Field evaluation of abundance estimates under binomial and multinomial N-mixture models

Assessing and modelling abundance from animal count data is a very common task in ecology and management. Detection is arguably never perfect, but modern hierarchical models can incorporate detection probability and yield abundance estimates that are corrected for imperfect detection. Two variants of these models rely on counts of unmarked individuals, or territories (binomial N-mixture models, or binmix), and on detection histories based on territory-mapping data (multinomial N-mixture models or multimix). However, calibration studies which evaluate these two N-mixture model approaches are needed. We analysed conventional territory-mapping data (three surveys in 2014 and four in 2015) using both binmix and multimix models to estimate abundance for two common avian cavity-nesting forest species (Great Tit Parus major and Eurasian Blue Tit Cyanistes caeruleus). In the same study area, we used two benchmarks: occupancy data from a dense nestbox scheme and total number of detected territories. To investigate variance in estimates due to the territory assignment, three independent ornithologists conducted territory assignments. Nestbox occupancy yields a minimum number of territories, as some natural cavities may have been used, and binmix model estimates were generally higher than this benchmark. Estimates using the multimix model were slightly more precise than binmix model estimates. Depending on the person assigning the territories, the multimix model estimates became quite different, either overestimating or underestimating the ‘truth’. We conclude that N-mixture models estimated abundance reliably, even for our very small sample sizes. Territory-mapping counts depended on territory assignment and this carried over to estimates under the multimix model. This limitation has to be taken into account when abundance estimates are compared between sites or years. Whenever possible, accounting for such hidden heterogeneity in the raw data of bird surveys, via including a ‘territory editor’ factor, is recommended. Distributing the surveys randomly (in time and space) to editors may also alleviate this problem.     

Evaluating abundance estimates and evidence of breeding for Bobolinks from transect and point-count surveys

Estimating the abundance and breeding success of territorial songbirds is challenging. Various types of surveys and analyses are available, but all receive some criticism in the literature, and most methods are rarely compared with results obtained using intensive monitoring efforts. We assessed the efficacy of transect and point-count surveys to estimate the abundance of male Bobolinks (Dolichonyx oryzivorus) and detect evidence of nesting and fledging by comparing the results of those surveys to results from more intensive monitoring (i.e., spot mapping and nest monitoring). We monitored 36 fields (254 ha) of late-harvest hay, restored grassland, and fallow fields in the Luther Marsh Wildlife Management Area and on four farms in southern Ontario, Canada, in 2018. Compared to the number of territories identified based on spot mapping (197), distance sampling analysis of transect survey data provided a more accurate estimate of the abundance of male Bobolinks (230, 95% CI: 187, 282) than N-mixture models of transect (668, 95% CI: 332, 1342) and point-count (337, 95% CI: 203, 559) data. Three visits to survey transects and five to point counts did not effectively detect evidence of Bobolink breeding (i.e., nesting or fledging) in fields compared to spot mapping and nest monitoring. Distance sampling analysis of transect data appears promising for estimating the number of Bobolink territories in an area, e.g., those impacted by conservation programs. If estimates of the number of nesting Bobolinks and frequency of fledging are of interest, spot mapping and nest monitoring could be implemented at a subset of sampled fields. Our results suggest that additional studies to evaluate model-based estimates of abundance with the best available information (e.g., from spot mapping of marked or unmarked populations and nest monitoring) would be useful to ensure that robust estimates are provided to support population estimates and conservation actions.     

Multinomial N-mixture models for removal sampling

Multinomial $N$-mixture models are commonly used to fit data from a removal sampling protocol. If the mixing distribution is negative binomial, the distribution of the counts does not appear to have been identified, and practitioners approximate the requisite likelihood by placing an upper bound on the embedded infinite sum. In this paper, the distribution which underpins the multinomial $N$-mixture model with a negative binomial mixing distribution is shown to belong to the broad class of multivariate negative binomial distributions. Specifically, the likelihood can be expressed in closed form as the product of conditional and marginal likelihoods and the information matrix shown to be block diagonal. As a consequence, the nature of the maximum likelihood estimates of the unknown parameters and their attendant standard errors can be examined and tests of the hypothesis of the Poisson against the negative binomial mixing distribution formulated. In addition, appropriate multinomial $N$-mixture models for data sets which include zero site totals can also be constructed. Two illustrative examples are provided.     

