S. Fish and Wildlife Service (USFWS), National Park Service, Appalachian Landscape Conservation Cooperative, Tennessee Wildlife Resources Agency, and U.S. Department of Agriculture?s U.S. Forest Service. Scenario-building was utilized to brainstorm cascading consequences (social, economic and ecological) of a Bsal disease outbreak in this region of Appalachia. This report highlights the management and science actions that should could be undertaken to ensure an effective, rapid response to a Bsal introduction to the United States.

Amphibian populations are exposed to multiple stressors, with potential for synergistic effects. These synergies can increase uncertainty in our ability to characterize the effects of each stressor and to understand the degree to which their effects interact to impact population processes. This uncertainty challenges our ability to identify appropriate management alternatives. Finding solutions that are robust to these uncertainties can improve management when knowledge is absent or equivocal and identify critical knowledge gaps. Bayesian Belief Networks (BBNs) are probabilistic graphical models that explicitly account for various sources of uncertainty and are used increasingly by environmental practitioners because of their broad applicability to ecological risk assessments. BBNs allow the user to: 1) generate a conceptual model to link actions to outcomes, 2) use a variety of source data (empirical or expert opinion), 3) explore robust management strategies under uncertainty, 4) use sensitivity analysis to identify opportunities for developing new management actions, and 5) guide the design of data collection for monitoring to improve management decisions. BBNs contribute considerably to environmental research and management because they are transparent and treat uncertainty explicitly. Because of the high level of uncertainty in stressor response, we developed a BBN to conceptually evaluate the effects of potential management actions on amphibian populations exposed to disease, environmental contaminants, and increasingly frequent and severe droughts

To better understand relations of annual calling phenophases for Pseudacris crucifer, and of the first calls of the season for Hyla chrysoscelis/versicolor, to the timing of the start of the calling season, we compared these dynamics for six wetlands in the St. Croix National Scenic Riverway from 2008 to 2012. We installed an acoustic recorder at each site prior to the start of each calling season and programmed it to record for five minutes at the top of every hour until late summer. We then used the Songscape option in Songscope software to generate annual summaries of all acoustic files recorded at each site. We created contour plots of the summarized median dB values across bandwidths in each recording and then assessed individual calls and calling peaks by visually examining these plots to identify first (and last) calls via the unique call signatures for these two species. We examined individual five-minute recordings aurally and visually as necessary when sound images represented on the contour plots were confounded and to ensure that the calling peaks described below were dates when calling activity was relatively intense. We also determined the daily median dB levels for frequencies across 2900 to 3200 Hz during 2100 to 2300 h, the bandwidth that typically encompassed the primary energy peak in P. crucifer calls and a time period during which P. crucifer typically called most consistently throughout their calling season. We did this for each day from the date when P. crucifer first called during each year to the date when they last called during each year. Because calling activity could vary from one hour to the next, we integrated the area under the curve for the daily median dB levels from 2900 to 3200 Hz during 2100 to 2300 h. We removed dates when overlapping sounds from storms or other sources rendered comparisons to calls of P. crucifer inaccurate. We used the resultant set of integrands to represent the relative sound intensity (as an indicator of calling activity) for P. crucifer across those hours for each date. We then used these integrands to determine the three highest peak calling dates for this species and used the median of those three dates as the overall median peak date for each site in each year.

To help determine when winter conditions were changing to spring conditions annually in our four study areas, we determined the first eight-day interval (in accordance with the scale limitations of satellite data we used to assess the presence of snow) during which the first amphibian of the season called at each of our study wetlands in those areas. To do this, we examined contour plots of summaries of all the acoustic data we collected at that site in a given year to identify the unique call signatures of individual amphibian species by date and time. When necessary due to potential confounding on a contour plot, we also examined relevant individual five-minute recordings aurally and visually to confirm whether a call occurred. When we confirmed the date of the first call we recorded in a given season, we identified the eight-day interval in which that date fell, with the first such interval beginning on January 1 of each year.

To relate water levels in our study wetlands to temperature, precipitation, wetland water depth, and amphibian calling activity, we installed one pressure logger in the deepest spot we could find in each wetland. Soon after thawing conditions allowed, we drove a plastic pipe (anchor pipe) into the sediments at the deepest location and secured another pipe to it that contained one pressure logger (Global Water Model 14 and 15 [College Station, TX, USA] or Onset Computer Corporation Model U20-001-04 [Bourne, MA, USA]) suspended approximately 2.5 cm above the sediments. We installed additional individual pressure loggers in the upper part of the logger pipes (in air) at select locations to measure barometric pressure for calibrating the submerged loggers’ readings. We measured pressure once per hour and used software supplied by the logger manufacturers to upload and convert data to depth at the end of each season.

