Evaluating the predictive abilities of community occupancy models using AUC while accounting for imperfect detection

Evaluation of contingency actions to control the spread of raccoon rabies in Ohio and Virginia

The raccoon (Procyon lotor) variant of the rabies virus (RRV) is enzootic in the eastern United States and oral rabies vaccination (ORV) is the primary strategy to prevent and control landscape spread. Breaches of ORV management zones occasionally occur, and emergency "contingency" actions may be implemented to enhance local control. Contingency actions are an integral part of landscape-scale wildlife rabies management but can be very costly and routinely involve enhanced rabies surveillance (ERS) around the index case. We investigated two contingency actions in Ohio (2017-2019 and 2018-2021) and one in Virginia (2017-2019) using a dynamic, multi-method occupancy approach to examine relationships between specific management actions and RRV occurrence, including whether ERS was sufficient around the index case. The RRV occupancy was assessed seasonally at 100-km2 grids and we examined relationships across three spatial scales (regional management zone, RRV free regions, and local contingency areas). The location of a grid relative to the ORV management zone was the strongest predictor of RRV occupancy at the regional scale. In RRV free regions, the neighbor effect and temporal variability were most important in influencing RRV occupancy. Parenteral (hand) vaccination of raccoons was important across all three contingency action areas, but more influential in the Ohio contingency action areas where more raccoons were hand vaccinated. In the Virginia contingency action area, ORV strategies were as important in reducing RRV occupancy as a hand vaccination strategy. The management action to trap, euthanize, and test (TET) raccoons was an important method to increase ERS, yet the impacts of TET on RRV occupancy are not clear. The probability of detecting additional cases of RRV was exceptionally high (>0.95) during the season the index case occurred. The probability of detecting RRV through ERS declined in the seasons following initial TET efforts but remained higher after the contingency action compared to the ERS detection probabilities prior to index case incidence. Local RRV cases were contained within one year and eliminated within 2-3 years of each contingency action.

Prospecting during egg laying informs incubation recess movements of eastern wild turkeys

BACKGROUND

Central place foragers must acquire resources and return to a central location after foraging bouts. During the egg laying (hereafter laying) period, females are constrained to a nest location, thus they must familiarize themselves with resources available within their incubation ranges after nest site selection. Use of prospecting behaviors by individuals to obtain knowledge and identify profitable (e.g., resource rich) locations on the landscape can impact demographic outcomes. As such, prospecting has been used to evaluate nest site quality both before and during the reproductive period for a variety of species.

METHODS

Using GPS data collected from female eastern wild turkeys (Meleagris gallopavo silvestris) across the southeastern United States, we evaluated if prospecting behaviors were occurring during laying and what landcover factors influenced prospecting. Specifically, we quantified areas prospected during the laying period using a cluster analysis and the return frequency (e.g., recess movements) to clustered laying patches (150-m diameter buffer around a clustered laying period location) during the incubation period.

RESULTS

The average proportion of recess movements to prospected locations was 56.9%. Nest fate was positively influenced (μ of posterior distribution with 95% credible 0.19, 0.06-0.37, probability of direction = 99.8%) by the number of patches (90-m diameter buffer around a clustered laying period location) a female visited during incubation recesses. Females selected for areas closer to the nest site, secondary roads, hardwood forest, mixed pine-hardwood forest, water, and shrub/scrub, whereas they avoided pine forest and open-treeless areas.

CONCLUSIONS

Our findings suggest that having a diverse suite of clustered laying patches to support incubation recesses is impactful to nest fate. As such, local conditions within prospected locations during incubation may be key to successful reproductive output by wild turkeys. We suggest that prospecting could be important to other phenological periods. Furthermore, future research should evaluate how prospecting for brood-rearing locations may occur before or during the incubation period.

King Rail (Rallus elegans) presence in the Midwestern United States is predicted by local-scale factors and avian community

The King Rail (Rallus elegans) is a wetland dependent species of conservation concern. Our objective was to gain a better understanding of the breeding habitat associations of King Rails in the Midwestern United States and the relationship of this species to other obligate marsh birds using occupancy and MaxEnt models. To collect data pertaining to occupancy, we placed trail cameras at 50 random points in coastal wetlands in the western Lake Erie basin where calls of King Rails were continuously broadcast at night. Data pertaining to other marsh bird species were collected via call-broadcast surveys and camera surveys at each sample point. For MaxEnt modeling, we obtained presence data for King Rails and other obligate marsh birds from eBird and habitat data from GIS databases. Trail cameras and call-broadcast surveys captured 10 detections of King Rails at nine sites, an 18% naive occupancy rate. King Rail occupancy was positively related to amount of interspersion, average water depth, and percent cover of emergent vegetation at local scales within a 5-m radius. Our MaxEnt models indicated that, at a broader scale, the presence of other rail species such as the Sora (Porzana carolina) may be more important for predicting King Rail presence than other marsh birds or coarse wetland categories such as "emergent vegetation." Our results could help wetland managers to predict where King Rails occur and to adapt management plans to incorporate King Rail conservation.