Experimental Investigation of Observation Error in Anuran Call Surveys

ABSTRACT Occupancy models that account for imperfect detection are often used to monitor anuran and songbird species occurrence. However, presence—absence data arising from auditory detections may be more prone to observation error (e.g., false-positive detections) than are sampling approaches utilizing physical captures or sightings of individuals. We conducted realistic, replicated field experiments using a remote broadcasting system to simulate simple anuran call surveys and to investigate potential factors affecting observation error in these studies. Distance, time, ambient noise, and observer abilities were the most important factors explaining false-negative detections. Distance and observer ability were the best overall predictors of false-positive errors, but ambient noise and competing species also affected error rates for some species. False-positive errors made up 5% of all positive detections, with individual observers exhibiting false-positive rates between 0.5% and 14%. Previous research suggests false-positive errors of these magnitudes would induce substantial positive biases in standard estimators of species occurrence, and we recommend practices to mitigate for false positives when developing occupancy monitoring protocols that rely on auditory detections. These recommendations include additional observer training, limiting the number of target species, and establishing distance and ambient noise thresholds during surveys.     

Making the most of survey data: Incorporating age uncertainty when fitting growth parameters

Individual growth is an important parameter and is linked to a number of other biological processes. It is commonly modeled using the von Bertalanffy growth function (VBGF), which is regularly fitted to age data where the ages of the animals are not known exactly but are binned into yearly age groups, such as fish survey data. Current methods of fitting the VBGF to these data treat all the binned ages as the actual ages. We present a new VBGF model that combines data from multiple surveys and allows the actual age of an animal to be inferred. By fitting to survey data for Atlantic herring (Clupea harengus) and Atlantic cod (Gadus morhua), we compare our model with two other ways of combining data from multiple surveys but where the ages are as reported in the survey data. We use the fitted parameters as inputs into a yield‐per‐recruit model to see what would happen to advice given to management. We found that each of the ways of combining the data leads to different parameter estimates for the VBGF and advice for policymakers. Our model fitted to the data better than either of the other models and also reduced the uncertainty in the parameter estimates and models used to inform management. Our model is a robust way of fitting the VBGF and can be used to combine data from multiple sources. The model is general enough to fit other growth curves for any taxon when the age of individuals is binned into groups.     

Evaluating the ability of regional models to predict local avian abundance

Spatial modeling over broad scales can potentially direct conservation efforts to areas with high species-specific abundances. We examined the performance of regional models for predicting bird abundance at spatial scales typically addressed in conservation planning. Specifically, we used point count data on wood thrush (Hylocichla mustelina) and blue-winged warbler (Vermivora cyanoptera) from 2 time periods (1995–1998 and 2006–2007) to evaluate the ability of regional models derived via Bayesian hierarchical techniques to predict bird abundance. We developed models for each species within Bird Conservation Region (BCR) 23 in the upper midwestern United States at 800-ha, 8,000-ha, and approximately 80,000-ha scales. We obtained count data from the Breeding Bird Survey and land cover data from the National Land Cover Dataset (1992). We evaluated predictions from the best models, as defined by an information-theoretic criterion, using point count data collected within an ecological subregion of BCR 23 at 131 count stations in the 1990s and again in 2006–2007. Competing (Deviance Information Criteria <5) blue-winged warbler models accounted for 67% of the variability and suggested positive associations with forest edge and proportion of forest at the 8,000-ha scale, and negative associations with forest patch area (800 ha) and wetness (800 ha and 80,000 ha). The regional model performed best for blue-winged warbler predicted abundances from point counts conducted in Iowa during 1995–1996 (r_s = 0.57; P = 0.14), the survey period that most closely aligned with the time period of data used for regional model construction. Wood thrush models exhibited positive correlations with point count data for all survey areas and years combined (r_s = 0.58, P ≤ 0.001). In comparison, blue-winged warbler models performed worse as time increased between the point count surveys and vintage of the model building data (r_s = 0.03, P = 0.92 for Iowa and r_s = 0.13, P = 0.51 for all areas, 2006–2007), likely related to the ephemeral nature of their preferred early successional habitat. Species abundance and sensitivity to changing habitat conditions seems to be an important factor in determining the predictive ability of regional models. Hierarchical models can be a useful tool for concentrating efforts at the scale of management units and should be one of many tools used by land managers, but we caution that the utility of such models may decrease over time for species preferring relatively ephemeral habitats if model inputs are not updated accordingly. © 2012 The Wildlife Society.     

