True Occupancy Research Articles

On the trade-off between accuracy and spatial resolution when estimating species occupancy from geographically biased samples

Species occupancy is often defined as the proportion of areal units (sites) in a landscape that the focal species occupies, but it is usually estimated from the subset of sites that have been sampled. Assuming no measurement error, we show that three quantities–the degree of sampling bias (in terms of site selection), the proportion of sites that have been sampled and the variability of true occupancy across sites–determine the extent to which a sample-based estimate of occupancy differs from its true value across the wider landscape. That these are the only three quantities (measurement error notwithstanding) to affect the accuracy of estimates of species occupancy is the fundamental insight of the “Meng equation”, an algebraic re-expression of statistical error. We use simulations to show how each of the three quantities vary with the spatial resolution of the analysis and that absolute estimation error is lower at coarser resolutions. Absolute error scales similarly with resolution regardless of the size and clustering of the virtual species’ distribution. Finely resolved estimates of species occupancy have the potential to be more useful than coarse ones, but this potential is only realised if the estimates are at least reasonably accurate. Consequently, wherever there is the potential for sampling bias, there is a trade-off between spatial resolution and accuracy, and the Meng equation provides a theoretical framework in which analysts can consider the balance between the two. An obvious next step is to consider the implications of the Meng equation for estimating a time trend in species occupancy, where it is the confounding of error and true change that is of most interest.

Leveraging Diffusion Perturbations for Measuring Fairness in Computer Vision

Computer vision models have been known to encode harmful biases, leading to the potentially unfair treatment of historically marginalized groups, such as people of color. However, there remains a lack of datasets balanced along demographic traits that can be used to evaluate the downstream fairness of these models. In this work, we demonstrate that diffusion models can be leveraged to create such a dataset. We first use a diffusion model to generate a large set of images depicting various occupations. Subsequently, each image is edited using inpainting to generate multiple variants, where each variant refers to a different perceived race. Using this dataset, we benchmark several vision-language models on a multi-class occupation classification task. We find that images generated with non-Caucasian labels have a significantly higher occupation misclassification rate than images generated with Caucasian labels, and that several misclassifications are suggestive of racial biases. We measure a model’s downstream fairness by computing the standard deviation in the probability of predicting the true occupation label across the different identity groups. Using this fairness metric, we find significant disparities between the evaluated vision-and-language models. We hope that our work demonstrates the potential value of diffusion methods for fairness evaluations.

Corridor-Level Auto Occupancy in Mobility Monitoring Efforts: A Crash-Based Approach

Vehicle occupancy (the number of persons per vehicle) is fundamental to evaluations of transportation system performance focused on the movement of people rather than solely vehicles. Historically, obtaining detailed estimates of occupancy has required manual (expensive) field data collection. This paper reports on a crash-based approach for estimating occupancy. Because some characteristics, such as the presence of multiple passengers aged 16 to 21, may affect the likelihood of crash involvement, the study examined ways to control possible crash bias using two methods. The first—heuristic bias correction, which entails synthesizing missing samples—is useful for smaller jurisdictions where there is no ground truth value readily available but an area-wide occupancy measure is needed. The second—statistical bias correction, which uses manual observations for calibration followed by the creation of a model—is suitable for obtaining occupancy at the corridor level. The approach is demonstrated with 12 years of fall and spring crash data for the Hampton Roads region of southeastern Virginia, U.S., representing 1.7 million people and 50,000 crashes. Testing showed estimated occupancies were within roughly 0.05 of true occupancy when corridor-specific calibration was performed. Further, the larger number of samples used for an area-wide occupancy means that heuristic bias correction is not always necessary: for cities with more than 200 samples, the difference between uncorrected and corrected occupancies was 0.02. For states that report the number of occupants for crashes with and without injury, this approach holds promise for estimating vehicle occupancy for both corridors and regions.

Combining expert knowledge of a threatened trout distribution with sparse occupancy data for climate‐related projection

AbstractObjectiveTo evaluate the vulnerability of Bull Trout Salvelinus confluentus to potential climate changes across its range in Oregon, we compiled disparate expert knowledge of the distribution of spawning and rearing and combined these probabilistic statements as data along with documented records of breeding and rearing in a joint occupancy model.MethodsThe joint expert knowledge–occupancy model, which was based on discrete patches of cold water (≤13°C) suitable for spawning and rearing, permitted the association of true occupancy with climate and other explanatory variables while accounting for variation in detection probability. We then applied estimated relationships of patch occupancy with explanatory variables to projected coldwater patch configurations in the years 2040 and 2080.ResultProjections of the kilometers of occupied coldwater patch in future decades suggest precipitous declines if current relationships of occupancy with environmental variables are maintained. Impacts of climate changes in future decades manifest directly through the outright loss of coldwater patches and increases in winter high flows but also indirectly by increased isolation.ConclusionCombining probabilistic statements of species distributions from knowledgeable experts with sparse occupancy data may be a robust and timely alternative when large numbers of repeated occupancy surveys are infeasible.

