Spatially-explicit Model Research Articles

The influence of shark behavior and environmental conditions on baited remote underwater video survey results

Baited remote underwater video systems (BRUVS) have become an important and frequently used tool by resource managers to monitor relative abundances for a variety of marine species. As with any abundance survey method, fundamental assumptions of the technique are that counts derived from videos accurately reflect true changes to abundances, while being robust to changes in density-independent factors. We tested these assumptions using a newly developed spatially-explicit individual-based simulation model for two of the most commonly used video survey metrics of relative abundance; MaxN and MeanCount. Simulating a 1-km2 area over a 60-min BRUV deployment targeting elasmobranch species, we evaluated how resulting estimates of each metric were influenced by swimming speed, relative directness of movement patterns, relative attraction strength to bait, bait plume size, and camera visibility range while shark density was held constant. By simulating both standard (120°) and full-spherical (FS; 360°) fields of view (FOV), we were also able to explore if newly developed FS cameras could help reduce the impact of these density-independent factors. Results from both generalized linear models and visual inspection of simulations indicated that density as well as non-density-related factors were highly significant in predicting camera counts, regardless of the camera's FOV or count metric used. Collectively, these findings illustrate that video survey metrics are sensitive to factors unrelated to changes in density, and researchers should carefully consider their potential influence on survey results as well as potential management decisions based on these data.

How much habitat does a river need? A spatially-explicit population dynamics model to assess ratios of ontogenetical habitat needs

Restoration of spawning and juvenile habitats is often used to restore fish abundances in rivers, although often with unclear results. To study the effects of habitat limitations on the common barbel (Barbus barbus), a riverine litophilic cyprinid fish, an age-structured population model was developed. Using a Bayesian modeling approach, spawning and fry (0+ juvenile) habitat availability was integrated in the model in a spatially explicit way. Using Beverton-Holt and Ricker recruitment models, density dependence was incorporated in the spawning process and the recruitment of 0+ juveniles. Model parameters and their uncertainty ranges were obtained from reviewing the existing literature. The uncertainty of the processes was intrinsically accounted for by the inherently probabilistic nature of the Bayesian model. By testing various scenarios of habitat availabilities for the barbel, we hypothesize that improvement of the fish stock will be reached only at a well specified ratio of spawning to fry habitat. Model simulations revealed substantial abundance improvements at rather equal amounts of about 10% cover of both habitats, while even substantial improvements of either spawning or fry habitats only will result in little or no increase of abundance. Higher ratios of spawning to fry habitat were found to lower population recovery times. This work provides a tool that serves the assessment and comparison of river restoration scenarios as well as benchmarking rehabilitation targets in the planning phase. When targeting restoration of fish stocks, focusing only on one key life stage or process (such as spawning), without considering potential bottlenecks in other stages, can result in little to no improvement.

End-to-End Modeling Reveals Species-Specific Effects of Large-Scale Coastal Restoration on Living Resources Facing Climate Change

Coastal erosion and wetland loss are affecting Louisiana to such an extent that the loss of land between 1932 and 2016 was close to 5,000 km2. To mitigate this decline, coastal protection and restoration projects are being planned and implemented by the State of Louisiana, United States. The Louisiana Coastal Master Plan (CMP) is an adaptive management approach that provides a suite of projects that are predicted to build or maintain land and protect coastal communities. Restoring the coast with this 50-year large-scale restoration and risk reduction plan has the potential to change the biomass and distribution of economically and ecologically important fisheries species in this region. However, not restoring the coast may have negative impacts on these species due to the loss of habitat. This research uses an ecosystem model to evaluate the effects of plan implementation versus a future without action (FWOA) on the biomass and distribution of fisheries species in the estuaries over 50 years of model simulations. By simulating effects using a spatially-explicit ecosystem model, not only can the changes in biomass in response to plan implementation be evaluated, but also the distribution of species in response to the planned restoration and risk reduction projects. Simulations are performed under two relative sea level rise (SLR) scenarios to understand the effects of climate change on project performance and subsequent fisheries species biomass and distribution. Simulation output of eight economically important fisheries species shows that the plan mostly results in increases in species biomass, but that the outcomes are species-specific and basin-specific. The SLR scenario highly affects the amount of wetland habitat maintained after 50 years (with higher levels of wetland loss under increased SLR) and, subsequently, the biomass of species depending on that habitat. Species distribution results can be used to identify expected changes for specific species on a regional basis. By making this type of information available to resource managers, precautionary measures of ecosystem management and adaptation can be implemented.

