skip to main content

Attention:

The NSF Public Access Repository (PAR) system and access will be unavailable from 11:00 PM ET on Thursday, February 13 until 2:00 AM ET on Friday, February 14 due to maintenance. We apologize for the inconvenience.


Title: Guidelines for estimating occupancy from autocorrelated camera trap detections
Abstract

Site occupancy models (SOMs) are a common tool for studying the spatial ecology of wildlife. When observational data are collected using passive monitoring field methods, including camera traps or autonomous recorders, detections of animals may be temporally autocorrelated, leading to biased estimates and incorrectly quantified uncertainty. We presently lack clear guidance for understanding and mitigating the consequences of temporal autocorrelation when estimating occupancy models with camera trap data.

We use simulations to explore when and how autocorrelation gives rise to biased or overconfident estimates of occupancy. We explore the impact of sampling design and biological conditions on model performance in the presence of autocorrelation, investigate the usefulness of several techniques for identifying and mitigating bias and compare performance of the SOM to a model that explicitly estimates autocorrelation. We also conduct a case study using detections of 22 North American mammals.

We show that a join count goodness‐of‐fit test previously proposed for identifying clustered detections is effective for detecting autocorrelation across a range of conditions. We find that strong bias occurs in the estimated occupancy intercept when survey durations are short and detection rates are low. We provide a reference table for assessing the degree of bias to be expected under all conditions. We further find that discretizing data with larger windows decreases the magnitude of bias introduced by autocorrelation. In our case study, we find that detections of most species are autocorrelated and demonstrate how larger detection windows might mitigate the resulting bias.

Our findings suggest that autocorrelation is likely widespread in camera trap data and that many previous studies of occupancy based on camera trap data may have systematically underestimated occupancy probabilities. Moving forward, we recommend that ecologists estimating occupancy from camera trap data use the join count goodness‐of‐fit test to determine whether autocorrelation is present in their data. If it is, SOMs should use large detection windows to mitigate bias and more accurately quantify uncertainty in occupancy model parameters. Ecologists should not use gaps between detection periods, which are ineffective at mitigating temporal structure in data and discard useful data.

 
more » « less
Award ID(s):
2206784 2211768
PAR ID:
10511144
Author(s) / Creator(s):
 ;  ;  ;  ;  ;  ;  ;  
Publisher / Repository:
Wiley-Blackwell
Date Published:
Journal Name:
Methods in Ecology and Evolution
Volume:
15
Issue:
7
ISSN:
2041-210X
Format(s):
Medium: X Size: p. 1177-1191
Size(s):
p. 1177-1191
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    Camera traps deployed in grids or stratified random designs are a well‐established survey tool for wildlife but there has been little evaluation of study design parameters.

    We used an empirical subsampling approach involving 2,225 camera deployments run at 41 study areas around the world to evaluate three aspects of camera trap study design (number of sites, duration and season of sampling) and their influence on the estimation of three ecological metrics (species richness, occupancy and detection rate) for mammals.

    We found that 25–35 camera sites were needed for precise estimates of species richness, depending on scale of the study. The precision of species‐level estimates of occupancy (ψ) was highly sensitive to occupancy level, with <20 camera sites needed for precise estimates of common (ψ > 0.75) species, but more than 150 camera sites likely needed for rare (ψ < 0.25) species. Species detection rates were more difficult to estimate precisely at the grid level due to spatial heterogeneity, presumably driven by unaccounted habitat variability factors within the study area. Running a camera at a site for 2 weeks was most efficient for detecting new species, but 3–4 weeks were needed for precise estimates of local detection rate, with no gains in precision observed after 1 month. Metrics for all mammal communities were sensitive to seasonality, with 37%–50% of the species at the sites we examined fluctuating significantly in their occupancy or detection rates over the year. This effect was more pronounced in temperate sites, where seasonally sensitive species varied in relative abundance by an average factor of 4–5, and some species were completely absent in one season due to hibernation or migration.

    We recommend the following guidelines to efficiently obtain precise estimates of species richness, occupancy and detection rates with camera trap arrays: run each camera for 3–5 weeks across 40–60 sites per array. We recommend comparisons of detection rates be model based and include local covariates to help account for small‐scale variation. Furthermore, comparisons across study areas or times must account for seasonality, which could have strong impacts on mammal communities in both tropical and temperate sites.

     
    more » « less
  2. Abstract

    Resource selection functions (RSFs) are among the most commonly used statistical tools in both basic and applied animal ecology. They are typically parameterized using animal tracking data, and advances in animal tracking technology have led to increasing levels of autocorrelation between locations in such data sets. Because RSFs assume that data are independent and identically distributed, such autocorrelation can cause misleadingly narrow confidence intervals and biased parameter estimates.

