Estimating occupancy and abundance using aerial images with imperfect detection

An easily implemented single-visit survey method for intermittently available and imperfectly detectable wildlife applied to the Florida east coast diamondback terrapin (Malaclemys terrapin tequesta)

Single-visit surveys of plots are often used for estimating the abundance of species of conservation concern. Less-than-perfect availability and detection of individuals can bias estimates if not properly accounted for. We developed field methods and a Bayesian model that accounts for availability and detection bias during single-visit visual plot surveys. We used simulated data to test the accuracy of the method under a realistic range of generating parameters and applied the method to Florida's east coast diamondback terrapin in the Indian River Lagoon system, where they were formerly common but have declined in recent decades. Simulations demonstrated that the method produces unbiased abundance estimates under a wide range of conditions that can be expected to occur in such surveys. Using terrapins as an example we show how to include covariates and random effects to improve estimates and learn about species-habitat relationships. Our method requires only counting individuals during short replicate surveys rather than keeping track of individual identity and is simple to implement in a variety of point count settings when individuals may be temporarily unavailable for observation. We provide examples in R and JAGS for implementing the model and to simulate and evaluate data to validate the application of the method under other study conditions.

Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys

Imagery from drones is becoming common in wildlife research and management, but processing data efficiently remains a challenge. We developed a methodology for training a convolutional neural network model on large-scale mosaic imagery to detect and count caribou (Rangifer tarandus), compare model performance with an experienced observer and a group of naïve observers, and discuss the use of aerial imagery and automated methods for large mammal surveys. Combining images taken at 75 m and 120 m above ground level, a faster region-based convolutional neural network (Faster-RCNN) model was trained in using annotated imagery with the labels: "adult caribou", "calf caribou", and "ghost caribou" (animals moving between images, producing blurring individuals during the photogrammetry processing). Accuracy, precision, and recall of the model were 80%, 90%, and 88%, respectively. Detections between the model and experienced observer were highly correlated (Pearson: 0.96-0.99, P value < 0.05). The model was generally more effective in detecting adults, calves, and ghosts than naïve observers at both altitudes. We also discuss the need to improve consistency of observers' annotations if manual review will be used to train models accurately. Generalization of automated methods for large mammal detections will be necessary for large-scale studies with diverse platforms, airspace restrictions, and sensor capabilities.

Leopard and spotted hyena densities in the Lake Mburo National Park, southwestern Uganda

Robust measures of animal densities are necessary for effective wildlife management. Leopards (Panthera pardus) and spotted hyenas (Crocuta Crocuta) are higher order predators that are data deficient across much of their East African range and in Uganda, excepting for one peer-reviewed study on hyenas, there are presently no credible population estimates for these species. A lack of information on the population status and even baseline densities of these species has ramifications as leopards are drawcards for the photo-tourism industry, and along with hyenas are often responsible for livestock depredations from pastoralist communities. Leopards are also sometimes hunted for sport. Establishing baseline density estimates for these species is urgently needed not only for population monitoring purposes, but in the design of sustainable management offtakes, and in assessing certain conservation interventions like financial compensation for livestock depredation. Accordingly, we ran a single-season survey of these carnivores in the Lake Mburo National Park of south-western Uganda using 60 remote camera traps distributed in a paired format at 30 locations. We analysed hyena and leopard detections under a Bayesian spatially explicit capture-recapture (SECR) modelling framework to estimate their densities. This small national park (370 km²) is surrounded by Bahima pastoralist communities with high densities of cattle on the park edge (with regular park incursions). Leopard densities were estimated at 6.31 individuals/100 km² (posterior SD = 1.47, 95% CI [3.75-9.20]), and spotted hyena densities were 10.99 individuals/100 km², but with wide confidence intervals (posterior SD = 3.35, 95% CI [5.63-17.37]). Leopard and spotted hyena abundance within the boundaries of the national park were 24.87 (posterior SD 7.78) and 39.07 individuals (posterior = SD 13.51) respectively. Leopard densities were on the middle end of SECR studies published in the peer-reviewed literature over the last 5 years while spotted hyena densities were some of the first reported in the literature using SECR, and similar to a study in Botswana which reported 11.80 spotted hyenas/100 km². Densities were not noticeably lower at the park edge, and in the southwest of our study site, despite repeated cattle incursions into these areas. We postulate that the relatively high densities of both species in the region could be owed to impala Aepyceros melampus densities ranging from 16.6-25.6 impala/km². Another, potential explanatory variable (albeit a speculative one) is the absence of interspecific competition from African lions (Panthera leo), which became functionally extinct (there is only one male lion present) in the park nearly two decades ago. This study provides the first robust population estimate of these species anywhere in Uganda and suggests leopards and spotted hyenas continue to persist in the highly modified landscape of Lake Mburo National Park.