Supporting non-target arthropods in agroecosystems: Modelling effects of insecticides and landscape structure on carabids in agricultural landscapes

Intensification of agricultural practices is one of the most important drivers of the dramatic decline of arthropod species. We do not know, however, the relative contribution to decline of different anthropogenic stressors that are part of this process. We used high-resolution dynamic landscape models and advanced spatially-explicit population modelling to estimate the relative importance of insecticide use and landscape structure for population dynamics of a widespread carabid beetle Bembidion lampros. The effects of in-crop mitigation measures through the application of insecticides with reduced lethality, and off-crop mitigation measures by increasing abundance of grassy field margins, were evaluated for the beetle along the gradient of landscape heterogeneity. Reducing the insecticide-driven lethality (from 90 to 10%) had larger positive impacts on beetle density and occupancy than increasing the abundance of field margins in a landscape. The effects of increasing field margins depended on their width and overall abundance in the landscape, but only field margins 4 m wide, applied to at least 40% of fields, resulted in an increase in beetle population density comparable to the scenario with the smallest reduction of insecticide-driven lethality we considered. Our findings suggest the importance of field margins rather as a supporting not stand-alone mitigation measure, as they generally improved effects of reduction of insecticide-driven lethality. Therefore, adding sufficiently broad off-field habitats should help to maintain viable beetle populations in agricultural landscapes even with moderate use of insecticides. In general, the less persistent the insecticides are in the environment, the larger positive impacts of applied mitigation measures on beetle populations were found. We also showed that the effectiveness of applied mitigation measures strongly depends on landscape and farmland heterogeneity. Thus, to achieve the same management or mitigation target in different landscapes might require different strategies.

Seasonal influence of snow conditions on Dall's sheep productivity in Wrangell-St Elias National Park and Preserve.

Dall's sheep (Ovis dalli dalli) are endemic to alpine areas of sub-Arctic and Arctic northwest America and are an ungulate species of high economic and cultural importance. Populations have historically experienced large fluctuations in size, and studies have linked population declines to decreased productivity as a consequence of late-spring snow cover. However, it is not known how the seasonality of snow accumulation and characteristics such as depth and density may affect Dall's sheep productivity. We examined relationships between snow and climate conditions and summer lamb production in Wrangell-St Elias National Park and Preserve, Alaska over a 37-year study period. To produce covariates pertaining to the quality of the snowpack, a spatially-explicit snow evolution model was forced with meteorological data from a gridded climate re-analysis from 1980 to 2017 and calibrated with ground-based snow surveys and validated by snow depth data from remote cameras. The best calibrated model produced an RMSE of 0.08 m (bias 0.06 m) for snow depth compared to the remote camera data. Observed lamb-to-ewe ratios from 19 summers of survey data were regressed against seasonally aggregated modelled snow and climate properties from the preceding snow season. We found that a multiple regression model of fall snow depth and fall air temperature explained 41% of the variance in lamb-to-ewe ratios (R2 = .41, F(2,38) = 14.89, p<0.001), with decreased lamb production following deep snow conditions and colder fall temperatures. Our results suggest the early establishment and persistence of challenging snow conditions is more important than snow conditions immediately prior to and during lambing. These findings may help wildlife managers to better anticipate Dall's sheep recruitment dynamics.

Fibersim, an Open-Source Spatially-Explicit Model of the Half-Sarcomere

FiberSim (https://campbell-muscle-lab.github.io/FiberSim/) is an open source spatially explicit model of a half-sarcomere, that represents a hexagonal lattice of compliant thin and thick filaments. Similar to prior frameworks developed by the Daniel, Tanner, and Mijailovich groups, FiberSim incorporates a variety of molecular mechanisms taking place during striated muscle contraction, such as thin filament activation and cross-bridge binding. Like any mathematical model of a biological system, FiberSim provides a convenient way to test hypotheses about the role of different molecular mechanisms and is a useful adjunct to experimental techniques.