Designing a Long-term Occupancy Monitoring Plan for a Cryptic Reptile

Monitoring populations of cryptic reptiles is challenging because of their crypticity. Occupancy monitoring is a useful technique for local populations, but seasonal unavailability for detection can result in large swings in apparent annual occupancy. We used data from 5 yr of occupancy surveys, and the observed sampling and process error, to evaluate the power to detect true change in occupancy under a range of sampling scenarios and occupancy trends for a cryptic reptile, Mojave Desert Tortoise (Gopherus agassizii) None of the sampling strategies that we tested had the power to detect a 1% per annum increase in true occupancy over a 10-yr period and required ≥70 sites to detect a 1% decline in occupancy. For a 2% per annum change in true occupancy, 50 sites were needed and for a 3–4% per annum change 20–30 sites were needed. Power to detect a 4% per annum decline in occupancy was >80% for all number of secondary visits and number of sites if the number of sites was ≥50, but required seven visits to ≥60 sites to detect a 4% increase in occupancy Multiple sampling designs provided equivalent power. For example, both designs of 80 sites with three secondary visits and 50 sites with seven secondary visits yielded >50% power on average to detect 1–4% per annum changes in true occupancy. These results can help local managers of Mojave Desert Tortoises and other cryptic reptiles to design optimally efficient occupancy monitoring strategies.

Monitoring Scaled Quail Occupancy and Colonization Post-Translocation on a Large Landscape

Large landscapes are important for sustaining quail populations in semiarid climates where annual variation in vital rates, and thus population volatility, tends to be larger than in subtropical climates. Translocations may need to be conducted on a similar scale to ensure long-term success. Large landscapes pose challenges for monitoring release sites in terms of costs and logistics. However, large landscapes also provide an excellent opportunity to evaluate habitat preferences and suitability because they inherently hold more variation in habitat type. Multiseason occupancy surveys are a potential monitoring tool for translocations where population persistence is a benchmark for success. Occupancy (i.e., presence–absence) data for quail are relatively easy to collect compared to more intensive surveying (i.e., mark-recapture or distance sampling) and can be analyzed in a framework that allows for the estimation of detection, colonization, and extinction as functions of spatial or temporal covariates. We used a multiseason occupancy survey to monitor a reintroduced population of scaled quail (Callipepla squamata) on a large landscape. Our objectives were to: 1) evaluate occupancy post-translocation, and 2) determine how landscape characteristics and distance from release points influenced colonization and extinction. Over 800 scaled quail were translocated to a >40,000-ha study area of contiguous rangeland in Knox County, Texas, USA from 2016–2017. We collected presence–absence data during a 10-day period in March just prior to release (2016) and for 2 years after first release (2017–2018). We sampled 73 locations on a 1.5-km × 1.5-km grid 3 times per year. We were unable to estimate true occupancy and thus any influence of covariates, because of low detection probability (p = 0.05, standard error = 0.02). However, we found that naïve occupancy based on detection within and outside of surveys increased from 1% in 2016 to 23% and 10% in 2017 and 2018, respectively. We recommend that monitoring programs prioritize survey methods that increase detection, such as sampling only during peak calling and call-back surveys, and using more than one method of detection.

Stationary and Moving Occupancy Detection Using the SLEEPIR Sensor Module and Machine Learning

Passive infrared (PIR) sensors, currently used for indoor lighting control report frequent false negative detects for stationary occupancy, which takes up almost 50% of the total occupancy rate. To address this, we developed a synchronized low-energy electronically chopped PIR (SLEEPIR) sensor that incorporates a liquid crystal (LC) shutter to chop the long-wave infrared signal received by the on-board PIR sensor. In this paper, we present a SLEEPIR sensor module, integrated with a PIR and machine learning for systematic evaluation of true occupancy detection in daily life. We design complex experimental scenarios containing a series of continuous daily activities and individual actions to simulate realistic environment to the maximum extend. We extract and down select key statistical features using recursive feature elimination with cross validation. In the end, we compared six machine learning models to evaluate the detection performance. Experiments involving continuous daily activities indicate an accuracy of 99.12% by using the support vector machine classifier.