Recursive Bayesian computation facilitates adaptive optimal design in ecological studies

Optimal design procedures provide a framework to leverage the learning generated by ecological models to flexibly and efficiently deploy future monitoring efforts. At the same time, Bayesian hierarchical models have become widespread in ecology and offer a rich set of tools for ecological learning and inference. However, coupling these methods with an optimal design framework can become computationally intractable. Recursive Bayesian computation offers a way to substantially reduce this computational burden, making optimal design accessible for modern Bayesian ecological models. We demonstrate the application of so-called prior-proposal recursive Bayes to optimal design using a simulated data binary regression and the real-world example of monitoring and modeling sea otters in Glacier Bay, Alaska. These examples highlight the computational gains offered by recursive Bayesian methods and the tighter fusion of monitoring and science that those computational gains enable.

Diffusion modeling reveals effects of multiple release sites and human activity on a recolonizing apex predator

BACKGROUND

Reintroducing predators is a promising conservation tool to help remedy human-caused ecosystem changes. However, the growth and spread of a reintroduced population is a spatiotemporal process that is driven by a suite of factors, such as habitat change, human activity, and prey availability. Sea otters (Enhydra lutris) are apex predators of nearshore marine ecosystems that had declined nearly to extinction across much of their range by the early 20th century. In Southeast Alaska, which is comprised of a diverse matrix of nearshore habitat and managed areas, reintroduction of 413 individuals in the late 1960s initiated the growth and spread of a population that now exceeds 25,000.

METHODS

Periodic aerial surveys in the region provide a time series of spatially-explicit data to investigate factors influencing this successful and ongoing recovery. We integrated an ecological diffusion model that accounted for spatially-variable motility and density-dependent population growth, as well as multiple population epicenters, into a Bayesian hierarchical framework to help understand the factors influencing the success of this recovery.

RESULTS

Our results indicated that sea otters exhibited higher residence time as well as greater equilibrium abundance in Glacier Bay, a protected area, and in areas where there is limited or no commercial fishing. Asymptotic spread rates suggested sea otters colonized Southeast Alaska at rates of 1-8 km/yr with lower rates occurring in areas correlated with higher residence time, which primarily included areas near shore and closed to commercial fishing. Further, we found that the intrinsic growth rate of sea otters may be higher than previous estimates suggested.

CONCLUSIONS

This study shows how predator recolonization can occur from multiple population epicenters. Additionally, our results suggest spatial heterogeneity in the physical environment as well as human activity and management can influence recolonization processes, both in terms of movement (or motility) and density dependence.

Estimating Reproduction and Survival of Unmarked Juveniles Using Aerial Images and Marked Adults

Methods for estimating juvenile survival of wildlife populations often rely on intensive data collection efforts to capture and uniquely mark individual juveniles and observe them through time. Capturing juveniles in a time frame sufficient to estimate survival can be challenging due to narrow and stochastic windows of opportunity. For many animals, juvenile survival depends on postnatal parental care (e.g., lactating mammals). When a marked adult gives birth to, and provides care for, juvenile animals, investigators can use the adult mark to locate and count unmarked juveniles. Our objective was to leverage the dependency between juveniles and adults and develop a framework for estimating reproductive rates, juvenile survival, and detection probability using repeated observations of marked adult animals with known fates, but imperfect detection probability, and unmarked juveniles with unknown fates. Our methods assume population closure for adults and that no juvenile births or adoptions take place after monitoring has begun. We conducted simulations to evaluate methods and then developed a field study to examine our methods using real data consisting of a population of mule deer in a remote area in central Nevada. Using simulations, we found that our methods were able to recover the true values used to generate the data well. Estimates of juvenile survival rates from our field study were 0.96, (95% CRI 0.83–0.99) for approximately 32-day periods between late June and late August. The methods we describe show promise for many applications and study systems with similar data types, and our methods can be easily extended to unmanned aerial platforms and cameras that are already commercially available for the types of images we used.