Habitat selection of Gould's wild turkeys in southeastern Arizona

In semi-arid environments, resources necessary for survival may be unevenly distributed across the landscape. Gould's wild turkeys (Meleagris gallopavo mexicana) are spatially restricted to mountainous semi-arid areas of southwestern United States and Mexico, and information on their distribution and habitat use is limited. We described how landcover type and topographical features influenced space use and habitat selection by Gould's wild turkeys in southeastern Arizona. We used GPS data from 51 Gould's wild turkeys to describe resource selection during 2016-2017 in southeastern Arizona, USA. We estimated home ranges and calculated resource selection functions using distance from landcover types, slope, aspect, and elevation at used locations and random locations within individual home ranges. Gould's wild turkeys selected areas closer to pine forest and water. Likewise, Gould's wild turkeys selected locations with moderate elevations of 1641 ± 235 m (range = 1223-2971 m), and on north and west facing slopes with a 10° ± 8.5 (range = 0.0-67.4°) incline. Our findings suggest that conserving portions of the landscape with appropriate topography and landcover types as described above will promote habitat availability for Gould's wild turkeys.

Behavioral-dependent recursive movements and implications for resource selection

Within home ranges, animals repeatedly visit certain areas. Recursive movement patterns are widespread throughout the animal kingdom, but are rarely considered when developing resource selection models. We examined how behavioral state-dependent recursive movements influenced reource selection of eastern wild turkey (Meleagris gallopavo silvestris) broods as they aged from day 1 to 28. Because broods become more plastic in behaviors once they begin roosting off the ground, we separated data into broods that were ground roosting (1-13 days) and tree roosting (14-28 days). We used Hidden Markov Models to identify 2 behavioral states (restricted and mobile). We extracted state-specific recursive movements based on states and specific step lengths, which we integrated into a step selection analysis to evaluate resource selection. We found that in a restricted state, ground roosting broods spent less time in areas of mixed pine-hardwoods and more time in areas with greater vegetation density. Tree roosting broods revisited areas closer to shrub/scrub landcover types, and areas with greater vegetation density. Tree roosting broods also spent less time near mixed pine-hardwoods, while spending more time in areas with greater vegetation density. We found that in a mobile state, ground roosting broods revisited areas closer to secondary roads and mixed pine-hardwoods, but farther from hardwoods. Tree roosting broods revisited areas farther from secondary roads and with greater vegetation density. Tree roosting broods also spent more time in areas closer to pine. Resource selection varied depending on behavioral state and recursive movements. However, revisitation and residence time impacted selection in both ground and tree roosting broods. Our findings highlight the need to consider how behaviors can influence movement decisions and ultimately resource selection.

A comparison of monitoring designs to assess wildlife community parameters across spatial scales

Dedicated long-term monitoring at appropriate spatial and temporal scales is necessary to understand biodiversity losses and develop effective conservation plans. Wildlife monitoring is often achieved by obtaining data at a combination of spatial scales, ranging from local to broad, to understand the status, trends, and drivers of individual species or whole communities and their dynamics. However, limited resources for monitoring necessitates tradeoffs in the scope and scale of data collection. Careful consideration of the spatial and temporal allocation of finite sampling effort is crucial for monitoring programs that span multiple spatial scales. Here we evaluate the ability of five monitoring designs-stratified random, weighted effort, indicator unit, rotating panel, and split panel-to recover parameter values that describe the status (occupancy), trends (change in occupancy), and drivers (spatially varying covariate and an autologistic term) of wildlife communities at two spatial scales. Using an amphibian monitoring program that spans a network of US national parks as a motivating example, we conducted a simulation study for a regional community occupancy sampling program to compare the monitoring designs across varying levels of sampling effort (ranging from 10% to 50%). We found that the stratified random design outperformed the other designs for most parameters of interest at both scales and was thus generally preferable in balancing the estimation of status, trends, and drivers across scales. However, we found that other designs had improved performance in specific situations. For example, the rotating panel design performed best at estimating spatial drivers at a regional level. Thus, our results highlight the nuanced scenarios in which various design strategies may be preferred and offer guidance as to how managers can balance common tradeoffs in large-scale and long-term monitoring programs in terms of the specific knowledge gained. Monitoring designs that improve accuracy in parameter estimates are needed to guide conservation policy and management decisions in the face of broad-scale environmental challenges, but the preferred design is sensitive to the specific objectives of a monitoring program.

Cool microrefugia accumulate and conserve biodiversity under climate change

A major challenge in climate change biology is to explain why the impacts of climate change vary around the globe. Microclimates could explain some of this variation, but climate change biologists often overlook microclimates because they are difficult to map. Here, we map microclimates in a freshwater rock pool ecosystem and evaluate how accounting for microclimates alters predictions of climate change impacts on aquatic invertebrates. We demonstrate that average maximum temperature during the growing season can differ by 9.9-11.6°C among microclimates less than a meter apart and this microclimate variation might increase by 21% in the future if deeper pools warm less than shallower pools. Accounting for this microclimate variation significantly alters predictions of climate change impacts on aquatic invertebrates. Predictions that exclude microclimates predict low future occupancy (0.08-0.32) and persistence probabilities (2%-73%) for cold-adapted taxa, and therefore predict decreases in gamma richness and a substantial shift toward warm-adapted taxa in local communities (i.e., thermophilization). However, predictions incorporating microclimates suggest cool locations will remain suitable for cold-adapted taxa in the future, no change in gamma richness, and 825% less thermophilization. Our models also suggest that cool locations will become suitable for warm-adapted taxa and will therefore accumulate biodiversity in the future, which makes cool locations essential for biodiversity conservation. Simulated protection of the 10 coolest microclimates (9% of locations on the landscape) results in a 100% chance of conserving all focal taxa in the future. In contrast, protecting the 10 currently most biodiverse locations, a commonly employed conservation strategy, results in a 3% chance of conserving all focal taxa in the future. Our study suggests that we must account for microclimates if we hope to understand the future impacts of climate change and design effective conservation strategies to limit biodiversity loss.