The 1st ACM SIGSPATIAL International Workshop on Modeling and Understanding the Spread of COVID-19

In response to the COVID-19 pandemic, a number of spatially-explicit models have been developed to better explain the pathways of the disease, to predict the trajectory of the disease, and to test the effect of different health guidelines and policies on the number of cases and deaths. The 1st ACM SIGSPATIAL International Workshop on Modeling and Understanding the Spread of COVID-19 workshop (COVID'2020) featured research efforts that aim to understand the spatial processes and patterns of COVID-19 spread using a variety of spatial modeling, simulation, and mining approaches. The goal of this workshop was to bring together a range of interdisciplinary researchers in the SIGSPATIAL community in the fields of computer science, spatial modeling, social sciences, and epidemiology. Also, this workshop was advertised for anyone interested in infectious disease data and modelling, including but not limited to COVID-19.

Bundles and Hotspots of Multiple Ecosystem Services for Optimized Land Management in Kentucky, United States

Ecosystem services are benefits that the natural environment provides to support human well-being. A thorough understanding and assessment of these services are critical to maintain ecosystem services flow through sustainable land management to optimize bundles of ecosystem services provision. Maximizing one particular ecosystem service may lead to reduction in another. Therefore, identifying ecosystem services tradeoffs and synergies is key in addressing this challenge. However, the identification of multiple ecosystem services tradeoffs and synergies is still limited. A previous study failed to effectively capture the spatial interaction among ecosystem services as it was limited by “space-to-time” substitution method used because of temporal data scarcity. The study was also limited by using land use types in creating ecosystem services, which could lead to some deviations. The broad objective of this study is therefore to examine the bundles and hotspots of multiple ecosystem services and their tradeoffs in Kentucky, U.S. The study combined geographic data and spatially-explicit models to identify multiple ecosystem services bundles and hotspots, and determined the spatial locations of ecosystem services hotspots. Results showed that the spatial interactions among ecosystem services were very high: of the 21 possible pairs of ecosystem services, 17 pairs were significantly correlated. The seven ecosystem services examined can be bundled into three groups, geographically clustered on the landscape. These results support the hypothesis that some groups of ecosystem services provision can present similar spatial patterns at a large mesoscale. Understanding the spatial interactions and bundles of the ecosystem services provides essential information for evidence-based sustainable land management.

Variation in feral cat density between two large adjacent islands in Australia’s monsoon tropics

Despite contributing to the ongoing collapse of native mammal populations across northern Australian savannas, we have limited understanding of the ecological constraints of feral cat population density in this system. Addressing such knowledge gaps is a crucial step towards mitigating the impacts of feral cats, and is particularly important for the large islands off northern Australia that remain as strongholds for numerous species vulnerable to cat predation. Here, we investigated cat density across Melville and Bathurst Island, two large islands in Australia’s monsoon tropics. We deployed large grids (~13 km2) of 70 camera-traps at four locations to investigate how feral cat density varies under different combinations of fire frequency, and feral herbivore presence. Using spatially-explicit capture-recapture models, we estimated feral cat density on Melville Island to be 0.15 cats km−2. We did not record any cat detections on Bathurst Island. Using simulations, we predicted that if cat density on Bathurst Island was equal to that on Melville Island, we would have expected to record 27.9 detections of 9.9 individual cats. Our results, coupled with other recent surveys, suggest that the density of cats is much lower on Bathurst Island than the adjacent Melville Island. The absence of feral herbivores on Bathurst Island may have contributed to this variation in cat density. Management that enhances understorey vegetation density, through feral herbivore control, as well as fire management, could help mitigate the impact of feral cats on northern Australian savanna biodiversity.

