That's not the Mona Lisa! How to interpret spatial capture-recapture density surface estimates.

Spatial capture-recapture methods are often used to produce density surfaces, and these surfaces are often misinterpreted. In particular, spatial change in density is confused with spatial change in uncertainty about density. We illustrate correct and incorrect inference visually by treating a grayscale image of the Mona Lisa as an activity center intensity or density surface and simulating spatial capture-recapture survey data from it. Inferences can be drawn about the intensity of the point process generating activity centers, and about the likely locations of activity centers associated with the capture histories obtained from a single survey of a single realization of this process. We show that treating probabilistic predictions of activity center locations as estimates of the intensity of the process results in invalid and misleading ecological inferences, and that predictions are highly dependent on where the detectors are placed and how much survey effort is used. Estimates of the activity center density surface should be obtained by estimating the intensity of a point process model for activity centers. Practitioners should state explicitly whether they are estimating the intensity or making predictions of activity center location, and predictions of activity center locations should not be confused with estimates of the intensity.

A flexible framework for spatial capture-recapture with unknown identities.

Camera traps or acoustic recorders are often used to sample wildlife populations. When animals can be individually identified, these data can be used with spatial capture-recapture (SCR) methods to assess populations. However, obtaining animal identities is often labor-intensive and not always possible for all detected animals. To address this problem, we formulate SCR, including acoustic SCR, as a marked Poisson process, comprising a single counting process for the detections of all animals and a mark distribution for what is observed (eg, animal identity, detector location). The counting process applies equally when it is animals appearing in front of camera traps and when vocalizations are captured by microphones, although the definition of a mark changes. When animals cannot be uniquely identified, the observed marks arise from a mixture of mark distributions defined by the animal activity centers and additional characteristics. Our method generalizes existing latent identity SCR models and provides an integrated framework that includes acoustic SCR. We apply our method to estimate density from a camera trap study of fisher (Pekania pennanti) and an acoustic survey of Cape Peninsula moss frog (Arthroleptella lightfooti). We also test it through simulation. We find latent identity SCR with additional marks such as sex or time of arrival to be a reliable method for estimating animal density.

A latent capture history model for digital aerial surveys.

We anticipate that unmanned aerial vehicles will become popular wildlife survey platforms. Because detecting animals from the air is imperfect, we develop a mark-recapture line transect method using two digital cameras, possibly mounted on one aircraft, which cover the same area with a short time delay between them. Animal movement between the passage of the cameras introduces uncertainty in individual identity, so individual capture histories are unobservable and are treated as latent variables. We obtain the likelihood for mark-recapture line transects without capture histories by automatically enumerating all possibilities within segments of the transect that contain ambiguous identities, instead of attempting to decide identities in a prior step. We call this method "Latent Capture-history Enumeration" (LCE). We include an availability model for species that are periodically unavailable for detection, such as cetaceans that are undetectable while diving. External data are needed to estimate the availability cycle length, but not the mean availability rate, if the full availability model is employed. We compare the LCE method with the recently developed cluster capture-recapture method (CCR), which uses a Palm likelihood approximation, providing the first comparison of CCR with maximum likelihood. The LCE estimator has slightly lower variance, more so as sample size increases, and close to nominal coverage probabilities. Both methods are approximately unbiased. We illustrate with semisynthetic data from a harbor porpoise survey.

Uncovering ecological state dynamics with hidden Markov models.

Ecological systems can often be characterised by changes among a finite set of underlying states pertaining to individuals, populations, communities or entire ecosystems through time. Owing to the inherent difficulty of empirical field studies, ecological state dynamics operating at any level of this hierarchy can often be unobservable or 'hidden'. Ecologists must therefore often contend with incomplete or indirect observations that are somehow related to these underlying processes. By formally disentangling state and observation processes based on simple yet powerful mathematical properties that can be used to describe many ecological phenomena, hidden Markov models (HMMs) can facilitate inferences about complex system state dynamics that might otherwise be intractable. However, HMMs have only recently begun to gain traction within the broader ecological community. We provide a gentle introduction to HMMs, establish some common terminology, review the immense scope of HMMs for applied ecological research and provide a tutorial on implementation and interpretation. By illustrating how practitioners can use a simple conceptual template to customise HMMs for their specific systems of interest, revealing methodological links between existing applications, and highlighting some practical considerations and limitations of these approaches, our goal is to help establish HMMs as a fundamental inferential tool for ecologists.

Cluster capture-recapture to account for identification uncertainty on aerial surveys of animal populations.