Climatic factors influence the spread of COVID-19 in Russia

The study is the first attempt to assess the role of climatic predictors in the rise of COVID-19 intensity in the Russian climatic region. The study used the Random Forest algorithm to understand the underlying associations and monthly scenarios. The results show that temperature seasonality (29.2 ± 0.9%) has the highest contribution for COVID-19 transmission in the humid continental region. In comparison, the diurnal temperature range (26.8 ± 0.4%) and temperature seasonality (14.6 ± 0.8%) had the highest impacts in the sub-arctic region. Our results also show that September and October have favorable climatic conditions for the COVID-19 spread in the sub-arctic and humid continental regions, respectively. From June to August, the high favorable zone for the spread of the disease will shift towards the sub-arctic region from the humid continental region. The study suggests that the government should implement strict measures for these months to prevent the second wave of COVID-19 outbreak in Russia.

Identifying mismatches between conservation area networks and vulnerable populations using spatial randomization

Grassland birds are among the most globally threatened bird groups due to substantial degradation of native grassland habitats. However, the current network of grassland conservation areas may not be adequate for halting population declines and biodiversity loss. Here, we evaluate a network of grassland conservation areas within Wisconsin, U.S.A., that includes both large Focal Landscapes and smaller targeted conservation areas (e.g., Grassland Bird Conservation Areas, GBCAs) established within them. To date, this conservation network has lacked baseline information to assess whether the current placement of these conservation areas aligns with population hot spots of grassland-dependent taxa. To do so, we fitted data from thousands of avian point-count surveys collected by citizen scientists as part of Wisconsin's Breeding Bird Atlas II with multinomial N-mixture models to estimate habitat-abundance relationships, develop spatially explicit predictions of abundance, and establish ecological baselines within priority conservation areas for a suite of obligate grassland songbirds. Next, we developed spatial randomization tests to evaluate the placement of this conservation network relative to randomly placed conservation networks. Overall, less than 20% of species statewide populations were found within the current grassland conservation network. Spatial tests demonstrated a high representation of this bird assemblage within the entire conservation network, but with a bias toward birds associated with moderately tallgrasses relative to those associated with shortgrasses or tallgrasses. We also found that GBCAs had higher representation at Focal Landscape rather than statewide scales. Here, we demonstrated how combining citizen science data with hierarchical modeling is a powerful tool for estimating ecological baselines and conducting large-scale evaluations of an existing conservation network for multiple grassland birds. Our flexible spatial randomization approach offers the potential to be applied to other protected area networks and serves as a complementary tool for conservation planning efforts globally.

Landscape Enhancements in Apple Orchards: Higher Bumble Bee Queen Species Richness, but No Effect on Apple Quality

Simple Summary

Pollinators are essential to produce fruits in apple production. Bumble bees are among the most effective pollinators in orchards during the blooming season, yet they are often threatened by the high levels of pesticide use in apple production. Hedgerows and flower strips are infrequently sprayed by pesticides and are thus potentially good shelter for bumble bees. This study evaluated the influence of landscaping in the form of hedgerows and flower strips on the abundance and number of bumble bee species found in apple orchards. The number of bumble bee species found in orchards with hedgerows or flower strips was higher than in orchards without such landscape enhancements. Similarly, three species were more abundant in orchards with landscaping than orchards without those enhancements. Our work provides additional evidence that landscaping in the form of hedgerows and/or flower strips improves bumble bee presence in apple orchards and should therefore be considered as a means to enhance and ensure pollination within farms.

Abstract

Bumble bees are among the most effective pollinators in orchards during the blooming period, yet they are often threatened by the high levels of pesticide use in apple production. This study aimed to evaluate the influence of landscape enhancements (e.g., hedgerows, flower strips) on bumble bee queens in apple orchards. Bumble bee queens from 12 orchards in southern Québec (Canada) were marked, released, and recaptured in the springs and falls of 2017 to 2019. Half of the 12 orchards had landscape enhancements. Apples were harvested in 2018 and 2019 to compare their quality (weight, diameter, sugar level, and seed number) in sites with and without landscape enhancements. Species richness, as well as the occurrence of three species out of eight, was higher in orchards with landscape enhancements than in orchards without such structures. The occurrence of Bombus ternarius was lower in orchards with high levels of pesticide use. Apples had fewer seeds when collected in orchards with landscape enhancements and were heavier in orchards that used more pesticides. Our work provides additional evidence that landscape enhancements improve bumble bee presence in apple orchards and should therefore be considered as a means to enhance pollination within farms.