Northern brown bandicoot (

Despite the ongoing collapse of native mammal populations across northern Australia, the paucity of robust estimates of population density limits our capacity to identify and understand population change. Here we aimed to provide the first estimates of native mammal density on the Tiwi Islands – one of Australia’s largest remaining refuge areas for native mammals. We conducted intensive live-trapping at four sites that represent varying combinations of fire frequency, feral cat density and feral herbivore presence. We used spatially-explicit capture-recapture models to investigate the density of common brushtail possum (Trichosurus vulpecula) and northern brown bandicoot (Isoodon macrourus). Compared with mainland northern Australia, populations of common brushtail possum and northern brown bandicoot have remained relatively healthy on the Tiwi Islands. Common brushtail possum density was significantly higher on Bathurst Island (1.06 possum ha−1) compared with Melville Island (0.32 possum ha−1), whereas northern brown bandicoot density varied across all four sites (ranging from 0.04 to 0.34 bandicoot ha−1). Unexpectedly, the very frequently burnt Ranku site (Bathurst Island) continues to support healthy populations of both species. These density estimates provide critical information for identifying and understanding future population change for two species that have suffered marked declines across the Australian monsoon tropics. Although the lack of replication limits our ability to draw conclusions regarding the ecological constraints of these mammal populations, our density observations align with a recent conceptual model postulating that the persistence of native mammal populations across northern Australian savannas reflects a complex, but spatially-variable interplay of ‘bottom-up’ and ‘top-down’ processes.

The effect of embayment complexity on ecological carrying capacity estimations in bivalve aquaculture sites

Bivalve aquaculture requires the alteration of natural populations of filter-feeders by artificially increasing their density. A bivalve farm could have negative consequences for the ecosystem if the filtration pressure of stocked biomass surpasses the capacity of the system to replenish the depleted resources. The concept of ecological carrying capacity, understood as the magnitude of aquaculture activity in a given area that can be supported without leading to unacceptable changes in the aquatic environment, is commonly used to inform management and regulatory decisions of bivalve aquaculture. In this study, a hydrodynamic model has been coupled to an ecological model that simulates the main dynamics of organic seston to evaluate the effects of bivalve aquaculture on seston supply and assess ecological carrying capacity. The spatially-explicit model allows the identification of areas where organic seston could be reduced beyond precautionary thresholds of ecosystem resilience. The model has been applied to three coastal embayments in Nova Scotia (Canada) that differ in water circulation and inlet/coastal complexity. The outcomes of the model suggest that the current aquaculture operations in Sober Island, Wine Harbour, and Whitehead are within the ecological carrying capacity of the ecosystem for bivalve aquaculture. The simulation of additional hypothetical stocking scenarios had demonstrated the relevance of local water circulation to the ecological carrying capacity of the system, and consequently for aquaculture operations. Accordingly, the placement of leases in areas with optimal circulation should be considered for planning purposes. The capability of the model to explore hypothetical scenarios could be used as a tool to guide management decisions in regard to site selection for new aquaculture sites.

Combining live and lethal trapping to inform the management of alien invasive rodent populations in a tropical montane forest

On large inhabited islands where complete eradication of alien invasive rodents through the use of poison delivery is often not practical or acceptable, mechanical trapping may represent the only viable option to reduce their impact in areas of high biodiversity value. However, the feasibility of sustained rodent control by trapping remains uncertain under realistic operational constraints. This study aimed to assess the effectiveness of non-toxic rat control strategies through a combination of lethal and live-trapping experiments, and scenario modelling, using the example of a remote montane rainforest of New Caledonia. Rat densities, estimated with spatially-explicit capture-recapture models, fluctuated seasonally (9.5–33.6 ind.ha-1). Capture probability (.01–.25) and home range sizes (HR95, .23–.75 ha) varied greatly according to trapping session, age class, sex and species. Controlling rats through the use of lethal trapping allowed maintaining rat densities at ca. 8 ind.ha-1 over a seven-month period in a 5.5-ha montane forest. Simulation models based on field parameter estimates over a 200-ha pilot management area indicated that without any financial and social constraints, trapping grids with the finest mesh sizes achieved cumulative capture probabilities &amp;gt; .90 after 15 trapping days, but were difficult to implement and sustain with the local workforce. We evaluated the costs and effectiveness of alternative trapping strategies taking into account the prevailing set of local constraints, and identified those that were likely to be successful. Scenario modelling, informed by trapping experiments, is a flexible tool for informing the design of sustainable control programs of island-invasive rodent populations, under idiosyncratic local circumstances.

