Spatial proximity moderates genotype uncertainty in genetic tagging studies

Modelling heterogeneity in the classification process in multi-species distribution models can improve predictive performance

Species distribution models and maps from large-scale biodiversity data are necessary for conservation management. One current issue is that biodiversity data are prone to taxonomic misclassifications. Methods to account for these misclassifications in multi-species distribution models have assumed that the classification probabilities are constant throughout the study. In reality, classification probabilities are likely to vary with several covariates. Failure to account for such heterogeneity can lead to biased prediction of species distributions. Here, we present a general multi-species distribution model that accounts for heterogeneity in the classification process. The proposed model assumes a multinomial generalised linear model for the classification confusion matrix. We compare the performance of the heterogeneous classification model to that of the homogeneous classification model by assessing how well they estimate the parameters in the model and their predictive performance on hold-out samples. We applied the model to gull data from Norway, Denmark and Finland, obtained from the Global Biodiversity Information Facility. Our simulation study showed that accounting for heterogeneity in the classification process increased the precision of true species' identity predictions by 30% and accuracy and recall by 6%. Since all the models in this study accounted for misclassification of some sort, there was no significant effect of accounting for heterogeneity in the classification process on the inference about the ecological process. Applying the model framework to the gull dataset did not improve the predictive performance between the homogeneous and heterogeneous models (with parametric distributions) due to the smaller misclassified sample sizes. However, when machine learning predictive scores were used as weights to inform the species distribution models about the classification process, the precision increased by 70%. We recommend multiple multinomial regression to be used to model the variation in the classification process when the data contains relatively larger misclassified samples. Machine learning prediction scores should be used when the data contains relatively smaller misclassified samples.

Occupancy data improves parameter precision in spatial capture–recapture models

Population size is one of the basic demographic parameters for species management and conservation. Among different estimation methods, spatially explicit capture–recapture (SCR) models allow the estimation of population density in a framework that has been greatly developed in recent years. The use of automated detection devices, such as camera traps, has impressively extended SCR studies for individually identifiable species. However, its application to unmarked/partially marked species remains challenging, and no specific method has been widely used. We fitted an SCR‐integrated model (SCR‐IM) to stone marten Martes foina data, a species for which only some individuals are individually recognizable by natural marks, and estimate population size based on integration of three submodels: (1) individual capture histories from live capture and transponder tagging; (2) detection/nondetection or “occupancy” data using camera traps in a bigger area to extend the geographic scope of capture–recapture data; and (3) telemetry data from a set of tagged individuals. We estimated a stone marten density of 0.352 (SD: 0.081) individuals/km². We simulated four dilution scenarios of occupancy data to study the variation in the coefficient of variation in population size estimates. We also used simulations with similar characteristics as the stone marten case study, comparing the accuracy and precision obtained from SCR‐IM and SCR, to understand how submodels' integration affects the posterior distributions of estimated parameters. Based on our simulations, we found that population size estimates using SCR‐IM are more accurate and precise. In our stone marten case study, the SCR‐IM density estimation increased the precision by 37% when compared to the standard SCR model as regards to the coefficient of variation. This model has high potential to be used for species in which individual recognition by natural markings is not possible, therefore limiting the need to rely on invasive sampling procedures.

Evaluating otter reintroduction outcomes using genetic spatial capture-recapture modified for dendritic networks

Monitoring the demographics and genetics of reintroduced populations is critical to evaluating reintroduction success, but species ecology and the landscapes that they inhabit often present challenges for accurate assessments. If suitable habitats are restricted to hierarchical dendritic networks, such as river systems, animal movements are typically constrained and may violate assumptions of methods commonly used to estimate demographic parameters. Using genetic detection data collected via fecal sampling at latrines, we demonstrate applicability of the spatial capture-recapture (SCR) network distance function for estimating the size and density of a recently reintroduced North American river otter (Lontra canadensis) population in the Upper Rio Grande River dendritic network in the southwestern United States, and we also evaluated the genetic outcomes of using a small founder group (n = 33 otters) for reintroduction. Estimated population density was 0.23-0.28 otter/km, or 1 otter/3.57-4.35 km, with weak evidence of density increasing with northerly latitude (β = 0.33). Estimated population size was 83-104 total otters in 359 km of riverine dendritic network, which corresponded to average annual exponential population growth of 1.12-1.15/year since reintroduction. Growth was ≥40% lower than most reintroduced river otter populations and strong evidence of a founder effect existed 8-10 years post-reintroduction, including 13-21% genetic diversity loss, 84%-87% genetic effective population size decline, and rapid divergence from the source population (F ST accumulation = 0.06/generation). Consequently, genetic restoration via translocation of additional otters from other populations may be necessary to mitigate deleterious genetic effects in this small, isolated population. Combined with non-invasive genetic sampling, the SCR network distance approach is likely widely applicable to demogenetic assessments of both reintroduced and established populations of multiple mustelid species that inhabit aquatic dendritic networks, many of which are regionally or globally imperiled and may warrant reintroduction or augmentation efforts.