Behavioral state resource selection in invasive wild pigs in the Southeastern United States

Elucidating correlations between wild pig (Sus scrofa) behavior and landscape attributes can aid in the advancement of management strategies for controlling populations. Using GPS data from 49 wild pigs in the southeastern U.S., we used hidden Markov models to define movement path characteristics and assign behaviors (e.g., resting, foraging, travelling). We then explored the connection between these behaviors and resource selection for both sexes between two distinct seasons based on forage availability (i.e., low forage, high forage). Females demonstrated a crepuscular activity pattern in the high-forage season and a variable pattern in the low-forage season, while males exhibited nocturnal activity patterns across both seasons. Wild pigs selected for bottomland hardwoods and dense canopy cover in all behavioral states in both seasons. Males selected for diversity in vegetation types while foraging in the low-forage season compared to the high-forage season and demonstrated an increased use of linear anthropogenic features across seasons while traveling. Wild pigs can establish populations and home ranges in an array of landscapes, but our results demonstrate male and female pigs exhibit clear differences in movement behavior and there are key resources associated with common behaviors that can be targeted to improve the efficiency of management programs.

Joint species distribution models of Everglades wading birds to inform restoration planning

Restoration of the Florida Everglades, a substantial wetland ecosystem within the United States, is one of the largest ongoing restoration projects in the world. Decision-makers and managers within the Everglades ecosystem rely on ecological models forecasting indicator wildlife response to changes in the management of water flows within the system. One such indicator of ecosystem health, the presence of wading bird communities on the landscape, is currently assessed using three species distribution models that assume perfect detection and report output on different scales that are challenging to compare against one another. We sought to use current advancements in species distribution modeling to improve models of Everglades wading bird distribution. Using a joint species distribution model that accounted for imperfect detection, we modeled the presence of nine species of wading bird simultaneously in response to annual hydrologic conditions and landscape characteristics within the Everglades system. Our resulting model improved upon the previous model in three key ways: 1) the model predicts probability of occupancy for the nine species on a scale of 0-1, making the output more intuitive and easily comparable for managers and decision-makers that must consider the responses of several species simultaneously; 2) through joint species modeling, we were able to consider rarer species within the modeling that otherwise are detected in too few numbers to fit as individual models; and 3) the model explicitly allows detection probability of species to be less than 1 which can reduce bias in the site occupancy estimates. These improvements are essential as Everglades restoration continues and managers require models that consider the impacts of water management on key indicator wildlife such as the wading bird community.

Using machine learning algorithms to map the groundwater recharge potential zones

Groundwater recharge is indispensable for the sustainable management of freshwater resources, especially in the arid regions. Here we address some of the important aspects of groundwater recharge through machine learning algorithms (MLAs). Three MLAs including, SVM, MARS, and RF were validated for higher prediction accuracies in generating groundwater recharge potential maps (GRPMs). Accordingly, soil permeability samples were prepared and are arbitrarily grouped into training (70%) and validation (30%) samples. The GRPMs are generated using sixteen effective factors, such as elevation (denoted using a digital elevation model; DEM), aspect, slope angle, TWI (topographic wetness index), fault density, MRVBF (multiresolution index of valley bottom flatness), rainfall, lithology, land use, drainage density, distance from rivers, distance from faults, annual ETP (evapo-transpiration), minimum temperature, maximum temperature, and rainfall 24-hr. Subsequently, the VI (variables importance) is assessed based on the LASSO algorithm. The GRPMs of three MLAs were validated using the ROC-AUC (receiver operating characteristic-area under curve) and various techniques including true positive rate (TPR), false positive rate (FPR), F-measures, fallout, sensitivity, specificity, true skill statistics (TSS), and corrected classified instances (CCI). Based on the validation, the RF algorithm performed better (AUC = 0.987) than the SVM (AUC = 0.963) and the MARS algorithm (AUC = 0.962). Furthermore, the accuracy of these MLAs are included in excellent class, based on the ROC curve threshold. Our case study shows that the GRPMs are potential guidelines for decision-makers in drafting policies related to the sustainable management of the groundwater resources.

When parks work: Effect of anthropogenic disturbance on occupancy of tropical forest mammals

Protected areas (PAs) in the tropics are vulnerable to human encroachment, and, despite formal protection, they do not fully mitigate anthropogenic threats to habitats and biodiversity. However, attempts to quantify the effectiveness of PAs and to understand the status and changes of wildlife populations in relation to protection efficiency remain limited. Here, we used camera-trapping data collected over 8 consecutive years (2009-2016) to investigate the yearly occurrences of medium-to-large mammals within the Udzungwa Mountains National Park (Tanzania), an area of outstanding importance for biological endemism and conservation. Specifically, we evaluated the effects of habitat and proxies of human disturbance, namely illegal hunting with snares and firewood collection (a practice that was banned in 2011 in the park), on species' occurrence probabilities. Our results showed variability in species' responses to disturbance: The only species that showed a negative effect of the number of snares found on occurrence probability was the Harvey's duiker, a relatively widespread forest antelope. Similarly, we found a moderate positive effect of the firewood collection ban on only the suni, another common antelope, and a negative effect on a large opportunistic rodent, the giant-pouched rat. Importantly, we found evidence of temporal stability in occurrence probability for all species over the 8-year study period. Our findings suggest that well-managed PAs can sustain mammal populations in tropical forests. However, variability among species in their responses to anthropogenic disturbance necessitates consideration in the design of conservation action plans for multiple taxa.