Modeling spatial variations of the invasibility of slash pine flatwoods to Chinese tallow (Triadica sebifera) invasion: Mechanisms and key factors at the microscale

Protecting endangered slash pine (Pinus elliottii) flatwood ecosystems from the invasion of Chinese tallow (Triadica sebifera (L.) Small) a community-transforming invasive tree species is of critical, regional importance to the biota of the Gulf Coastal Plain. We developed a conceputal model to decipher the mechanism of tallow invasion at the microscale based on the community assembly theory, modeling the spatial variations of invasibility of slash pine flatwoods to Chinese tallow using a spatially-explicit point process model. A set of idenitified factors (filters) including the nearest distance to roads and seed sources (dispersal filters), canopy closure and pine: hardwood ratio (overstory filters), and grass cover and microtopography (understory filters) as well as tallow abundance and distribution (input for the model) were measured within 281 contiguous 30 m2 quadrats with a total area of 0.86-ha in a slash pine flatwood. The identified filters vary spatially and affect tallow invasion in a nonlinear “concave down” manner. Additive polynomial functions of these identified filters provide a viable approach to model the spatial variations of tallow invasion probability (invasibility). The full models which include all filters, as well as the sub-models which include only the dispersal filters and either of the overstory and understory filters, appear to have an adequate predictive power based on R2, RSE and skewness as well as the scatter- and qq-plots. In contrast, the sub-models that include only one of the filters failed to precisely predict the observed invasion outcomes. The results support the importance of dispersal filters and reducing propagule pressure levels for effective tallow control and slowing of spread.

Estimating the location of individual livestock holdings and their populations in two developing countries for use in spatial disease spread models

Infectious diseases of food animals, such as Foot-and-Mouth Disease (FMD), pose severe impacts on animal trade, animal products and subsequently endanger global food security. FMD is endemic in many parts of the world and is associated with substantial economic losses, which require risk assessments, preparedness planning, and evaluation of the effectiveness of mitigation strategies that fit within a country’s sociopolitical and socioeconomic constraints. Spatially-explicit stochastic simulation models (SESS) have become a common tool for estimating the spread and impact of FMD. SESS models incorporate uncertainty in the input and output parameters, heterogeneity in disease processes, and integrate geographic locations and spatial proximity of holdings that affect their relative exposure and transmission risk. An essential input to these models is locational data for holdings of animals and associated animal populations in each holding.Several efforts have been made to predict the location and population of livestock holdings or population density at different spatial resolutions. These methods or data cannot be used in developing countries because either the data is too coarse, or the inputs required for the methods are not available in resource-limited countries. As such, there is a need to adapt the practical and reliable existing methods to generate simulated datasets depicting the location and population of individual livestock holdings in developing countries for use in SESS models.We generated spatially-resolved simulated datasets for the location and population density of individual livestock holdings in Pakistan and Thailand. Firstly, we microsimulated and downscaled the census data to individual holdings based on statistical distributions. Second, geospatial probability surfaces were created based on a survey of expert veterinarians and empirical holding locations. Third, holdings were randomly placed on the probability surface based on a set of rules. These holdings were assigned population of livestock by joining downscaled data and random holdings. The combined dataset on the location and population of individual livestock holdings was, finally, used to generate the density of holdings.To our knowledge, this was the first attempt to estimate the locations and populations of individual livestock holdings in developing countries. These data pave the way for the application of SESS models in developing countries to understand the spread of FMD and evaluate mitigation strategies. The control of such an important animal disease would improve livestock health, improve economic gains for producers, and help alleviate poverty and hunger, which will complement efforts to attain the 2030 Sustainable Development Goals.

Combining refuges with transgenic insect releases for the management of an insect pest with non-recessive resistance to Bt crops in agricultural landscapes

