New technologies in the mix: Assessing N‐mixture models for abundance estimation using automated detection data from drone surveys

1. Reliable estimates of abundance are critical in effectively managing threatened species, but the feasibility of integrating data from wildlife surveys completed using advanced technologies such as remotely piloted aircraft systems (RPAS) and machine learning into abundance estimation methods such as N‐mixture modeling is largely unknown due to the unique sources of detection errors associated with these technologies.
2. We evaluated two modeling approaches for estimating the abundance of koalas detected automatically in RPAS imagery: (a) a generalized N‐mixture model and (b) a modified Horvitz–Thompson (H‐T) estimator method combining generalized linear models and generalized additive models for overall probability of detection, false detection, and duplicate detection. The final estimates from each model were compared to the true number of koalas present as determined by telemetry‐assisted ground surveys.
3. The modified H‐T estimator approach performed best, with the true count of koalas captured within the 95% confidence intervals around the abundance estimates in all 4 surveys in the testing dataset (n = 138 detected objects), a particularly strong result given the difficulty in attaining accuracy found with previous methods.
4. The results suggested that N‐mixture models in their current form may not be the most appropriate approach to estimating the abundance of wildlife detected in RPAS surveys with automated detection, and accurate estimates could be made with approaches that account for spurious detections.

     

Spatial and temporal factors influencing sightability of elk

Few tracking studies consider seasonal changes in ability to re-sight wildlife, despite potential for biases in sightability to mislead our interpretation of models of movement and abundance. We developed seasonal sightability models based on visual observations of radio-collared elk (Cervus elaphus) in Manitoba, Canada, through 6 seasons. We located 377 elk 8,862 times using aerial telemetry from 2002 to 2009. We tested the hypothesis that sites where we were able to visually observe radio-collared elk during aerial telemetry differed from sites where collared elk were known to be present but could not be sighted. Relationships varied with season and elk sightability was influenced by forest type, habitat openness, distance to edge, and time of day. Our results confirm that observers have the highest probability of detecting elk in early and late winter. However, factors such as day length, which increases by 64% during this period, suggest that fewer impediments to detection exist in late winter. Our findings reinforce the need to account for seasonal as well as spatial changes in habitat-specific sightability models. © 2011 The Wildlife Society.     

Aerial flow of barley yellow dwarf viruses and of their vectors in western France

During the years 1989–1992 cereal aphids were caught alive in a low level (1.5 m high) suction trap operated in Le Rheu (Brittany, France) and tested for BYDV transmission. In most cases comparisons with data collected simultaneously by a 12.2 m suction trap operating in the same site resulted in good relationships between weekly catches at both heights. Results from transmission tests showed that: (i) the two main BYDV vectors were Rhopalosiphum padi and Metopolophium dirhodum during the years of experiment; (ii) PAV and MAV were the commonest viruses and RPV was relatively scarce; (iii) during spring M. dirhodum appeared to be the most important MAV vector and nearly as good a PAV vector as R. padi; (iv) during autumn R. padi was the only vector of the three viruses with mixed transmission allowing it to transmit also MAV probably by heteroencapsidation. To give an indication of the risk of infection, infectivity indices were calculated by multiplying the numbers of aphids caught by the 12.2 m suction trap by the proportion that were infective. These infectivity indices agreed with field records of primary infections.     

Comparative efficacy of two controlled-release gypsy moth mating disruption formulations