Camera trap surveys to evaluate pest animal control operations

Many camera trap surveys aiming to detect the effects of management actions on pest animal populations are not as useful as they could be, because they do not produce informative results. We used mixed effects models, signed ranks tests and occupancy models to examine changes in the activity and occurrence of European red fox (Vulpes vulpes L.) in areas subjected to repeated fox management operations. In agricultural land, mixed effects models identified a decline in fox activity after baiting in 1 year, but not the next. Occupancy models revealed a decrease in fox occurrence in a National Park following control operations. These methods, combined with appropriate survey designs, could greatly enhance the value of many pest animal surveys conducted by management agencies.     

Integration of ground survey and remote sensing derived data: Producing robust indicators of habitat extent and condition

The availability of suitable habitat is a key predictor of the changing status of biodiversity. Quantifying habitat availability over large spatial scales is, however, challenging. Although remote sensing techniques have high spatial coverage, there is uncertainty associated with these estimates due to errors in classification. Alternatively, the extent of habitats can be estimated from ground‐based field survey. Financial and logistical constraints mean that on‐the‐ground surveys have much lower coverage, but they can produce much higher quality estimates of habitat extent in the areas that are surveyed. Here, we demonstrate a new combined model which uses both types of data to produce unified national estimates of the extent of four key habitats across Great Britain based on Countryside Survey and Land Cover Map. This approach considers that the true proportion of habitat per km² (Z_i) is unobserved, but both ground survey and remote sensing can be used to estimate Z_i. The model allows the relationship between remote sensing data and Z_i to be spatially biased while ground survey is assumed to be unbiased. Taking a statistical model‐based approach to integrating field survey and remote sensing data allows for information on bias and precision to be captured and propagated such that estimates produced and parameters estimated are robust and interpretable. A simulation study shows that the combined model should perform best when error in the ground survey data is low. We use repeat surveys to parameterize the variance of ground survey data and demonstrate that error in this data source is small. The model produced revised national estimates of broadleaved woodland, arable land, bog, and fen, marsh and swamp extent across Britain in 2007.     

Plant cover estimation based on the beta distribution in grassland vegetation

Cover is the most frequently used measure of abundance in vegetation surveys of grasslands, and various qualitative and semi-quantitative methods have been developed for visual estimation of this metric. Field survey is usually made with a point-grid plate. The frequency distributions of cover derived from point-grid counts follow a beta distribution. Combining point-grid counts from a field survey and the beta distribution for a statistical analysis, we developed an effort-saving cover-measurement method. Cover is measured with a transparent plastic plate on which, for example, 10 × 10 = 100 points are arranged in a lattice with 1-cm grid spacing (thus, one point count represents 1 cm² of cover). N quadrats are set out at randomly dispersed sites in a grassland, and, in each, the plastic plate is used for making counts. The number of grid points located above a given species is counted in every quadrat until the number of counted points reaches a given value c, which is determined in advance. If the number of counted points reaches c in a quadrat, the count is stopped and the quadrat is classified in the category “>c”. In quadrats where c is not attained, full point counts above the species bodies are made. Let g be the number of observed quadrats whose cover is ≤c. Using these g cover measurements and the number of quadrats (N − g) with cover >c, we can quantitatively estimate cover for each species and the spatial pattern index value based on the maximum likelihood method. In trial counts using this method, the time savings varied between 5% and 41%, depending on the shape of the cover frequency distribution. The mean cover value estimates agreed well with conventional measures without a stopping point (i.e., based on full counts of all points in each quadrat).     