Supplementary materials accompanying this paper appear online.

The effect of a multi-target protocol on cetacean detection and abundance estimation in aerial surveys

A double-platform protocol was implemented in the Bay of Biscay and English Channel during the SCANS-III survey (2016). Two observation platforms using different protocols were operating on board a single aircraft: the reference platform (Scans), targeting cetaceans, and the 'Megafauna' platform, recording all the marine fauna visible at the sea surface (jellyfish to seabirds). We tested for a potential bias in small cetacean detection and density estimation when recording all marine fauna. At a small temporal scale (30 s, roughly 1.5 km), our results provided overall similar perception probabilities for both platforms. Small cetacean perception was higher following the detection of another cetacean within the previous 30 s in both platforms. The only prior target that decreased small cetacean perception during the subsequent 30 s was seabirds, in the Megafauna platform. However, at a larger scale (study area), this small-scale perception bias had no effect on the density estimates, which were similar for the two protocols. As a result, there was no evidence of lower performance regarding small cetacean population monitoring for the multi-target protocol in our study area. Because our study area was characterized by moderate cetacean densities and small spatial overlap of cetaceans and seabirds, any extrapolation to other areas or time requires caution. Nonetheless, by permitting the collection of cost-effective quantitative data for marine fauna, anthropogenic activities and marine litter at the sea surface, the multi-target protocol is valuable for optimizing logistical and financial resources to efficiently monitor biodiversity and study community ecology.

Understanding meta-population trends of the Australian fur seal, with insights for adaptive monitoring

Effective ecosystem-based management requires estimates of abundance and population trends of species of interest. Trend analyses are often limited due to sparse or short-term abundance estimates for populations that can be logistically difficult to monitor over time. Therefore it is critical to assess regularly the quality of the metrics in long-term monitoring programs. For a monitoring program to provide meaningful data and remain relevant, it needs to incorporate technological improvements and the changing requirements of stakeholders, while maintaining the integrity of the data. In this paper we critically examine the monitoring program for the Australian fur seal (AFS) Arctocephalus pusillus doriferus as an example of an ad-hoc monitoring program that was co-ordinated across multiple stakeholders as a range-wide census of live pups in the Austral summers of 2002, 2007 and 2013. This 5-yearly census, combined with historic counts at individual sites, successfully tracked increasing population trends as signs of population recovery up to 2007. The 2013 census identified the first reduction in AFS pup numbers (14,248 live pups, -4.2% change per annum since 2007), however we have limited information to understand this change. We analyse the trends at breeding colonies and perform a power analysis to critically examine the reliability of those trends. We then assess the gaps in the monitoring program and discuss how we may transition this surveillance style program to an adaptive monitoring program than can evolve over time and achieve its goals. The census results are used for ecosystem-based modelling for fisheries management and emergency response planning. The ultimate goal for this program is to obtain the data we need with minimal cost, effort and impact on the fur seals. In conclusion we identify the importance of power analyses for interpreting trends, the value of regularly assessing long-term monitoring programs and proper design so that adaptive monitoring principles can be applied.

Monitoring dynamic spatio-temporal ecological processes optimally

Population dynamics vary in space and time. Survey designs that ignore these dynamics may be inefficient and fail to capture essential spatio-temporal variability of a process. Alternatively, dynamic survey designs explicitly incorporate knowledge of ecological processes, the associated uncertainty in those processes, and can be optimized with respect to monitoring objectives. We describe a cohesive framework for monitoring a spreading population that explicitly links animal movement models with survey design and monitoring objectives. We apply the framework to develop an optimal survey design for sea otters in Glacier Bay. Sea otters were first detected in Glacier Bay in 1988 and have since increased in both abundance and distribution; abundance estimates increased from 5 otters to >5,000 otters, and they have spread faster than 2.7 km/yr. By explicitly linking animal movement models and survey design, we are able to reduce uncertainty associated with forecasting occupancy, abundance, and distribution compared to other potential random designs. The framework we describe is general, and we outline steps to applying it to novel systems and taxa.