The effects of aerial applications of the gypsy moth sex pheromone, disparlure, on mating disruption and suppression of growth of populations of the gypsy moth, Lymantria dispar (L.), were investigated. Two formulations of disparlure, plastic laminate flakes applied in a single application and polymethacrylate beads applied in two applications, were compared in two separate tests conducted in 1993 and 1994. The beads were applied in two applications spaced 2 weeks apart because preliminary tests had indicated that they released pheromone too rapidly to maintain adequate emission rates throughout the period of male flight. In 1993, the flakes were applied at a rate of 50 g a.i./ha, and the beads were applied at a rate of 15 g a.i./ha for each application. In 1994, the flakes were applied at a rate of 75 g a.i./ha and the beads were applied at rates of 32.5 and 42.5 g a.i./ha for the two applications. Beads with larger average particle size were used in 1994 to prolong disparlure release. The treatments applied in 1993 resulted in >97% reduction in mating and >82% suppression of population growth in the following year. Because of a 1995 collapse of gypsy moth populations in the vicinity of the tests, reliable population growth data were not available for the treatments applied in 1994, but significant mating disruption did occur under both treatments. Based on measurements of residual disparlure after field aging, the flakes released 32 and 48% of their disparlure content during the 6 weeks of male moth flight in 1993 and 1994, respectively. The smaller beads used in 1993 released 75% of their disparlure content, and the larger beads used in 1994 released 52% of their disparlure content, during the 6 weeks of male flight. The biological efficacy data suggest that the bead and flake formulations, as applied in these tests, have similar effects on gypsy moth mating disruption and subsequent population growth. Based on the observed release rates from both 1993 and 1994, a single application of the beads would provide emission rates equal to or greater than those provided by the flakes when applied at an equal dose.     

Fine‐scale bird monitoring from light unmanned aircraft systems

Unmanned aircraft systems (UAS) are remote‐controlled devices capable of collecting information from difficult‐to‐access places while minimizing disturbance. Although UAS are increasingly used in many research disciplines, their application to wildlife research remains to be explored in depth. Here, we report on the use of a small UAS to monitor temporal changes in breeding population size in a Black‐headed Gull Chroicocephalus ridibundus colony. This method makes it possible to obtain georeferenced data on nest locations without causing colony disturbance, which would not otherwise be possible via direct ground observations.     

Invasive Snake Activity Before and After Automated Aerial Baiting

Aerially delivered toxic baits have proven effective for landscape-level control of numerous invasive vertebrate populations with major benefits for conservation and ecosystem function, but this technique has not been broadly adapted for control of invasive reptiles. Nonnative brown treesnakes (Boiga irregularis) on the Pacific island of Guam have caused severe ecological and economic damage and pose an invasion risk on other islands, making them a high-profile candidate for application of aerial baiting methods. Although terrestrial applications of traps, toxicants, and hand-removal are standard brown treesnake management practices, these methods are not cost-effective for control in the island's large tracts of remote, rugged forest. In 2016, the first major in situ evaluation of a helicopter-borne automated aerial bait delivery system applied snake-targeted toxic baits at an effective rate of approximately 120 baits/ha over a 110-ha forested test plot on Guam. We evaluated the extent and duration of the suppressive effect of this toxic bait application on brown treesnakes by measuring nontoxic bait take rates as a proxy index of relative snake abundance before and after toxic bait application in a treatment plot and surrounding reference area. We placed 4,420 nontoxic baits in random transects at georeferenced locations, from 1 month before until nearly 12 months after toxic bait application, allowing temporal analysis of the suppressive effect and spatial analysis of treatment plot reinvasion. Over the first 30 days after toxic bait application, average nontoxic bait take rate in the treatment plot was 41.2% lower than the pre-application rate, and there was no immediate decrease in bait take in the reference area. Reduced snake activity was still evident nearly 12 months after bait application. Roads forming a portion of the treatment boundary appeared to slow snake movement between treated area and surrounding untreated area. Trail cameras monitoring a subset of bait tubes showed that 97.5% of baits removed were taken by snakes rather than nontarget species. We indexed rodent abundance in the treatment plot and reference area, and found no indication of a rodent population increase following toxic bait application. Our results show that automated aerial bait applications can suppress brown treesnake abundance over a large area and that reinvasion from surrounding untreated habitat occurs over several months. We anticipate that repeated bait applications could achieve and maintain greatly reduced brown treesnake abundance on a landscape scale, potentially improving biosecurity and enabling experimental reintroduction of native birds extirpated by brown treesnake predation. Published 2019. This article is a U.S. Government work and is in the public domain in the USA. The Journal of Wildlife Management published by Wiley Periodicals, Inc. on behalf of The Wildlife Society.     