To describe calling activity of Pseudacris crucifer in relation to temperature, precipitation, and wetland water levels, we programmed an acoustic recorder (Wildlife Acoustics) to sample seasonal amphibian calls remotely at study site SC4DAI2 in the St. Croix National Scenic Riverway from 2008 to 2012. We programmed the recorder to sample for five minutes at the top of every hour of every day from late winter/early spring through late summer. We used the Songscape option in Songscope software to generate annual summaries of all of our acoustic samples from SC4DAI2. These summaries included a median dB level for each prescribed frequency within each recording. Pseudacris crucifer, the spring peeper, inhabited SC4DAI2 and typically called over several weeks each year, depending upon weather conditions and surface-water availability. Most of the energy in their individual calls occurred between 2900 and 3200 Hz, which provided a unique acoustic signature compared with the other anurans that called from the site. We used this information as part of a case study to better understand how the daily calling activity of P. crucifer varied relative to air temperature, precipitation, and water depth at SC4DAI2 across years. We first determined the daily median dB levels for frequencies across 2900 to 3200 Hz during 2100 to 2300 h, a time period during which P. crucifer typically called throughout their calling season. We did this for each day from the date when P. crucifer first called each year to the date when they last called each year and considered any day in this range as one during which they potentially could call. Because calling activity could vary from one hour to the next, we integrated the area under the curve for the daily median dB levels from 2900 to 3200 Hz during 2100 to 2300 h. We removed dates when overlapping sounds from storms or other sources rendered comparisons to calls of P. crucifer inaccurate. We used the resultant set of integrands to represent the relative sound intensity (as an indicator of calling activity) for P. crucifer across those hours for each date. Those integrands are contained in this data set. These data enabled us to then compare daily integrand values with daily measurements of air temperature, precipitation totals, and water depth.

Assessing climate-related ecological changes across spatiotemporal scales meaningful to resource managers is challenging because no one method reliably produces essential data at both fine and broad scales. We recently confronted such challenges while integrating data from ground- and satellite-based sensors for an assessment of four wetland-rich study areas in the U.S. Midwest. We examined relations between temperature and precipitation and a set of variables measured on the ground at individual wetlands and another set measured via satellite sensors within surrounding 4 km2 landscape blocks. At the block scale, we used evapotranspiration and vegetation greenness as remotely sensed proxies for water availability and to estimate seasonal photosynthetic activity. We used sensors on the ground to coincidentally measure surface-water availability and amphibian calling activity at individual wetlands within blocks. Responses of landscape blocks generally paralleled changes in conditions measured on the ground, but the latter were more dynamic, and changes in ecological conditions on the ground that were critical for biota were not always apparent in measurements of related parameters in blocks. Here, we evaluate the effectiveness of decisions and assumptions we made in applying the remotely sensed data for the assessment and the value of integrating observations across scales, sensors, and disciplines.

Long-term, interdisciplinary studies of relations between climate and ecological conditions on wetland-upland landscapes have been lacking, especially studies integrated across scales meaningful for adaptive resource management. We collected data in situ at individual wetlands, and via satellite for surrounding 4-km2 landscape blocks, to assess relations between annual weather dynamics, snow duration, phenology, wetland surface-water availability, amphibian presence and calling activity, greenness, and evapotranspiration in four U.S. conservation areas from 2008 to 2012. Amid recent decades of relatively warm growing seasons, 2012 and 2010 were the first and second warmest seasons, respectively, dating back to 1895. Accordingly, we observed the earliest starts of springtime biological activity during those two years. In all years, early-season amphibians first called soon after daily mean air temperatures were ? 0°C and snow had mostly melted. Similarly, satellite-based indicators suggested seasonal leaf-out happened soon after snowmelt and temperature thresholds for plant growth had occurred. Daily fluctuations in weather and water levels were related to amphibian calling activity, including decoupling the timing of the onset of calling at the start of season from the onset of calling events later in the season. Within-season variation in temperature and precipitation also was related to vegetation greenness and evapotranspiration, but more at monthly and seasonal scales. Wetland water levels were moderately to strongly associated with precipitation and early or intermittent wetland drying likely reduced amphibian reproduction success in some years, even though Pseudacris crucifer occupied sites at consistently high levels. Notably, satellite-based indicators of landscape water availability did not suggest such consequential, intra-seasonal variability in wetland surface-water availability. Our cross-disciplinary data show how temperature and precipitation interacted to affect key ecological relations and outcomes on our study landscapes. These results demonstrate the value of multi-year studies and the importance of scale for understanding actual climate-related effects in these areas.

Understanding the distribution of pathogens across landscapes and within host populations is a common aim of wildlife managers. Despite the need for unbiased inferences about these parameters to plan effective management interventions, many researchers fail to account for imperfect pathogen detection and instead report only raw data, which may lead to improper management. We demonstrate analyses of ranavirus detection data in the Patuxent Research Refuge using an occupancy modeling approach, which yields unbiased estimates of pathogen occurrence and prevalence. To improve inference and reporting of pathogen and disease parameters, we describe our approach in-text and provide a written tutorial on how to fit these models using the freely available software Program PRESENCE. In our case study, ranavirus prevalence was underestimated up to 30% if imperfect detection was ignored. After accounting for imperfect detection, estimates of ranavirus prevalence in larval wood frogs (Lithobates sylvaticus) were higher than in larval spotted salamanders (Ambystoma maculatum). In addition, we found that the odds of detecting ranavirus in tail samples were 6.7 times higher than detecting ranavirus in liver samples. We discuss how this information can be used to design ranavirus surveillance studies. Our tutorial provides a clear guide for practitioners and researchers wishing to make unbiased inference in a variety of host-pathogen systems.