A Sightability Model for Mountain Goats

ABSTRACT Unbiased estimates of mountain goat (Oreamnos americanus) populations are key to meeting diverse harvest management and conservation objectives. We developed logistic regression models of factors influencing sightability of mountain goat groups during helicopter surveys throughout the Cascades and Olympic Ranges in western Washington during summers, 2004–2007. We conducted 205 trials of the ability of aerial survey crews to detect groups of mountain goats whose presence was known based on simultaneous direct observation from the ground (n = 84), Global Positioning System (GPS) telemetry (n = 115), or both (n = 6). Aerial survey crews detected 77% and 79% of all groups known to be present based on ground observers and GPS collars, respectively. The best models indicated that sightability of mountain goat groups was a function of the number of mountain goats in a group, presence of terrain obstruction, and extent of overstory vegetation. Aerial counts of mountain goats within groups did not differ greatly from known group sizes, indicating that under-counting bias within detected groups of mountain goats was small. We applied Horvitz-Thompson-like sightability adjustments to 1,139 groups of mountain goats observed in the Cascade and Olympic ranges, Washington, USA, from 2004 to 2007. Estimated mean sightability of individual animals was 85% but ranged 0.75–0.91 in areas with low and high sightability, respectively. Simulations of mountain goat surveys indicated that precision of population estimates adjusted for sightability biases increased with population size and number of replicate surveys, providing general guidance for the design of future surveys. Because survey conditions, group sizes, and habitat occupied by goats vary among surveys, we recommend using sightability correction methods to decrease bias in population estimates from aerial surveys of mountain goats.     

Using Resource Selection Functions to Improve Estimation of Elk Population Numbers

Abstract: Stratification is commonly used to improve sampling efficiency of aerial surveys of ungulate populations with strata typically based on a priori information, such as preflight animal observations or vegetation attributes as surrogates for animal densities. We evaluated the usefulness of stratifying survey units for elk (Cervus elaphus) in the Rocky Mountain foothills of Alberta, Canada, using a resource selection function (RSF). We compared precision and design efficiency (DEFF) of population estimates from stratification approaches based on an RSF model to the past approach using amount of forest cover. We used a sample of telemetry relocations taken over a 3-year period from 165 elk, rarified to times of the day and months of the year when aerial surveys are conducted, to develop the RSF. We then used the top RSF model, based on Akaike's Information Criterion, to derive the average RSF value for an 8-km² survey unit. Using survey data from the first year, we evaluated binning schemes to define RSF-oriented strata based on poststratification and showed that Jenks natural breaks in the RSF values provided the greatest improvement in DEFF and increased precision, compared to 2 other stratification schemes. We then used this approach with data from 2 additional surveys to find that stratification by RSF consistently improves relative precision and design efficiency of elk population estimates, whether we employ pre- or poststratification. Where a RSF is available it could be used as a surrogate for animal densities when conducting stratified sampling for population surveys.     