Perspectives on Individual Animal Identification from Biology and Computer Vision

Identifying individual animals is crucial for many biological investigations. In response to some of the limitations of current identification methods, new automated computer vision approaches have emerged with strong performance. Here, we review current advances of computer vision identification techniques to provide both computer scientists and biologists with an overview of the available tools and discuss their applications. We conclude by offering recommendations for starting an animal identification project, illustrate current limitations, and propose how they might be addressed in the future.

Augustine B, Linden DW, B. Chandler R, Royle JA. Spatial capture-recapture with random thinning for unidentified encounters

Spatial capture-recapture (SCR) models have increasingly been used as a basis for combining capture-recapture data types with variable levels of individual identity information to estimate population density and other demographic parameters. Recent examples are the unmarked SCR (or spatial count model), where no individual identities are available and spatial mark-resight (SMR) where individual identities are available for only a marked subset of the population. Currently lacking, though, is a model that allows unidentified samples to be combined with identified samples when there are no separate classes of "marked" and "unmarked" individuals and when the two sample types cannot be considered as arising from two independent observation models. This is a common scenario when using noninvasive sampling methods, for example, when analyzing data on identified and unidentified photographs or scats from the same sites.Here we describe a "random thinning" SCR model that utilizes encounters of both known and unknown identity samples using a natural mechanistic dependence between samples arising from a single observation model. Our model was fitted in a Bayesian framework using NIMBLE.We investigate the improvement in parameter estimates by including the unknown identity samples, which was notable (up to 79% more precise) in low-density populations with a low rate of identified encounters. We then applied the random thinning SCR model to a noninvasive genetic sampling study of brown bear (Ursus arctos) density in Oriental Cantabrian Mountains (North Spain).Our model can improve density estimation for noninvasive sampling studies for low-density populations with low rates of individual identification, by making use of available data that might otherwise be discarded.

Generalized model-based solutions to false-positive error in species detection/nondetection data

Detection/nondetection data are widely collected by ecologists interested in estimating species distributions, abundances, and phenology, and are often imperfect. Recent model development has focused on accounting for both false-positive and false-negative errors given evidence that misclassification is common across many sampling protocols. To date, however, model-based solutions to false-positive error have largely addressed occupancy estimation. We describe a generalized model structure that allows investigators to account for false-positive error in detection/nondetection data across a broad range of ecological parameters and model classes, and demonstrate that previously developed model-based solutions are special cases of the generalized model. Simulation results demonstrate that estimators for abundance and migratory arrival time ignoring false-positive error exhibit severe (20-70%) relative bias even when only 5-10% of detections are false positives. Bias increased when false-positive detections were more likely to occur at sites or within occasions in which true positive detections were unlikely to occur. Models accounting for false-positive error following the site-confirmation or observation-confirmation designs generally reduced bias substantially, even when few detections were confirmed as true or false positives or when the process model for false-positive error was misspecified. Results from an empirical example focusing on gray fox (Urocyon cinereoargenteus) abundance in Wisconsin, USA reinforce concerns that biases induced by false-positive error can also distort spatial predictions often used to guide decision making. Model sensitivity to false-positive error extends well beyond occupancy estimation, but encouragingly, model-based solutions developed for occupancy estimators are generalizable and effective across a range of models widely used in ecological research.

Spatial proximity moderates genotype uncertainty in genetic tagging studies

Accelerating declines of an increasing number of animal populations worldwide necessitate methods to reliably and efficiently estimate demographic parameters such as population density and trajectory. Standard methods for estimating demographic parameters from noninvasive genetic samples are inefficient because lower-quality samples cannot be used, and they assume individuals are identified without error. We introduce the genotype spatial partial identity model (gSPIM), which integrates a genetic classification model with a spatial population model to combine both spatial and genetic information, thus reducing genotype uncertainty and increasing the precision of demographic parameter estimates. We apply this model to data from a study of fishers (Pekania pennanti) in which 37% of hair samples were originally discarded because of uncertainty in individual identity. The gSPIM density estimate using all collected samples was 25% more precise than the original density estimate, and the model identified and corrected three errors in the original individual identity assignments. A simulation study demonstrated that our model increased the accuracy and precision of density estimates 63 and 42%, respectively, using three replicated assignments (e.g., PCRs for microsatellites) per genetic sample. Further, the simulations showed that the gSPIM model parameters are identifiable with only one replicated assignment per sample and that accuracy and precision are relatively insensitive to the number of replicated assignments for high-quality samples. Current genotyping protocols devote the majority of resources to replicating and confirming high-quality samples, but when using the gSPIM, genotyping protocols could be more efficient by devoting more resources to low-quality samples.