Improving Area of Occupancy Estimates for Parapatric Species Using Distribution Models and Support Vector Machines

As geographic range estimates for the IUCN Red List guide conservation actions, accuracy and ecological realism are crucial. IUCN's extent of occurrence (EOO) is the general region including the species' range, while area of occupancy (AOO) is the subset of EOO occupied by the species. Data-poor species with incomplete sampling present particular difficulties, but species distribution models (SDMs) can be used to predict suitable areas. Nevertheless, SDMs typically employ abiotic variables (i.e., climate) and do not explicitly account for biotic interactions that can impose range constraints. We sought to improve range estimates for data-poor, parapatric species by masking out areas under inferred competitive exclusion. We did so for two South American spiny pocket mice: Heteromys australis (Least Concern) and Heteromys teleus (Vulnerable due to especially poor sampling), whose ranges appear restricted by competition. For both species, we estimated EOO using SDMs and AOO with four approaches: occupied grid cells, abiotic SDM prediction, and this prediction masked by approximations of the areas occupied by each species' congener. We made the masks using support vector machines (SVMs) fit with two data types: occurrence coordinates alone; and coordinates along with SDM predictions of suitability. Given the uncertainty in calculating AOO for low-data species, we made estimates for the lower and upper bounds for AOO, but only make recommendations for H. teleus as its full known range was considered. The SVM approaches (especially the second one) had lower classification error and made more ecologically realistic delineations of the contact zone. For H. teleus, the lower AOO bound (a strongly biased underestimate) corresponded to Endangered (occupied grid cells), while the upper bounds (other approaches) led to Near Threatened. As we currently lack data to determine the species' true occupancy within the post-processed SDM prediction, we recommend that an updated listing for H. teleus include these bounds for AOO. This study advances methods for estimating the upper bound of AOO and highlights the need for better ways to produce unbiased estimates of lower bounds. More generally, the SVM approaches for post-processing SDM predictions hold promise for improving range estimates for other uses in biogeography and conservation.

Improving area of occupancy estimates for parapatric species using distribution models and support vector machines.

As geographic range estimates for the IUCN Red List guide conservation actions, accuracy and ecological realism are crucial. IUCN’s extent of occurrence (EOO) is the general region including the species’ range, while area of occupancy (AOO) is the subset of EOO occupied by the species. Data‐poor species with incomplete sampling present particular difficulties, but species distribution models (SDMs) can be used to predict suitable areas. Nevertheless, SDMs typically employ abiotic variables (i.e., climate) and do not explicitly account for biotic interactions that can impose range constraints. We sought to improve range estimates for data‐poor, parapatric species by masking out areas under inferred competitive exclusion. We did so for two South American spiny pocket mice: Heteromys australis (Least Concern) and Heteromys teleus (Vulnerable due to especially poor sampling), whose ranges appear restricted by competition. For both species, we estimated EOO using SDMs and AOO with four approaches: occupied grid cells, abiotic SDM prediction, and this prediction masked by approximations of the areas occupied by each species’ congener. We made the masks using support vector machines (SVMs) fit with two data types: occurrence coordinates alone; and coordinates along with SDM predictions of suitability. Given the uncertainty in calculating AOO for low‐data species, we made estimates for the lower and upper bounds for AOO, but only make recommendations for H. teleus as its full known range was considered. The SVM approaches (especially the second one) had lower classification error and made more ecologically realistic delineations of the contact zone. For H. teleus, the lower AOO bound (a strongly biased underestimate) corresponded to Endangered (occupied grid cells), while the upper bounds (other approaches) led to Near Threatened. As we currently lack data to determine the species’ true occupancy within the post‐processed SDM prediction, we recommend that an updated listing for H. teleus include these bounds for AOO. This study advances methods for estimating the upper bound of AOO and highlights the need for better ways to produce unbiased estimates of lower bounds. More generally, the SVM approaches for post‐processing SDM predictions hold promise for improving range estimates for other uses in biogeography and conservation.