Temporal pooling of point transect data increases precision in density estimates of southern chamois

Estimating animal abundances in small areas is a difficult task and because a limited number of observations often results in low-precision estimates whose inaccuracies may even be exacerbated if surveys are focussed on clustered populations and/or are only carried out once a year. In an attempt to overcome this problem, we used point transects to monthly survey two small areas of a game reserve to assess the density of Pyrenean chamois (Rupicapra p. pyrenaica). The coefficient of variation associated with the density estimates after pooling observations by season was still high but decreased to reasonable values (<20%) when observations were over 29 chamois groups (clusters). Our results suggest that Distance Sampling may be a useful way of estimating the population density of mountain ungulates such as Pyrenean chamois in small rugged areas where only a small or moderate number of observations are to be expected.     

Integrating variability in detection probabilities when designing wildlife surveys: a case study of amphibians from south-eastern Australia

Occupancy-based monitoring programs rely on survey data to infer presence or absence of the target species. However, species may occupy a site and go undetected, leading to erroneous inference of absence (‘false absence’). If detectability is influenced by the time of year or weather conditions, survey protocols can be adjusted to minimize the chance of false absences. In this study, detection probabilities for three amphibian species from south-eastern Australia were modelled using a Bayesian approach. For aural surveys, we compared basic models, which only included effects of survey date, duration and time of day on detection, to models including additional effects of weather. Model selection using deviance information criterion (DIC) suggested that the basic model was the most parsimonious for Crinia signifera, while models including relative humidity and water temperature were most supported for Limnodynastes dumerilii and L. tasmaniensis respectively. When predictive performance was assessed by cross validation, DIC results were largely matched for C. signifera and L. dumerilii, while models of detection for L. tasmaniensis were indistinguishable, AUC scores suggesting inadequate performance. We show how results such as these can be used to design surveys, developing protocols for individual surveys and estimating the number of surveys required under those protocols to achieve a threshold cumulative probability of detection. Conservation managers can use these models to maximize the efficiency of surveys. This will improve the accuracy of occupancy data, and reduce the risk of misdirected conservation actions resulting from false absences.     

Predicting the Geographic Distribution of a Species from Presence‐Only Data Subject to Detection Errors

Summary Several models have been developed to predict the geographic distribution of a species by combining measurements of covariates of occurrence at locations where the species is known to be present with measurements of the same covariates at other locations where species occurrence status (presence or absence) is unknown. In the absence of species detection errors, spatial point‐process models and binary‐regression models for case‐augmented surveys provide consistent estimators of a species’ geographic distribution without prior knowledge of species prevalence. In addition, these regression models can be modified to produce estimators of species abundance that are asymptotically equivalent to those of the spatial point‐process models. However, if species presence locations are subject to detection errors, neither class of models provides a consistent estimator of covariate effects unless the covariates of species abundance are distinct and independently distributed from the covariates of species detection probability. These analytical results are illustrated using simulation studies of data sets that contain a wide range of presence‐only sample sizes. Analyses of presence‐only data of three avian species observed in a survey of landbirds in western Montana and northern Idaho are compared with site‐occupancy analyses of detections and nondetections of these species.     

A practical sampling design for acoustic surveys of bats

Acoustic surveys are widely used for describing bat occurrence and activity patterns and are increasingly important for addressing concerns for habitat management, wind energy, and disease on bat populations. Designing these surveys presents unique challenges, particularly when a probabilistic sample is required for drawing inference to unsampled areas. Sampling frame errors and other logistical constraints often require survey sites to be dropped from the sample and new sites added. Maintaining spatial balance and representativeness of the sample when these changes are made can be problematic. Spatially balanced sampling designs recently developed to support aquatic surveys along rivers provide solutions to a number of practical challenges faced by bat researchers and allow for sample site additions and deletions, support unequal-probability selection of sites, and provide an approximately unbiased local neighborhood-weighted variance estimator that is efficient for spatially structured populations such as is typical for bats. We implemented a spatially balanced design to survey canyon bat (Parastrellus hesperus) activity along a stream network. The spatially balanced design accommodated typical logistical challenges and yielded a 25% smaller estimated standard error for the mean activity level than the usual simple random sampling estimator. Spatially balanced designs have broad application to bat research and monitoring programs and will improve studies relying on model-based inference (e.g., occupancy models) by providing flexibility and protection against violations of the independence assumption, even if design-based estimators are not used. Our approach is scalable and can be used for pre- and post-construction surveys along wind turbine arrays and for regional monitoring programs. © 2011 The Wildlife Society.     