Small mammal community composition varies among Ozark glades

Island biogeography theory (IBT) explains and estimates large-scale ecological patterns among islands and isolated habitat patches. Specifically, IBT predicts that the number of species per habitat patch differs as a function of area and isolation as a result of local colonization and extinction. Accurate estimates of species richness are essential for testing predictions of IBT, but differences in detectability of species can lead to bias in empirical data. Hierarchical community models correct for imperfect detection by leveraging information from across the community to estimate species-specific occupancy and detection probabilities. Using the fragmented Ozark glades as our model system, we constructed a hierarchical community model to 1) estimate site-level and regional species richness of small mammals while correcting for detection error, and 2) determine environmental covariates driving occupancy. We sampled 16 glades in southwestern Missouri in summer 2016–2017 and quantified mammal community structure within the glade network. The detected species pool included eight species, and the model yielded a regional species estimate of 8.6 species, with a mean of 3.47 species per glade. Species richness increased with patch area but not isolation, and effects of patch shape varied between species in the community.

Rabies Surveillance Identifies Potential Risk Corridors and Enables Management Evaluation

Intensive efforts are being made to eliminate the raccoon variant of rabies virus (RABV) from the eastern United States and Canada. The United States Department of Agriculture (USDA) Wildlife Services National Rabies Management Program has implemented enhanced rabies surveillance (ERS) to improve case detection across the extent of the raccoon oral rabies vaccination (ORV) management area. We evaluated ERS and public health surveillance data from 2006 to 2017 in three northeastern USA states using a dynamic occupancy modeling approach. Our objectives were to examine potential risk corridors for RABV incursion from the U.S. into Canada, evaluate the effectiveness of ORV management strategies, and identify surveillance gaps. ORV management has resulted in a decrease in RABV cases over time within vaccination zones (from occupancy ( ψ ¯ ) of 0.60 standard error (SE) = 0.03 in the spring of 2006 to ψ ¯ of 0.33 SE = 0.10 in the spring 2017). RABV cases also reduced in the enzootic area (from ψ ¯ of 0.60 SE = 0.03 in the spring of 2006 to ψ ¯ of 0.45 SE = 0.05 in the spring 2017). Although RABV occurrence was related to habitat type, greater impacts were associated with ORV and trap-vaccinate-release (TVR) campaigns, in addition to seasonal and yearly trends. Reductions in RABV occupancy were more pronounced in areas treated with Ontario Rabies Vaccine Bait (ONRAB) compared to RABORAL V-RG^®. Our approach tracked changes in RABV occurrence across space and time, identified risk corridors for potential incursions into Canada, and highlighted surveillance gaps, while evaluating the impacts of management actions. Using this approach, we are able to provide guidance for future RABV management.

Evidence of region-wide bat population decline from long-term monitoring and Bayesian occupancy models with empirically informed priors

Strategic conservation efforts for cryptic species, especially bats, are hindered by limited understanding of distribution and population trends. Integrating long-term encounter surveys with multi-season occupancy models provides a solution whereby inferences about changing occupancy probabilities and latent changes in abundance can be supported. When harnessed to a Bayesian inferential paradigm, this modeling framework offers flexibility for conservation programs that need to update prior model-based understanding about at-risk species with new data. This scenario is exemplified by a bat monitoring program in the Pacific Northwestern United States in which results from 8 years of surveys from 2003 to 2010 require updating with new data from 2016 to 2018. The new data were collected after the arrival of bat white-nose syndrome and expansion of wind power generation, stressors expected to cause population declines in at least two vulnerable species, little brown bat (Myotis lucifugus) and the hoary bat (Lasiurus cinereus). We used multi-season occupancy models with empirically informed prior distributions drawn from previous occupancy results (2003-2010) to assess evidence of contemporary decline in these two species. Empirically informed priors provided the bridge across the two monitoring periods and increased precision of parameter posterior distributions, but did not alter inferences relative to use of vague priors. We found evidence of region-wide summertime decline for the hoary bat ( λ ^  = 0.86 ± 0.10) since 2010, but no evidence of decline for the little brown bat ( λ ^  = 1.1 ± 0.10). White-nose syndrome was documented in the region in 2016 and may not yet have caused regional impact to the little brown bat. However, our discovery of hoary bat decline is consistent with the hypothesis that the longer duration and greater geographic extent of the wind energy stressor (collision and barotrauma) have impacted the species. These hypotheses can be evaluated and updated over time within our framework of pre-post impact monitoring and modeling. Our approach provides the foundation for a strategic evidence-based conservation system and contributes to a growing preponderance of evidence from multiple lines of inquiry that bat species are declining.

Identifying occupancy model inadequacies: can residuals separately assess detection and presence?