Occupation density estimation for noisy high-frequency data

This paper studies the nonparametric estimation of occupation densities for semimartingale processes observed with noise. As leading examples we consider the stochastic volatility of a latent efficient price process, the volatility of the latent noise that separates the efficient price from the actually observed price, and nonlinear transformations of these processes. Our estimation methods are decidedly nonparametric and consist of two steps: the estimation of the spot price and noise volatility processes based on pre-averaging techniques and in-fill asymptotic arguments, followed by a kernel-type estimation of the occupation densities. Our spot volatility estimates attain the optimal rate of convergence, and are robust to leverage effects, price and volatility jumps, general forms of serial dependence in the noise, and random irregular sampling. The convergence rates of our occupation density estimates are directly related to that of the estimated spot volatilities and the smoothness of the true occupation densities. An empirical application involving high-frequency equity data illustrates the usefulness of the new methods in illuminating time-varying risks, market liquidity, and informational asymmetries across time and assets.

A Low-Power Electric-Mechanical Driving Approach for True Occupancy Detection Using a Shuttered Passive Infrared Sensor

Passive infrared (PIR) sensors are the most popular deployed sensors in buildings for individual presence detection. However, PIR sensors are motion detectors in nature, responding only to incident radiation variation, which leads to false negative detections, inaccurate occupancy estimation, and uncomfortable lighting swings and waste of energy. To address this issue, an optical shutter driven by a Lavet motor PIR (LAMPIR) sensor is developed for true presence detection. In comparison with our previously developed chopped PIR (C-PIR) and rotationally chopped PIR (Ro-PIR) sensors, a low-power single-phase electro-mechanical driving approach is introduced for LAMPIR to replace traditional servo and stepper motors and thus significantly reduce the power consumption by up to 89%, size by up to 60%, weight by up to 75%, cost by up to 31%, and acoustic noise by 12 dBA. More specifically, driven by pulsed signal from a microcontroller unit, the electro-mechanical vibrator drives a semi-transparent long-wave infrared optical shutter to chop the field of view (FOV) of a PIR sensor periodically. By monitoring and analyzing the voltage outputs generated by the LAMPIR senor, high-accuracy presence detection can be achieved by optimizing the shutter width and shuttering period through parametric studies. Experimental results show that a classification accuracy of 100% can be reached for detecting stationary occupants up to 4.5 m and moving occupants up to 10 m, suggesting a detection range improvement from both the C-PIR and the Ro-PIR sensors (4.0 m for stationary and 8.0 m for moving occupancy detection for both sensors). Additionally, the LAMPIR sensor has an FOV of 90° in horizontal and 100° in vertical, which is sufficient for most applications. For a 31-h-long presence detection test, an accuracy of 97% is achieved when classifying unoccupied and occupied scenarios, while an accuracy of 93% is achieved when classifying unoccupied, stationary and moving occupant scenarios.

Vindicating “True Occupation” for the Progress of Society

Vindicating “True Occupation” for the Progress of Society

Sampling scales define occupancy and underlying occupancy-abundance relationships in animals.

Occupancy-abundance (OA) relationships are a foundational ecological phenomenon and field of study, and occupancy models are increasingly used to track population trends and understand ecological interactions. However, these two fields of ecological inquiry remain largely isolated, despite growing appreciation of the importance of integration. For example, using occupancy models to infer trends in abundance is predicated on positive OA relationships. Many occupancy studies collect data that violate geographical closure assumptions due to the choice of sampling scales and application to mobile organisms, which may change how occupancy and abundance are related. Little research, however, has explored how different occupancy sampling designs affect OA relationships. We develop a conceptual framework for understanding how sampling scales affect the definition of occupancy for mobile organisms, which drives OA relationships. We explore how spatial and temporal sampling scales, and the choice of sampling unit (areal vs. point sampling), affect OA relationships. We develop predictions using simulations, and test them using empirical occupancy data from remote cameras on 11 medium-large mammals. Surprisingly, our simulations demonstrate that when using point sampling, OA relationships are unaffected by spatial sampling grain (i.e., cell size). In contrast, when using areal sampling (e.g., species atlas data), OA relationships are affected by spatial grain. Furthermore, OA relationships are also affected by temporal sampling scales, where the curvature of the OA relationship increases with temporal sampling duration. Our empirical results support these predictions, showing that at any given abundance, the spatial grain of point sampling does not affect occupancy estimates, but longer surveys do increase occupancy estimates. For rare species (low occupancy), estimates of occupancy will quickly increase with longer surveys, even while abundance remains constant. Our results also clearly demonstrate that occupancy for mobile species without geographical closure is not true occupancy. The independence of occupancy estimates from spatial sampling grain depends on the sampling unit. Point-sampling surveys can, however, provide unbiased estimates of occupancy for multiple species simultaneously, irrespective of home-range size. The use of occupancy for trend monitoring needs to explicitly articulate how the chosen sampling scales define occupancy and affect the occupancy-abundance relationship.