Comparative Analysis of Mourning Dove Population Change in North America

ABSTRACT Mourning doves (Zenaida macroura) are surveyed in North America with a Call-Count Survey (CCS) and the North American Breeding Bird Survey (BBS). Analyses in recent years have identified inconsistencies in results between surveys, and a need exists to analyze the surveys using modern methods and examine possible causes of differences in survey results. Call-Count Survey observers collect separate information on number of doves heard and number of doves seen during counting, whereas BBS observers record one index containing all doves observed. We used hierarchical log-linear models to estimate trend and annual indices of abundance for 1966–2007 from BBS data, CCS-heard data, and CCS-seen data. Trend estimates from analyses provided inconsistent results for several states and for eastern and central dove-management units. We examined differential effects of change in land use and noise-related disturbance on the CCS indices. Changes in noise-related disturbance along CCS routes had a larger influence on the heard index than on the seen index, but association analyses among states of changes in temperature and of amounts of developed land suggest that CCS indices are differentially influenced by changes in these environmental features. Our hierarchical model should be used to estimate population change from dove surveys, because it provides an efficient framework for estimating population trends from dove indices while controlling for environmental features that differentially influence the indices.     

Integrating sign surveys and telemetry data for estimating brown bear (Ursus arctos) density in the Romanian Carpathians

Accurate population size estimates are important information for sustainable wildlife management. The Romanian Carpathians harbor the largest brown bear (Ursus arctos) population in Europe, yet current management relies on estimates of density that lack statistical oversight and ignore uncertainty deriving from track surveys. In this study, we investigate an alternative approach to estimate brown bear density using sign surveys along transects within a novel integration of occupancy models and home range methods. We performed repeated surveys along 2‐km segments of forest roads during three distinct seasons: spring 2011, fall‐winter 2011, and spring 2012, within three game management units and a Natura 2000 site. We estimated bears abundances along transects using the number of unique tracks observed per survey occasion via N‐mixture hierarchical models, which account for imperfect detection. To obtain brown bear densities, we combined these abundances with the effective sampling area of the transects, that is, estimated as a function of the median (± bootstrapped SE) of the core home range (5.58 ± 1.08 km²) based on telemetry data from 17 bears tracked for 1‐month periods overlapping our surveys windows. Our analyses yielded average brown bear densities (and 95% confidence intervals) for the three seasons of: 11.5 (7.8–15.3), 11.3 (7.4–15.2), and 12.4 (8.6–16.3) individuals/100 km². Across game management units, mean densities ranged between 7.5 and 14.8 individuals/100 km². Our method incorporates multiple sources of uncertainty (e.g., effective sampling area, imperfect detection) to estimate brown bear density, but the inference fundamentally relies on unmarked individuals only. While useful as a temporary approach to monitor brown bears, we urge implementing DNA capture–recapture methods regionally to inform brown bear management and recommend increasing resources for GPS collars to improve estimates of effective sampling area.     

Using a species distribution model to guide NSW surveys of the long‐footed potoroo (Potorous longipes)