Accounting for residual heterogeneity in double-observer sightability models decreases bias in burro abundance estimates

Feral burros (Equus asinus) and horses (E. ferus caballus) inhabiting public land in the western United States are intended to be managed at population levels established to promote a thriving, natural ecological balance. Double-observer sightability (M_DS) models, which use detection records from multiple observers and sighting covariates, perform well for estimating feral horse abundances, but their effectiveness for use in burro populations is less understood. These M_DS models help minimize detection bias, yet bias can be further reduced with models that account for unmodeled variation, or residual heterogeneity, in detection probability. In populations containing radio-marked individuals, residual heterogeneity can be estimated with M_DS models by including a covariate that corresponds to the marked status of a group (M_H models). Another approach is to use information from detections missed by both observers to account for the characteristics that make groups more or less likely to be detected, or recaptured, by the second observer (M_R models). We used aerial survey data from 3 burro populations (Sinbad Herd Management Area, UT [2016–2018], Lake Pleasant Herd Management Area, AZ [2017], and Fort Irwin National Training Center, CA [2016–2017]) to develop M_DS models applicable for feral burros in the southwestern United States. Our objectives were to quantify precision and bias of standard M_DS surveys for feral burros and to examine which model type for incorporating residual heterogeneity (M_H or M_R) would result in the least-biased estimates of burro populations relative to the minimum number known alive (MNKA) within the Sinbad Herd Management Area. Standard M_DS model estimates achieved a mean coefficient of variation of 0.08, while underestimating MNKA by an average of 27.1%. Accounting for residual heterogeneity through recapture probability in M_R models resulted in estimates closer to MNKA than M_H models (9.5% vs. 16.5% less than MNKA). Our results indicate that M_DS models can achieve precise enough estimates to monitor feral burro populations, but they routinely produce negatively biased estimates. We encourage the use of radio-collars to reduce bias in future burro surveys by accounting for residual heterogeneity through M_R models.     

Demographic response of brown treesnakes to extended population suppression

From a management perspective, reptiles are relatively novel invasive taxa. Few methods for reptile control have been developed and very little is known about their effectiveness for reducing reptile populations, particularly when the goal is eradication. Many reptiles, and especially snakes, are cryptic, secretive, and undergo extended periods of inactivity, traits that decrease detection probabilities and create challenges in estimating population size or evaluating management effects. The brown treesnake (Boiga irregularis) is a notorious invasive species that continues to cause major ecological and economic harm following their introduction to the island of Guam after World War II. They have been the subject of intensive research on the effectiveness of various techniques to control snakes, including the first ever aerial system for the distribution of toxic acetaminophen baits for reptile control. We provide a cohort-based life table for a cryptic and invasive reptile undergoing extended population control using toxic baits from March 2017–2020. We also evaluated the effects of single (toxic bait) versus multi-tool (toxic bait and live trapping) management efforts on population trajectories, and estimated which population vital rates are most important for influencing population growth or decline in a treated landscape. Treatment of the population with acetaminophen-laced baits resulted in an immediate reduction followed by a gradual population decline that suggested that eradication was the probable outcome given sufficient treatment time but that the period of treatment was decades in magnitude. Inclusion of live trapping reduced the predicted time required to achieve eradication by more than half. Preventing the transition of 1,000-mm snout-vent length (SVL) females to larger sizes was predicted to have the greatest effect on population reduction based on integral projection modeling. Our results suggest that toxic baits are capable of eradicating brown treesnakes in an enclosure, although inclusion of trapping reduced overall treatment time required. Tools that effectively target females >1,000 mm SVL may have the greatest effect on reducing overall treatment timelines.     

Accuracy and precision of dolphin group size estimates

Estimating the number of dolphins in a group is a challenging task. To assess the accuracy and precision of dolphin group size estimates, observer estimates were compared to counts from large‐format vertical aerial photographs. During 11 research cruises, a total of 2,435 size estimates of 434 groups were made by 59 observers. Observer estimates were modeled as a function of the photo count in a hierarchical Bayesian framework. Accuracy varied widely among observers, and somewhat less widely among dolphin species. Most observers tended to underestimate, and the tendency increased with group size. Groups of 25, 50, 100, and 500 were underestimated by <1%, 16%, 27%, and 47%, respectively, on average. Precision of group size estimates was low, and estimates were highly variable among observers for the same group. Predicted true group size, given an observer estimate, was larger than the observer estimate for groups of more than about 25 dolphins. Predicted group size had low precision, with coefficients of variation ranging from 0.7 to 1.9. Studies which depend on group size estimates will be improved if the tendency to underestimate group size and the high uncertainty of group size estimates are included in the analysis.