Occupancy surveys with conditional replicates: An alternative sampling design for rare species

Abstract Occupancy models are widely used to describe the distribution of rare and cryptic species—those that occur on only a small portion of the landscape and cannot be detected reliably during a single survey. However, model estimates of occupancy (ψ) and detection probabilities (p) are often least accurate under these circumstances. Available sampling designs for occupancy surveys include standard design, wherein each of S sites is visited K times, and removal design, wherein S sites are visited K times each or until the species of interest is detected. We propose a new conditional design, wherein each of S sites is visited one time, and sites where the species of interest is encountered during the first survey are visited an additional (K−1) times to better estimate detection probability. We used large sample properties of maximum‐likelihood estimators and Markov chain Monte Carlo (MCMC) simulations to characterise our proposed conditional design and compare it to standard and removal designs across a wide range of true occupancy and detection probabilities (ψ, p = 0.1 to 0.9 by 0.1 increments), maximum visits (K) and total sampling effort (E, the number of surveys accrued across all sites). The conditional design provided more accurate estimates (lower standard or root mean squared error) of occupancy than standard or removal designs in our calculations and simulations when species were rare (ψ ≤ 0.3) as well as more accurate estimates of detection probability over most combinations of ψ and p. These low‐occupancy improvements are achieved by expending a greater proportion of effort at occupied sites, improving estimates of p and thus ψ. When species are common (ψ ≥ 0.5) the removal design generally provided the most accurate occupancy estimates, whereas the standard design performed best when ψ was intermediate and during MCMC simulations when p and K were low. We recommend the conditional design for surveys of rare species and pilot studies. For multi‐species surveys that include mixtures of rare and common species, a hybrid standard‐conditional design with 2–3 replicates at all sites and additional replicates at sites where rare species are detected improves occupancy estimates of rare species.

How bad is occupational coding error? A task-based approach

Studies of occupational choice and mobility are often plagued by rampant occupational coding error. Use of task-based occupation measures, such as O*Net, may mitigate the bias caused by coding error if the occupation is misclassified as an occupation similar to the true occupation. Measuring occupational changes in “task space”, I find that task-based measures reduce the problems of coding error, but only slightly. If one does not correct for coding error, one overestimates traditional occupational mobility rates by about 90%; using task-based measures, the overestimate of mobility is still 75%. I also show that when tasks are used as regressors and coding error is not corrected, estimates will be attenuated by 15%–20%. Task-based measures are a slight improvement over census occupation codes but are no panacea for dealing with coding error.

Performance of species occurrence estimators when basic assumptions are not met: a test using field data where true occupancy status is known

Summary Populations are rarely censused. Instead, observations are subject to incomplete detection, misclassification and detection heterogeneity that result from human and environmental constraints. Though numerous methods have been developed to deal with observational uncertainty, validation under field conditions is rare because truth is rarely known in these cases. We present the most comprehensive test of occupancy estimation methods to date, using more than 33 000 auditory call observations collected under standard field conditions and where the true occupancy status of sites was known. Basic occupancy estimation approaches were biased when two key assumptions were not met: that no false positives occur and that no unexplained heterogeneity in detection parameters occurs. The greatest bias occurred for dynamic parameters (i.e. local colonization and extinction), and in many cases, the degree of inaccuracy would render results largely useless. We examined three approaches to increase adherence or relax these assumptions: modifying the sampling design, employing estimators that account for false‐positive detections and using covariates to account for site‐level heterogeneity in both false‐negative and false‐positive detection probabilities. We demonstrate that bias can be substantially reduced by modifications to sampling methods and by using estimators that simultaneously account for false‐positive detections and site‐level covariates to explain heterogeneity. Our results demonstrate that even small probabilities of misidentification and among‐site detection heterogeneity can have severe effects on estimator reliability if ignored. We challenge researchers to place greater attention on both heterogeneity and false positives when designing and analysing occupancy studies. We provide 9 specific recommendations for the design, implementation and analysis of occupancy studies to better meet this challenge.