Anthropogenic climate change presents challenges and opportunities to the growth, reproduction, and survival of individuals throughout their life cycles. Demographic compensation among life-history stages has the potential to buffer populations from decline, but alternatively, compounding negative effects can lead to accelerated population decline and extinction. In montane ecosystems of the US Pacific Northwest, increasing temperatures are resulting in a transition from snow-dominated to rain-dominated precipitation events, reducing snowpack. For ectotherms such as amphibians, warmer winters can reduce the frequency of critical minimum temperatures and increase the length of summer growing seasons, benefiting post-metamorphic stages, but may also increase metabolic costs during winter months, which could decrease survival. Lower snowpack levels also result in wetlands that dry sooner or more frequently in the summer, increasing larval desiccation risk. To evaluate how these challenges and opportunities compound within a species? life history, we collected demographic data on Cascades frog (Rana cascadae) in Olympic National Park in Washington state to parameterize stage-based stochastic matrix population models under current and future (A1B, 2040s and 2080s) environmental conditions. We estimated the proportion of reproductive effort lost each year due to drying using watershed-specific hydrologic models, and coupled this with an analysis that relates 15-years of R. cascadae abundance data with a suite of climate variables. We estimated the current population growth (λs) to be https://0.98 (95% CI: 0.97-0.99), but predict that λs will decline under continued climate warming, resulting in a 62% chance of extinction by the 2080s because of compounding negative effects on early and late life history stages. By the 2080s, our models predict that larval mortality will increase by 17% as a result of increased pond drying, and adult survival will decrease by 7% as winter length and summer precipitation continue to decrease. We find that reduced larval survival drives initial declines in the 2040s, but further declines in the 2080s are compounded by decreases in adult survival. Our results demonstrate the need to understand the potential for compounding or compensatory effects within different life history stages to exacerbate or buffer the effects of climate change on population growth rates through time.

A research priority can be defined as a knowledge gap that, if resolved, identifies the optimal course of conservation action. We (a group of geographically distributed and multidisciplinary research scientists) used tools from nominal group theory and decision analysis to collaboratively identify and prioritize information needs within the context of disease-associated amphibian decline, in order to develop a strategy that would support US management agency needs. We developed iterated influence diagrams to create and assess a unified research strategy. We illustrated a transparent process for identifying specific knowledge gaps in amphibian disease ecology relevant to environmental management, and then constructed a research plan to address these uncertainties. The resulting priorities include a need to: (1) understand the drivers of the community-disease relationship, (2) determine the mechanisms by which exposure to contaminants influence disease outcomes, (3) identify elements of terrestrial and aquatic habitats that stabilize host-pathogen dynamics, (4) discuss how metapopulations may be managed to reduce the speed and intensity of disease outbreaks, and (5) define the relationship between habitat management and the environmental and host microbiomes. Along with identifying research priorities for disease management, we present the details of the process used to develop a consensus plan for addressing disease-related declines in amphibians on federally managed lands of the United States.

Occupancy models provide a reliable measure of species distributions while accounting for imperfect detectability. The cost of accounting for false absences is that occupancy surveys typically require repeated visits to a site or multiple-observer techniques. More efficient methods of estimating detection probabilities would allow more sites to be surveyed for the same effort, resulting in more information about the ecological processes leading to occupancy. Time-to-detection surveys allow the estimation of detection probability based on a single site visit by one observer, and therefore might be an efficient technique for herpetological occupancy studies. We evaluated the use of time-to-detection surveys to estimate the occupancy of pond-breeding amphibians at Point Reyes National Seashore, California, USA, including variables that affected detection rates and the probability of occurrence. We found that detection times were short enough and occupancy high enough to reliably estimate the probability of occurrence of three pond-breeding amphibians at Point Reyes National Seashore, and that survey and site conditions had species-specific effects on detection rates. In particular, relative abundance was negatively related to the time to initial detection of all species, and pond area was positively related to time to initial detection for Sierran Treefrogs (Hyliola sierra) and Rough-skinned Newts (Taricha granulosa). Rough-skinned newt time to initial detection also was affected by date, with lowest initial detection time in early summer. California Red-legged Frog (Rana draytonii) time to detection was lowest in ponds with a mean depth of 0.6 m, and higher in shallower and deeper ponds. Probability of occurrence of Sierran Treefrogs and Rough-skinned Newts was negatively related to the presence of fish and pond area. Rarely detected species required constraints on priors to fit time-to-detection models. Time-to-detection surveys can provide an efficient method of estimating detection probabilities and accounting for false absences in occupancy studies of reptiles and amphibians.