Occupancy models are widely applied to estimate species distributions, but few methods exist for model checking. Thorough model assessments can uncover inadequacies and allow for deeper ecological insight by exploring structure in the observed data not accounted for by a model. We introduce occupancy model residual definitions that utilize the posterior distribution of the partially latent occupancy states. Residual-based assessments are valuable because they can target specific assumptions and identify ways to improve a model, such as adding spatial correlation or meaningful covariates. Our approach defines separate residuals for occupancy and detection, and we use simulation to examine whether missing structure for modeling detection probabilities can be distinguished from that for occupancy probabilities. In many scenarios, our residual diagnostics were able to separate inadequacies at the different model levels successfully, but we describe other situations when this may not be the case. Applying Moran's I residual diagnostics to assess models for silver-haired (Lasionycteris noctivagans) and little brown (Myotis lucifugus) bats only provided evidence of residual spatial correlation among detections. Targeting specific model assumptions using carefully chosen residual diagnostics is valuable for any analysis, and we remove previous barriers for occupancy analyses-lack of examples and practical advice.

Multistate occupancy modeling improves understanding of amphibian breeding dynamics in the Greater Yellowstone Area

Discerning the determinants of species occurrence across landscapes is fundamental to their conservation and management. In spatially and climatologically complex landscapes, explaining the dynamics of occurrence can lead to improved understanding of short- vs. long-term trends and offer novel insight on local vs. regional change. We examined the changes in occupancy for two species of anurans with different life histories over a decade using hundreds of wetland sites in Yellowstone and Grand Teton National Parks. To account for the joint dynamics of wetland drying and amphibian breeding, we adopted a multistate occupancy model as a means to investigate mechanistic relationships of observed occurrence patterns with climatological drivers of wetland hydrologic variability. This approach allowed us to decompose occupancy dynamics into habitat changes caused by wetland drying and amphibian breeding activity, conditional on available water and previous breeding state. Over our 10-yr time series, we observed considerable variability in climate drivers and the proportion of dry wetlands. Boreal chorus frogs (Pseudacris maculata) were more responsive to changes in wetland inundation status than Columbia spotted frogs (Rana luteiventris), as indicated by higher breeding colonization probabilities under favorable (wet) conditions. Both species had high probabilities of breeding persistence in permanently inundated wetlands with prior breeding. Despite the absence of multi-year drought in our time series, mechanistic relationships described here offer insights on how future climate variation may result in reduced and/or shifted occurrence patterns for pond-breeding anurans in the Greater Yellowstone Area. Further, our modeling approach may prove valuable in evaluating determinants of occurrence for other species that are dependent on wetlands or other dynamic habitats.

Variation in regional and landscape effects on occupancy of temperate bats in the southeastern U.S

Habitat loss, wind energy development, and the disease white-nose syndrome are major threats contributing to declines in bat populations in North America. In the southeastern US in particular, the recent arrival of white-nose syndrome and changes in landscape composition and configuration have driven shifts in bat species populations and distributions. Effective management strategies which address these large-scale, community-level threats require landscape-scale analyses. Our objective was to model the relationship between ecoregional and landscape factors and occupancy by all bat species in South Carolina, USA, during summer. We conducted acoustic surveys from mid-May through July 2015 and 2016 and evaluated temporally dynamic occupancy models for eight bat species or species groups at the 100 km2 level. We found significant effects of landscape factors such as ecoregion and forest edge density for three species, but habitat condition effects were not statistically significant for five other species. Thus, for some species, site-use analyses may be more appropriate than larger scale occupancy analyses. However, our occupancy predictions generally matched statewide historical distributions for all species, suggesting our approach could be useful for monitoring landscape-level trends in bat species. Thus, while our scale of study was likely too coarse for assessing fine-scale habitat associations for all bat species, our findings can improve future monitoring efforts, inform conservation priorities, and guide subsequent landscape-scale studies for bat species and community-level responses to global change.

Effects of agriculture and topography on tropical amphibian species and communities

Habitat loss is the greatest threat to the persistence of forest-dependent amphibians, but it is not the only factor influencing species occurrences. The composition of the surrounding matrix, structure of stream networks, and topography are also important landscape characteristics influencing amphibian distributions. Tropical forests have high diversity and endemism of amphibians, but little is known about the specific responses of many of these species to landscape features. In this paper, we quantify the response of amphibian species and communities to landscape-scale characteristics in streams within the fragmented Brazilian Atlantic Forest. We surveyed amphibian communities during a rainy season in 50 independent stream segments using Standardized Acoustic and Visual Transect Sampling (active) and Automated Acoustic Recorders (passive) methods. We developed a hierarchical multi-species occupancy model to quantify the influence of landscape-scale characteristics (forest cover, agriculture, catchment area, stream density, and slope) on amphibian occurrence probabilities while accounting for imperfect detection of species using the two survey methods. At the community level, we estimated an overall mean positive relationship between amphibian occurrence probabilities and forest cover, and a negative relationship with agriculture. Catchment area and slope were negatively related with amphibian community structure (95% credible interval [CI] did not overlap zero). The species-level relationships with landscape covariates were highly variable but showed similar patterns to those at the community level. Species detection probabilities varied widely and were influenced by the sampling method. For most species, the active method resulted in higher detection probabilities than the passive approach. Our findings suggest that small streams and flat topography lead to higher amphibian occurrence probabilities for many species in Brazil's Atlantic Forest. Our results combined with land use and topographic maps can be used to make predictions of amphibian occurrences and distributions beyond our study area. Such projections can be useful to determine where to conduct future research and prioritize conservation efforts in human-modified landscapes.