    Data thinning, generalized estimating equations and step selection functions (SSFs) have been suggested as techniques for mitigating the statistical problems posed by autocorrelation, but these approaches have notable limitations that include statistical inefficiency, unclear or arbitrary targets for adequate levels of statistical independence, constraints in input data and (in the case of SSFs) scale‐dependent inference. To remedy these problems, we introduce a method for likelihood weighting of animal locations to mitigate the negative consequences of autocorrelation on RSFs.

    In this study, we demonstrate that this method weights each observed location in an animal's movement track according to its level of non‐independence, expanding confidence intervals and reducing bias that can arise when there are missing data in the movement track.

    Ecologists and conservation biologists can use this method to improve the quality of inferences derived from RSFs. We also provide a complete, annotated analytical workflow to help new users apply our method to their own animal tracking data using thectmm Rpackage.

     
    more » « less
  3. Abstract

    Understanding patterns of diversity is central to ecology and conservation, yet estimates of diversity are often biased by imperfect detection. In recent years, multi‐species occupancy models (MSOM) have been developed as a statistical tool to account for species‐specific heterogeneity in detection while estimating true measures of diversity. Although the power of these models has been tested in various ways, their ability to estimate gamma diversity—or true community size,Nis a largely unrecognized feature that needs rigorous evaluation.

    We use both simulations and an empirical dataset to evaluate the bias, precision, accuracy and coverage of estimates ofNfrom MSOM compared to the widely applied iChao2 non‐parametric estimator. We simulated 5,600 datasets across seven scenarios of varying average occupancy and detectability covariates, as well as varying numbers of sites, replicates and true community size. Additionally, we use a real dataset of surveys over 9 years (where species accumulation reached an asymptote, indicating trueN), to estimateNfrom each annual survey.

    Simulations showed that both MSOM and iChao2 estimators are generally accurate (i.e. unbiased and precise) except under unideal scenarios where mean species occupancy is low. In such scenarios, MSOM frequently overestimatedN. Across all scenarios, MSOM estimates were less certain than iChao2, but this led to over‐confident iChao2 estimates that showed poor coverage. Results from the real dataset largely confirmed the simulation findings, with MSOM estimates showing greater accuracy and coverage than iChao2.

    Community ecologists have a wide choice of analytical methods, and both iChao2 and MSOM estimates ofNare substantially preferable to raw species counts. The simplicity of non‐parametric estimators has obvious advantages, but our results show that in many cases, MSOM may provide superior estimates that also account more accurately for uncertainty. Both methods can show strong bias when average occupancy is very low, and practitioners should show caution when using estimates derived from either method under such conditions.

     
    more » « less
  4. Abstract

    Occupancy modelling is a common approach to assess species distribution patterns, while explicitly accounting for false absences in detection–nondetection data. Numerous extensions of the basic single‐species occupancy model exist to model multiple species, spatial autocorrelation and to integrate multiple data types. However, development of specialized and computationally efficient software to incorporate such extensions, especially for large datasets, is scarce or absent.

    We introduce thespOccupancy Rpackage designed to fit single‐species and multi‐species spatially explicit occupancy models. We fit all models within a Bayesian framework using Pólya‐Gamma data augmentation, which results in fast and efficient inference.spOccupancyprovides functionality for data integration of multiple single‐species detection–nondetection datasets via a joint likelihood framework. The package leverages Nearest Neighbour Gaussian Processes to account for spatial autocorrelation, which enables spatially explicit occupancy modelling for potentially massive datasets (e.g. 1,000s–100,000s of sites).

    spOccupancyprovides user‐friendly functions for data simulation, model fitting, model validation (by posterior predictive checks), model comparison (using information criteria and k‐fold cross‐validation) and out‐of‐sample prediction. We illustrate the package's functionality via a vignette, simulated data analysis and two bird case studies.

    ThespOccupancypackage provides a user‐friendly platform to fit a variety of single and multi‐species occupancy models, making it straightforward to address detection biases and spatial autocorrelation in species distribution models even for large datasets.

     
    more » « less
  5. Abstract

    Historical museum records provide potentially useful data for identifying drivers of change in species occupancy. However, because museum records are typically obtained via many collection methods, methodological developments are needed to enable robust inferences. Occupancy–detection models, a relatively new and powerful suite of statistical methods, are a potentially promising avenue because they can account for changes in collection effort through space and time.

    We use simulated datasets to identify how and when patterns in data and/or modelling decisions can bias inference. We focus primarily on the consequences of contrasting methodological approaches for dealing with species' ranges and inferring species' non‐detections in both space and time.

    We find that not all datasets are suitable for occupancy–detection analysis but, under the right conditions (namely, datasets that are broken into more time periods for occupancy inference and that contain a high fraction of community‐wide collections, or collection events that focus on communities of organisms), models can accurately estimate trends. Finally, we present a case study on eastern North American odonates where we calculate long‐term trends of occupancy using our most robust workflow.

    These results indicate that occupancy–detection models are a suitable framework for some research cases and expand the suite of available tools for macroecological analysis available to researchers, especially where structured datasets are unavailable.

     
    more » « less