Reinforcing the high-dose/refuge strategy with releases of transgenic insects has been suggested as a method for simultaneously managing agricultural pest populations and resistance to transgenic crops. Theoretical and empirical studies have shown that these approaches can work when deployed against closed populations and the assumptions of the HDR strategy are met. However, field-evolved resistance is often linked to non-recessive resistance or refuge non-compliance, and pest management regimes are likely to take place at the landscape-level. It is therefore important to understand how effective such strategies are when resistance is non-recessive, and how they could be employed in agricultural landscapes. We developed a spatially-explicit model to investigate the efficacy of strategies combining refuges with transgenic insect releases to manage a pest with non-recessive resistance in agricultural landscapes. We compared two release strategies, area-wide releases and localised releases targeted at population hotspots, and analysed the effects of refuge and release parameters on population and resistance dynamics. Area-wide releases reliably achieved landscape-level pest eradication. Localised releases also eradicated the pest when low release thresholds were combined with high release ratios, and maintained the pest at low densities when insufficient to achieve extinction. Reinforcing refuges with localised releases also greatly enhanced the probability of resistance extinction. However, when resistance remained in the population, localised releases prevented resistance from reaching fixation rather than greatly delaying or reversing resistance evolution. Our work indicates that combining refuges with simple release policies is effective for landscape-level pest suppression when the HDR assumptions are violated, but more nuanced release strategies may be required to enhance the benefits to resistance management.

Why Hantavirus Prevalence Does Not Always Increase With Host Density: Modeling the Role of Host Spatial Behavior and Maternal Antibodies.

For wildlife diseases, one often relies on host density to predict host infection prevalence and the subsequent force of infection to humans in the case of zoonoses. Indeed, if transmission is mainly indirect, i.e., by way of the environment, the force of infection is expected to increase with host density, yet the laborious field data supporting this theoretical claim are often absent. Hantaviruses are among those zoonoses that have been studied extensively over the past decades, as they pose a significant threat to humans. In Europe, the most widespread hantavirus is the Puumala virus (PUUV), which is carried by the bank vole and causes nephropathia epidemica (NE) in humans. Extensive field campaigns have been carried out in Central Finland to shed light on this supposed relationship between bank vole density and PUUV prevalence and to identify other drivers for the infection dynamics. This resulted in the surprising observation that the relationship between bank vole density and PUUV prevalence is not purely monotonic on an annual basis, contrary to what previous models predicted: a higher vole density does not necessary result in a higher infection prevalence, nor in an increased number of humans reported having NE. Here, we advance a novel individual-based spatially-explicit model which takes into account the immunity provided by maternal antibodies and which simulates the spatial behavior of the host, both possible causes for this discrepancy that were not accounted for in previous models. We show that the reduced prevalence in peak years can be attributed to transient immunity, and that the density-dependent spatial vole behavior, i.e., the fact that home ranges are smaller in high density years, plays only a minor role. The applicability of the model is not limited to the study and prediction of PUUV (and NE) occurrence in Europe, as it could be easily adapted to model other rodent-borne diseases, either with indirect or direct transmission.

CO2 emissions from forest degradation in Brazilian Amazon

Forest degradation is widespread around the world, due to multiple factors such as unsustainable logging, agriculture, invasive species, fire, fuelwood gathering, and livestock grazing. In the Brazilian Amazon forest degradation from August 2006 to July 2016 reached 1,1 869 800 ha. The processes of forest degradation are still poorly understood, being a missing component in anthropogenic CO2 emission estimates in tropical forests. In this work, we analyzed temporal trajectories of forest degradation from August 2006 to July 2016 in the Brazilian Amazon and assessed their impact on the regional carbon balance. We combined the degradation process with deforestation-related processes (clear-cut deforestation and secondary vegetation dynamics), using the spatially-explicit INPE-EM carbon emission model. The trajectory analysis showed that 13% of the degraded area ended up being cleared and converted in the period and 61% of the total degraded area experienced only one event of degradation throughout the whole period. Net emissions added up to 5.4 GtCO2, considering the emissions from forest degradation and deforestation, absorption from degraded forest recovery, and secondary vegetation dynamics. The results show an increase in the contribution of forest degradation to net emissions towards the end of the period, related to the decrease in clear-cut deforestation rates, decoupled from the forest degradation rates. The analysis also indicates that the regeneration of degraded forests absorbed 1.8 GtCO2 from August 2006 and July 2016—a component typically overlooked in the regional carbon balance.