Knowledge of threatened species’ distributions is essential for effective conservation decision‐making. Species distribution models (SDMs) are widely used to map species’ geographic ranges, identify new areas of suitable habitat and guide field surveys. In New South Wales (NSW), Australia, there are grave doubts about whether populations of the critically endangered long‐footed potoroo (Potorous longipes) remain extant, and identification of occupied sites is a high priority for its conservation. We used an SDM (Maxent) to identify regions in NSW that may have suitable habitat for the potoroo. The SDM was built with seven climate layers and had strong predictive performance (cross‐validated AUC = 0.94). We then combined this information on habitat suitability with vegetation and topography, to identify 58 survey sites across NSW. From April 2016 to May 2017, we undertook six field trips deploying six to eight cameras at each site for 52–63 days, resulting in 25 120 camera trap nights. A total of 215 759 images captured 43 native and feral animal species, but no long‐footed potoroos. Following the survey, newly available, independent presence and absence data were used to validate our model. A Kruskal–Wallis H test indicated that habitat suitability values were significantly higher at presence locations than absence locations (H = 58.66, d.f. = 1, P < 0.001). Finally, we refitted the Maxent model with the new data and identified additional regions that future surveys could explore. We conclude, however, that if the long‐footed potoroo remains extant in NSW, it is extremely rare.     

Errors in aerial survey count data: Identifying pitfalls and solutions

Accurate estimates of animal abundance are essential for guiding effective management, and poor survey data can produce misleading inferences. Aerial surveys are an efficient survey platform, capable of collecting wildlife data across large spatial extents in short timeframes. However, these surveys can yield unreliable data if not carefully executed. Despite a long history of aerial survey use in ecological research, problems common to aerial surveys have not yet been adequately resolved. Through an extensive review of the aerial survey literature over the last 50 years, we evaluated how common problems encountered in the data (including nondetection, counting error, and species misidentification) can manifest, the potential difficulties conferred, and the history of how these challenges have been addressed. Additionally, we used a double‐observer case study focused on waterbird data collected via aerial surveys and an online group (flock) counting quiz to explore the potential extent of each challenge and possible resolutions. We found that nearly three quarters of the aerial survey methodology literature focused on accounting for nondetection errors, while issues of counting error and misidentification were less commonly addressed. Through our case study, we demonstrated how these challenges can prove problematic by detailing the extent and magnitude of potential errors. Using our online quiz, we showed that aerial observers typically undercount group size and that the magnitude of counting errors increases with group size. Our results illustrate how each issue can act to bias inferences, highlighting the importance of considering individual methods for mitigating potential problems separately during survey design and analysis. We synthesized the information gained from our analyses to evaluate strategies for overcoming the challenges of using aerial survey data to estimate wildlife abundance, such as digital data collection methods, pooling species records by family, and ordinal modeling using binned data. Recognizing conditions that can lead to data collection errors and having reasonable solutions for addressing errors can allow researchers to allocate resources effectively to mitigate the most significant challenges for obtaining reliable aerial survey data.     

Composite analysis of black duck breeding population surveys in eastern North America

Waterfowl are monitored in eastern Canada and the northeastern United States with 2 surveys: a transect survey from fixed-wing aircraft and a plot survey conducted from helicopters. The surveys vary in extent, but overlap exists in a core area of 9 strata covering portions of all provinces from Ontario east to Newfoundland. We estimated population change for American black ducks (Anas rubripes) and mallards (Anas platyrhynchos) from these surveys using a log-linear hierarchical model that accommodates differences in sample design and visibility associated with these survey methods. Using a combined analysis of the surveys based on total indicated birds, we estimate the American black duck population to be 901,700 (95% CI: 715,200–1,274,000) in 2011, with 526,900 (95% CI: 357,500–852,300) mallards in the surveyed area. Precision of estimates varies widely by species and region, with transect surveys providing less precise results than plot surveys for black ducks in areas of overlap. The combined survey analysis for black ducks in the eastern survey region produced estimates with an average yearly coefficient of variation (CV) of 12.1% for the entire area and an average CV of 6.9% in the plot survey area. Mallards, which had a more limited distribution in the region, had an average yearly CV of 22.1% over the entire region, and an average CV of 27.7% in the plot survey area. Hierarchical models provide a rich framework for analyzing and combining results from complex survey designs, providing useful spatial and temporal information on population size and change in these economically important species. © 2012 The Wildlife Society.