Host-pathogen metapopulation dynamics suggest high elevation refugia for boreal toads

Emerging infectious diseases are an increasingly common threat to wildlife. Chytridiomycosis, caused by the fungal pathogen Batrachochytrium dendrobatidis (Bd), is an emerging infectious disease that has been linked to amphibian declines around the world. Few studies exist that explore amphibian-Bd dynamics at the landscape scale, limiting our ability to identify which factors are associated with variation in population susceptibility and to develop effective in situ disease management. Declines of boreal toads (Anaxyrus boreas boreas) in the southern Rocky Mountains are largely attributed to chytridiomycosis but variation exists in local extinction of boreal toads across this metapopulation. Using a large-scale historic data set, we explored several potential factors influencing disease dynamics in the boreal toad-Bd system: geographic isolation of populations, amphibian community richness, elevational differences, and habitat permanence. We found evidence that boreal toad extinction risk was lowest at high elevations where temperatures may be suboptimal for Bd growth and where small boreal toad populations may be below the threshold needed for efficient pathogen transmission. In addition, boreal toads were more likely to recolonize high elevation sites after local extinction, again suggesting that high elevations may provide refuge from disease for boreal toads. We illustrate a modeling framework that will be useful to natural resource managers striving to make decisions in amphibian-Bd systems. Our data suggest that in the southern Rocky Mountains high elevation sites should be prioritized for conservation initiatives like reintroductions.

Parasite metacommunities: Evaluating the roles of host community composition and environmental gradients in structuring symbiont communities within amphibians

Ecologists increasingly report the structures of metacommunities for free-living species, yet far less is known about the composition of symbiont communities through space and time. Understanding the drivers of symbiont community patterns has implications ranging from emerging infectious disease to managing host microbiomes. Using symbiont communities from amphibian hosts sampled from wetlands of California, USA, we quantified the effects of spatial structure, habitat filtering and host community components on symbiont occupancy and overall metacommunity structure. We built upon a statistical method to describe metacommunity structure that accounts for imperfect detection in survey data-detection error-corrected elements of metacommunity structure-by adding an analysis to identify covariates of community turnover. We applied our model to a metacommunity of eight parasite taxa observed in 3,571 Pacific chorus frogs (Pseudacris regilla) surveyed from 174 wetlands over 5 years. Symbiont metacommunity structure varied across years, showing nested structure in 3 years and random structure in 2 years. Species turnover was most consistently influenced by spatial and host community components. Occupancy generally increased in more southeastern wetlands, and snail (intermediate host) community composition had strong effects on most symbiont taxa. We have used sophisticated but accessible statistical methods to reveal that spatial components-which influence colonization-and host community composition-which mediates transmission-both drive symbiont community composition in this system. These methods allow us to associate broad patterns of community turnover to local, species-level effects, ultimately improving our understanding of spatial community dynamics.

A near half-century of temporal change in different facets of avian diversity

Assessments of spatial patterns of biodiversity change are essential to detect a signature of anthropogenic impacts, inform monitoring and conservation programs, and evaluate implications of biodiversity loss to humans. While taxonomic diversity (TD) is the most commonly assessed attribute of biodiversity, it misses the potential functional or phylogenetic implications of species losses or gains for ecosystems. Functional diversity (FD) and phylogenetic diversity (PD) are able to capture these important trait-based and phylogenetic attributes of species, but their changes have to date only been evaluated over limited spatial and temporal extents. Employing a novel framework for addressing detectability, we here comprehensively assess a near half-century of changes in local TD, FD, and PD of breeding birds across much of North America to examine levels of congruency in changes among these biodiversity facets and their variation across spatial and environmental gradients. Time-series analysis showed significant and continuous increases in all three biodiversity attributes until ca. 2000, followed by a slow decline since. Comparison of avian diversity at the beginning and end of the temporal series revealed net increase in TD, FD, and PD, but changes in TD were larger than those in FD and PD, suggesting increasing biotic homogenization of avian assemblages throughout the United States. Changes were greatest at high elevations and latitudes - consistent with purported effects of ongoing climate change on biodiversity. Our findings highlight the potential of combining new types of data with novel statistical models to enable a more integrative monitoring and assessment of the multiple facets of biodiversity.

Modeling stream fish distributions using interval-censored detection times

Controlling for imperfect detection is important for developing species distribution models (SDMs). Occupancy-detection models based on the time needed to detect a species can be used to address this problem, but this is hindered when times to detection are not known precisely. Here, we extend the time-to-detection model to deal with detections recorded in time intervals and illustrate the method using a case study on stream fish distribution modeling. We collected electrofishing samples of six fish species across a Mediterranean watershed in Northeast Portugal. Based on a Bayesian hierarchical framework, we modeled the probability of water presence in stream channels, and the probability of species occupancy conditional on water presence, in relation to environmental and spatial variables. We also modeled time-to-first detection conditional on occupancy in relation to local factors, using modified interval-censored exponential survival models. Posterior distributions of occupancy probabilities derived from the models were used to produce species distribution maps. Simulations indicated that the modified time-to-detection model provided unbiased parameter estimates despite interval-censoring. There was a tendency for spatial variation in detection rates to be primarily influenced by depth and, to a lesser extent, stream width. Species occupancies were consistently affected by stream order, elevation, and annual precipitation. Bayesian P-values and AUCs indicated that all models had adequate fit and high discrimination ability, respectively. Mapping of predicted occupancy probabilities showed widespread distribution by most species, but uncertainty was generally higher in tributaries and upper reaches. The interval-censored time-to-detection model provides a practical solution to model occupancy-detection when detections are recorded in time intervals. This modeling framework is useful for developing SDMs while controlling for variation in detection rates, as it uses simple data that can be readily collected by field ecologists.