Capture-recapture methods for estimating wildlife population sizes almost always require their users to identify every detected animal. Many modern-day wildlife surveys detect animals without physical capture-visual detection by cameras is one such example. However, for every pair of detections, the surveyor faces a decision that is often fraught with uncertainty: are they linked to the same individual? An inability to resolve every such decision to a high degree of certainty prevents the use of standard capture-recapture methods, impeding the estimation of animal density. Here, we develop an estimator for aerial surveys, on which two planes or unmanned vehicles (drones) fly a transect over the survey region, detecting individuals via high-definition cameras. Identities remain unknown, so one cannot discern if two detections match to the same animal; however, detections in close proximity are more likely to match. By modeling detection locations as a clustered point process, we extend recently developed methodology and propose a precise and computationally efficient estimator of animal density that does not require individual identification. We illustrate the method with an aerial survey of porpoise, on which cameras detect individuals at the surface of the sea, and we need to take account of the fact that they are not always at the surface.

Open population maximum likelihood spatial capture-recapture.

Open population capture-recapture models are widely used to estimate population demographics and abundance over time. Bayesian methods exist to incorporate open population modeling with spatial capture-recapture (SCR), allowing for estimation of the effective area sampled and population density. Here, open population SCR is formulated as a hidden Markov model (HMM), allowing inference by maximum likelihood for both Cormack-Jolly-Seber and Jolly-Seber models, with and without activity center movement. The method is applied to a 12-year survey of male jaguars (Panthera onca) in the Cockscomb Basin Wildlife Sanctuary, Belize, to estimate survival probability and population abundance over time. For this application, inference is shown to be biased when assuming activity centers are fixed over time, while including a model for activity center movement provides negligible bias and nominal confidence interval coverage, as demonstrated by a simulation study. The HMM approach is compared with Bayesian data augmentation and closed population models for this application. The method is substantially more computationally efficient than the Bayesian approach and provides a lower root-mean-square error in predicting population density compared to closed population models.

Double-observer line transect methods: levels of independence.

Double-observer line transect methods are becoming increasingly widespread, especially for the estimation of marine mammal abundance from aerial and shipboard surveys when detection of animals on the line is uncertain. The resulting data supplement conventional distance sampling data with two-sample mark-recapture data. Like conventional mark-recapture data, these have inherent problems for estimating abundance in the presence of heterogeneity. Unlike conventional mark-recapture methods, line transect methods use knowledge of the distribution of a covariate, which affects detection probability (namely, distance from the transect line) in inference. This knowledge can be used to diagnose unmodeled heterogeneity in the mark-recapture component of the data. By modeling the covariance in detection probabilities with distance, we show how the estimation problem can be formulated in terms of different levels of independence. At one extreme, full independence is assumed, as in the Petersen estimator (which does not use distance data); at the other extreme, independence only occurs in the limit as detection probability tends to one. Between the two extremes, there is a range of models, including those currently in common use, which have intermediate levels of independence. We show how this framework can be used to provide more reliable analysis of double-observer line transect data. We test the methods by simulation, and by analysis of a dataset for which true abundance is known. We illustrate the approach through analysis of minke whale sightings data from the North Sea and adjacent waters.

Using Continuous-Time Spatial Capture-Recapture models to make inference about animal activity patterns.

Quantifying the distribution of daily activity is an important component of behavioral ecology. Historically, it has been difficult to obtain data on activity patterns, especially for elusive species. However, the development of affordable camera traps and their widespread usage has led to an explosion of available data from which activity patterns can be estimated.Continuous-time spatial capture-recapture (CT SCR) models drop the occasion structure seen in traditional spatial and nonspatial capture-recapture (CR) models and use the actual times of capture. In addition to estimating density, CT SCR models estimate expected encounters through time. Cyclic splines can be used to allow flexible shapes for modeling cyclic activity patterns, and the fact that SCR models also incorporate distance means that space-time interactions can be explored. This method is applied to a jaguar dataset.Jaguars in Belize are most active and range furthest in the evening and early morning and when they are located closer to the network of trails. There is some evidence that females have a less variable pattern than males. The comparison between sexes demonstrates how CT SCR can be used to explore hypotheses about animal behavior within a formal modeling framework.SCR models were developed primarily to estimate and model density, but the models can be used to explore processes that interact across space and time, especially when using the CT SCR framework that models the temporal dimension at a finer resolution.

Population density estimated from locations of individuals on a passive detector array.

The density of a closed population of animals occupying stable home ranges may be estimated from detections of individuals on an array of detectors, using newly developed methods for spatially explicit capture-recapture. Likelihood-based methods provide estimates for data from multi-catch traps or from devices that record presence without restricting animal movement ("proximity" detectors such as camera traps and hair snags). As originally proposed, these methods require multiple sampling intervals. We show that equally precise and unbiased estimates may be obtained from a single sampling interval, using only the spatial pattern of detections. This considerably extends the range of possible applications, and we illustrate the potential by estimating density from simulated detections of bird vocalizations on a microphone array. Acoustic detection can be defined as occurring when received signal strength exceeds a threshold. We suggest detection models for binary acoustic data, and for continuous data comprising measurements of all signals above the threshold. While binary data are often sufficient for density estimation, modeling signal strength improves precision when the microphone array is small.