Mind the gaps: testing for hiatuses in regional radiocarbon date sequences

Long gaps in regional radiocarbon sequences are often considered evidence for occupation hiatuses, but they might also be a product of stochastic processes of occupation and limited numbers of dates. Here we show that, if radiocarbon dates over a span of time are distributed as a Poisson random variable (such that any point in that span has an equal probability of being dated), the gaps between dates will approximate a negative exponential distribution, with many short gaps and a few long ones. Long gaps between dates are to be expected under these conditions, even in the absence of true occupation hiatuses. This exponential distribution of gap lengths is robust even when the uniform probability assumption is relaxed, though true hiatuses have a distinctive, if subtle, signature. We use this model to assess the regional radiocarbon sequence from Qinghai Lake basin, western China, which shows two long possible occupation hiatuses during the period 12,500–4200 14C BP.

Habitat fragmentation reduces occupancy of nest boxes by an open-country raptor

SummaryDespite the recent rapid decline of many grassland bird species, the relative importance of habitat configuration to population persistence is unclear. We used Southeastern American KestrelsFalco sparverius paulusin north-central Florida as a model system to explore the relative influence of landscape structure components on site occupancy patterns at two spatial scales, and for two different time periods. We focused on the dynamic processes of site-level population expansion or contraction. We modelled the occupancy of 131 American Kestrel nest boxes with Bayesian state-space dynamic occupancy models that considered both the partially observed process of true occupancy and the probability of detection of occupancy. We used reversible jump Markov chain Monte Carlo (RJMCMC) algorithms to identify variables that described the continued occupancy of nest boxes, or ϕ, and the probability of colonisation of nest boxes between time periods, or γ3. Changes in open habitat patch isolation at a fine scale, as estimated by the variability of nearest neighbour distance, predicted site colonisation between decades, and patch shape variability was related to ϕ during the early time period (1992–93). We found no strong effects of landscape structure on ϕ during the later time period (2008–2010). We also found no evidence for effects of loss of open habitat on box occupancy or colonization. Our results indicate that continued habitat fragmentation would be deleterious for this threatened subspecies. Additionally, certain land cover management practices recommended for the Florida sandhills, such as frequent low-intensity controlled burns, will likely help conservation attempts.

Measuring indoor occupancy in intelligent buildings using the fusion of vision sensors

In intelligent buildings, practical sensing systems designed to gather indoor occupancy information play an indispensable role in improving occupant comfort and energy efficiency. In this paper, we propose a novel method for occupancy measurement based on the video surveillance now widely used in buildings. In our method, we analyze occupant detection both at the entrance and inside the room. A two-stage static detector is presented based on both appearances and shapes to find the human heads in rooms, and motion-based technology is used for occupant detection at the entrance. To model the change of occupancy and combine the detection results from multiple vision sensors located at entrances and inside rooms for more accurate occupancy estimation, we propose a dynamic Bayesian network-based method. The detection results of each vision sensor play the role of evidence nodes of this network, and thus, we can estimate the true occupancy at time t using the evidence prior to (and including) time t. Experimental results demonstrate the effectiveness and efficiency of the proposed method.

Range margin shifts of birds revisited – the role of spatiotemporally varying survey effort

Global climate warming is predicted to lead to global and regional changes in the distribution of organisms. One influential approach to test this prediction using temporally repeated mapping surveys of organisms was suggested in a seminal paper by Thomas & Lennon (1999, Nature). The Thomas & Lennon approach corrects observed changes in the range margin for changes in the range size, and thus potentially controls for other broad-scale environmental changes between surveys, however the approach does not necessarily account for potential biases in sampling effort. To verify whether the issue of variation in sampling effort affects empirical estimates of shifts in range margin, we reanalyzed all three published studies exploring range margin changes of breeding birds in Great Britain (GB), Finland, and New York State (NY). Accounting for changes in survey effort on range margins lowered the estimated shift for breeding birds in New York, but the shift remained statistically significant. For Great Britain and Finland, for which no direct estimate of survey effort is available, we used species richness (a strong correlate of survey effort in New York) as a proxy and found that in both cases the estimated shift in range margin was significantly reduced and became nonsignificant. To understand how robust the approach is to sampling biases, we use a simulation model to show that the Thomas & Lennon approach is, under certain conditions, sensitive to changes in detection probability (probability to detect true occupancy) which in turn may be affected by changes in surveying effort between surveys. We thus found evidence that temporal changes in the distribution of breeding birds based on repeated mapping surveys may be inflated by changes in survey effort along range boundaries. We discuss possible approaches to deal with this issue in the analysis and design of national or regional surveys.