A goodness-of-fit test for occupancy models with correlated within-season revisits

Occupancy modeling is important for exploring species distribution patterns and for conservation monitoring. Within this framework, explicit attention is given to species detection probabilities estimated from replicate surveys to sample units. A central assumption is that replicate surveys are independent Bernoulli trials, but this assumption becomes untenable when ecologists serially deploy remote cameras and acoustic recording devices over days and weeks to survey rare and elusive animals. Proposed solutions involve modifying the detection‐level component of the model (e.g., first‐order Markov covariate). Evaluating whether a model sufficiently accounts for correlation is imperative, but clear guidance for practitioners is lacking. Currently, an omnibus goodness‐of‐fit test using a chi‐square discrepancy measure on unique detection histories is available for occupancy models (MacKenzie and Bailey, Journal of Agricultural, Biological, and Environmental Statistics, 9, 2004, 300; hereafter, MacKenzie–Bailey test). We propose a join count summary measure adapted from spatial statistics to directly assess correlation after fitting a model. We motivate our work with a dataset of multinight bat call recordings from a pilot study for the North American Bat Monitoring Program. We found in simulations that our join count test was more reliable than the MacKenzie–Bailey test for detecting inadequacy of a model that assumed independence, particularly when serial correlation was low to moderate. A model that included a Markov‐structured detection‐level covariate produced unbiased occupancy estimates except in the presence of strong serial correlation and a revisit design consisting only of temporal replicates. When applied to two common bat species, our approach illustrates that sophisticated models do not guarantee adequate fit to real data, underscoring the importance of model assessment. Our join count test provides a widely applicable goodness‐of‐fit test and specifically evaluates occupancy model lack of fit related to correlation among detections within a sample unit. Our diagnostic tool is available for practitioners that serially deploy survey equipment as a way to achieve cost savings.

GIS-based groundwater potential mapping using boosted regression tree, classification and regression tree, and random forest machine learning models in Iran

Groundwater is considered one of the most valuable fresh water resources. The main objective of this study was to produce groundwater spring potential maps in the Koohrang Watershed, Chaharmahal-e-Bakhtiari Province, Iran, using three machine learning models: boosted regression tree (BRT), classification and regression tree (CART), and random forest (RF). Thirteen hydrological-geological-physiographical (HGP) factors that influence locations of springs were considered in this research. These factors include slope degree, slope aspect, altitude, topographic wetness index (TWI), slope length (LS), plan curvature, profile curvature, distance to rivers, distance to faults, lithology, land use, drainage density, and fault density. Subsequently, groundwater spring potential was modeled and mapped using CART, RF, and BRT algorithms. The predicted results from the three models were validated using the receiver operating characteristics curve (ROC). From 864 springs identified, 605 (≈70 %) locations were used for the spring potential mapping, while the remaining 259 (≈30 %) springs were used for the model validation. The area under the curve (AUC) for the BRT model was calculated as 0.8103 and for CART and RF the AUC were 0.7870 and 0.7119, respectively. Therefore, it was concluded that the BRT model produced the best prediction results while predicting locations of springs followed by CART and RF models, respectively. Geospatially integrated BRT, CART, and RF methods proved to be useful in generating the spring potential map (SPM) with reasonable accuracy.

Estimating occupancy dynamics for large-scale monitoring networks: amphibian breeding occupancy across protected areas in the northeast United States

Regional monitoring strategies frequently employ a nested sampling design where a finite set of study areas from throughout a region are selected and intensive sampling occurs within a subset of sites within the individual study areas. This sampling protocol naturally lends itself to a hierarchical analysis to account for dependence among subsamples. Implementing such an analysis using a classic likelihood framework is computationally challenging when accounting for detection errors in species occurrence models. Bayesian methods offer an alternative approach for fitting models that readily allows for spatial structure to be incorporated. We demonstrate a general approach for estimating occupancy when data come from a nested sampling design. We analyzed data from a regional monitoring program of wood frogs (Lithobates sylvaticus) and spotted salamanders (Ambystoma maculatum) in vernal pools using static and dynamic occupancy models. We analyzed observations from 2004 to 2013 that were collected within 14 protected areas located throughout the northeast United States. We use the data set to estimate trends in occupancy at both the regional and individual protected area levels. We show that occupancy at the regional level was relatively stable for both species. However, substantial variation occurred among study areas, with some populations declining and some increasing for both species. In addition, When the hierarchical study design is not accounted for, one would conclude stronger support for latitudinal gradient in trends than when using our approach that accounts for the nested design. In contrast to the model that does not account for nesting, the nested model did not include an effect of latitude in the 95% credible interval. These results shed light on the range‐level population status of these pond‐breeding amphibians, and our approach provides a framework that can be used to examine drivers of local and regional occurrence dynamics.