Estimating the encounter rate variance in distance sampling.

The dominant source of variance in line transect sampling is usually the encounter rate variance. Systematic survey designs are often used to reduce the true variability among different realizations of the design, but estimating the variance is difficult and estimators typically approximate the variance by treating the design as a simple random sample of lines. We explore the properties of different encounter rate variance estimators under random and systematic designs. We show that a design-based variance estimator improves upon the model-based estimator of Buckland et al. (2001, Introduction to Distance Sampling. Oxford: Oxford University Press, p. 79) when transects are positioned at random. However, if populations exhibit strong spatial trends, both estimators can have substantial positive bias under systematic designs. We show that poststratification is effective in reducing this bias.

An Efficient Acoustic Density Estimation Method with Human Detectors Applied to Gibbons in Cambodia.

Some animal species are hard to see but easy to hear. Standard visual methods for estimating population density for such species are often ineffective or inefficient, but methods based on passive acoustics show more promise. We develop spatially explicit capture-recapture (SECR) methods for territorial vocalising species, in which humans act as an acoustic detector array. We use SECR and estimated bearing data from a single-occasion acoustic survey of a gibbon population in northeastern Cambodia to estimate the density of calling groups. The properties of the estimator are assessed using a simulation study, in which a variety of survey designs are also investigated. We then present a new form of the SECR likelihood for multi-occasion data which accounts for the stochastic availability of animals. In the context of gibbon surveys this allows model-based estimation of the proportion of groups that produce territorial vocalisations on a given day, thereby enabling the density of groups, instead of the density of calling groups, to be estimated. We illustrate the performance of this new estimator by simulation. We show that it is possible to estimate density reliably from human acoustic detections of visually cryptic species using SECR methods. For gibbon surveys we also show that incorporating observers' estimates of bearings to detected groups substantially improves estimator performance. Using the new form of the SECR likelihood we demonstrate that estimates of availability, in addition to population density and detection function parameters, can be obtained from multi-occasion data, and that the detection function parameters are not confounded with the availability parameter. This acoustic SECR method provides a means of obtaining reliable density estimates for territorial vocalising species. It is also efficient in terms of data requirements since since it only requires routine survey data. We anticipate that the low-tech field requirements will make this method an attractive option in many situations where populations can be surveyed acoustically by humans.

A spatially explicit capture-recapture estimator for single-catch traps.

Single-catch traps are frequently used in live-trapping studies of small mammals. Thus far, a likelihood for single-catch traps has proven elusive and usually the likelihood for multicatch traps is used for spatially explicit capture-recapture (SECR) analyses of such data. Previous work found the multicatch likelihood to provide a robust estimator of average density. We build on a recently developed continuous-time model for SECR to derive a likelihood for single-catch traps. We use this to develop an estimator based on observed capture times and compare its performance by simulation to that of the multicatch estimator for various scenarios with nonconstant density surfaces. While the multicatch estimator is found to be a surprisingly robust estimator of average density, its performance deteriorates with high trap saturation and increasing density gradients. Moreover, it is found to be a poor estimator of the height of the detection function. By contrast, the single-catch estimators of density, distribution, and detection function parameters are found to be unbiased or nearly unbiased in all scenarios considered. This gain comes at the cost of higher variance. If there is no interest in interpreting the detection function parameters themselves, and if density is expected to be fairly constant over the survey region, then the multicatch estimator performs well with single-catch traps. However if accurate estimation of the detection function is of interest, or if density is expected to vary substantially in space, then there is merit in using the single-catch estimator when trap saturation is above about 60%. The estimator's performance is improved if care is taken to place traps so as to span the range of variables that affect animal distribution. As a single-catch likelihood with unknown capture times remains intractable for now, researchers using single-catch traps should aim to incorporate timing devices with their traps.

Estimating Consumption to Biomass Ratio in Non-Stationary Harvested Fish Populations.

The food consumption to biomass ratio (C) is one of the most important population parameters in ecosystem modelling because its quantifies the interactions between predator and prey. Existing models for estimating C in fish populations are per-recruit cohort models or empirical models, valid only for stationary populations. Moreover, empirical models lack theoretical support. Here we develop a theory and derive a general modelling framework to estimate C in fish populations, based on length frequency data and the generalised von Bertalanffy growth function, in which models for stationary populations with a stable-age distributions are special cases. Estimates using our method are compared with estimates from per-recruit cohort models for C using simulated harvested fish populations of different lifespans. The models proposed here are also applied to three fish populations that are targets of commercial fisheries in southern Chile. Uncertainty in the estimation of C was evaluated using a resampling approach. Simulations showed that stationary and non-stationary population models produce different estimates for C and those differences depend on the lifespan, fishing mortality and recruitment variations. Estimates of C using the new model exhibited smoother inter-annual variation in comparison with a per-recruit model estimates and they were also smaller than C predicted by the empirical equations in all population assessed.