Spatial Heterogeneity of Surface Topography in Peatlands: Assessing Overwintering Habitat Availability for the Eastern Massasauga Rattlesnake

The eastern massasauga rattlesnake (Sistrurus catenatus) relies on small-scale differences in peatland surface elevation to survive harsh overwintering conditions at the northern limit of its range. We characterized the spatial heterogeneity of surface topography in peatlands within the eastern Georgian Bay rock barrens landscape of Ontario, Canada, to assess available snake overwintering habitat. At six peatlands, we used a differential global positioning system to collect surface-elevation data. We created spatially-explicit surface models to map peatland surface topography, quantify habitat that was likely to remain unflooded during the overwintering period, and identify key characteristics associated with greater habitat availability. While surface elevations were spatially heterogeneous within and among sites, larger peatlands were associated with greater surface spatial variability relative to the lowest elevation measured within each site. However, even peatlands with very little spatial heterogeneity (average of 0.24 m above lowest elevation), provided unflooded overwintering habitat. Inter-annual weather conditions and peatland and watershed characteristics likely control the availability and distribution of unflooded overwintering habitat. We found that trees, specifically white pine (Pinus strobus) and maple (Acer spp.), were spatially associated with higher surface elevations and could be used to identify areas of unflooded winter habitat. Our findings are useful for landscape-scale assessments of available overwintering habitat to prioritize conservation and management efforts.

When the average hides the risk of Bt-corn pollen on non-target Lepidoptera: Application to Aglais io in Catalonia

Field cultivation of Genetically Modified (GM) Bt-plants has a potential environmental risk toward non-target Lepidoptera (NTLs) larvae through the consumption of Bt-maize pollen. The Bt-maize Cry protein targeting Lepidoptera species detrimental to the crop is also expressed in pollen which is dispersed by wind and can thus reach habitats of NTLs. To better assess the current ecological risk of Bt-maize at landscape scales, we developed a spatially-explicit exposure-hazard model considering (i) the dynamics of pollen dispersal obtained by convolving GM plants emission with a dispersal kernel and (ii) a toxicokinetic-toxicodynamic (TKTD) model accounting for the impact of toxin ingestion on individual lethal effects. We simulated the model using real landscape observations in Catalonia (Spain): GM-maize locations, flowering dates, rainfall time series and larvae emergence date of the European peacock butterfly Aglais io. While in average, the additional mortality appears to be negligible, we show significant additional mortality at sub-population level, with for instance a mortality higher than 40% within the 10m for the 10% most Bt-sensitive individuals. Also, using Pareto optimality we capture the best trade-off between isolation distance and additional mortality: up to 50 m are required to significantly buffer Bt-pollen impact on NTLs survival at the individual level. Our study clears up the narrow line between diverging conclusions: those claiming no risk by only looking at the average regional effect of Bt on NTLs survival and those pointing out a significant threaten when considering the variability of individuals mortality.

Counting cats for conservation: seasonal estimates of leopard density and drivers of distribution in the Serengeti

Large carnivore conservation is important for ecosystem integrity and understanding drivers of their abundance is essential to guide conservation efforts. Leopard (Panthera pardus) populations are in a general state of decline, although local studies demonstrated large variation in their population trends and density estimates vary widely across their range. We used spatially-explicit capture-recapture models for unmarked populations with camera trap data from a citizen science project to estimate previously-unknown leopard population densities in Serengeti National Park, Tanzania, and determine potential biological drivers of their abundance and distribution. We estimated leopard densities, at 5.41 (95% CrI = 2.23–9.26) and 5.72 (95% CrI = 2.44–9.55) individuals/100 km2, in the dry and wet season, respectively, which confirmed Serengeti National Park as one of the strongholds of this species in Africa. In contrast to abundance estimates, we found that drivers of leopard abundance and distribution varied among the dry and wet seasons, and were primarily affected by interactions with other larger carnivores and cover. The underlying driver of leopard distribution may be the dynamic prey availability which shifts between seasons, leading to an avoidance of dominant carnivores when prey availability is low in the dry season but an association with dominant carnivores when prey availability is high in the wet season. As efforts to conserve large carnivore populations increase worldwide, our results highlight the benefits of using data from citizen science projects, including large camera-trapping surveys, to estimate local carnivore abundances. Using a Bayesian framework allows of estimation of population density, but it is also important to understand the factors that dictate their distribution across the year to inform conservation efforts.