Variation In Detection Probability Research Articles

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.

Experimental feeding validates nanofluidic array technology for DNA detection of ungulate prey in wolf scats

AbstractThe study of carnivores' diet is a key component to enhance knowledge on the ecology of predators and their effect on prey populations. Although molecular approaches to detect prey DNA in carnivore scats are improving, the validation of their accuracy, a prerequisite for reliable applications within ecological frameworks, is still lagging behind the methodological advances. Indeed, variation in detection probability among prey species can occur, representing a potentially insidious source of bias in food‐habit studies of carnivores. Calibration of DNA‐based methods involves the optimization of specificity and sensitivity and, whereas priority is usually given to the former to avoid false positives, sensitivity is rarely investigated so that false negatives may be largely overlooked. We conducted feeding trials with captive wolves (Canis lupus) to validate a nanofluidic array technology recently developed for the detection of multiple prey species in scats. Using 371 scat samples from 12 wolves fed with a single‐prey diet, the sensitivity of our nanofluidic array method varied between 0.45 and 0.95 for the six main ungulate prey species. The method sensitivity was enhanced by using multiple markers per species and by a relatively low threshold of number of amplifying markers required to confirm a detection. Yet, at least two markers should be used to avoid false positives. By acknowledging sources of bias in sensitivity to reliably interpret the results of DNA‐based dietary methods, our study highlights the relevance of feeding experiments to optimally calibrate the relative thresholds to define a positive detection and investigate the occurrence and extent of biases in sensitivity.

Increasing sampling efficiency of Bombus communities with rare and endangered species by optimizing detection probabilities: A multi-species occupancy modelling approach using roadsides as a case study

Nearly one-quarter of bumble bee species assessed by the International Union for the Conservation of Nature are considered to be in decline, necessitating monitoring to support recovery planning. A key element of robust wildlife monitoring is accounting for imperfect detection in the observation process and efforts to understand sources of detection uncertainty in bumble bee surveys began recently in response to the listing of the rusty patched bumble bee, Bombus affinis, as federally endangered in the U.S. and Canada. We investigated within-community variation in detection probabilities among eight species that vary in occupancy, including B. affinis. We sampled bumble bees along roadsides in the greater metropolitan area of Minneapolis-St. Paul, Minnesota in the United States in 2018. We generated estimates of occupancy, detection probability, and cumulative detection probability from a multispecies occupancy model using a Bayesian framework that included covariates on occupancy and detection. Covariates selected a priori included floral cover and impervious surface effects on occupancy, and date, time of day, and observer effects on detection. We found that Bombus species with low occurrence are not necessarily less detectable than common species with higher occurrence, and that species reach their peak detection probability at different dates. Detection was higher in the afternoon and varied among observers. Well-designed Bombus monitoring programs can reduce sampling effort and optimize detection by sampling on dates near species detection peaks, sampling in the afternoon, and using cumulative detection probabilities to determine optimum sampling effort.

A spatially explicit model for density that accounts for availability: a case study withMojavedesert tortoises

AbstractEstimating population density and identifying those areas where density is changing through time are central to prioritizing conservation and management strategies. Obtaining reliable estimates of density and trends can be challenging, however, especially for long‐lived species that are rare, have broad geographic distributions, and are difficult to detect reliably during field surveys. We developed a hierarchical model for distance‐sampling data that characterizes spatial variation in density at two scales and simultaneously estimates regional trends while accounting for variation in detection probability and availability across surveys. We applied the model to data collected over a 20‐year period (2001–2020) in an area that encompassed most of the geographic range of the Mojave desert tortoise (Gopherus agassizii). Density of adult tortoises varied with multiple biotic and abiotic features, including topography, aspect, geology, and seasonal precipitation and temperature regimes. Across the entire period and study area, the density of adult tortoises decreased by an average of 1.8% per year (95% CI = −3.5% to −0.2%). Trends varied geographically, however, with the steepest declines in the western part of the range (−4.1%, −6.9% to −1.3%). Accounting for habitat loss across our study area, the abundance of this threatened species declined by an estimated 129,000 adults (36%) between 2001 and 2020. Our modeling approach extends traditional distance‐sampling frameworks by accounting for ecological and observational processes that could mask spatiotemporal variation in density and, at the same time, provides spatially explicit estimates to guide conservation and management strategies for tortoises and other rare species.

Pooling robustness in distance sampling: Avoiding bias when there is unmodelled heterogeneity

The pooling robustness property of distance sampling results in unbiased abundance estimation even when sources of variation in detection probability are not modeled. However, this property cannot be relied upon to produce unbiased subpopulation abundance estimates when using a single pooled detection function that ignores subpopulations. We investigate by simulation the effect of differences in subpopulation detectability upon bias in subpopulation abundance estimates. We contrast subpopulation abundance estimates using a pooled detection function with estimates derived using a detection function model employing a subpopulation covariate. Using point transect survey data from a multispecies songbird study, species-specific abundance estimates are compared using pooled detection functions with and without a small number of adjustment terms, and a detection function with species as a covariate. With simulation, we demonstrate the bias of subpopulation abundance estimates when a pooled detection function is employed. The magnitude of the bias is positively related to the magnitude of disparity between the subpopulation detection functions. However, the abundance estimate for the entire population remains unbiased except when there is extreme heterogeneity in detection functions. Inclusion of a detection function model with a subpopulation covariate essentially removes the bias of the subpopulation abundance estimates. The analysis of the songbird point count surveys shows some bias in species-specific abundance estimates when a pooled detection function is used. Pooling robustness is a unique property of distance sampling, producing unbiased abundance estimates at the level of the study area even in the presence of large differences in detectability between subpopulations. In situations where subpopulation abundance estimates are required for data-poor subpopulations and where the subpopulations can be identified, we recommend the use of subpopulation as a covariate to reduce bias induced in subpopulation abundance estimates.

Coverage analysis and a new metaheuristic approach using the Elfes Probabilistic detection model in Wireless sensor networks

The dynamic deployment of sensors in AoI (Area of Interest) plays an important role in ensuring the quality of services (QoS) of Wireless Sensor Networks (WSNs) by optimising the coverage and lifetime of the network. In this paper, a new dynamic deployment approach using metaheuristics based on Electromagnetism-Like (EM-L) algorithm and Elfes Probabilistic Detection Model (EPDM) was proposed to optimise the coverage of WSNs based on the sensor coverage problem, and coverage analysis of AoI was performed. A variable threshold detection probability (TDP) was defined instead of defining a fixed TDP as in the literature. Thus, a more realistic modelling environment was created by considering the signal-to-noise ratio (SNR) in the coverage calculation. The simulation results show that the sensors are always effectively deployed in different scenarios with variable TDPs by the proposed approach compared to a random distribution.

Power to the people: Analysis of occupancy models informed by local knowledge

AbstractPower analyses help to improve the cost‐effectiveness of monitoring strategies for wildlife populations, but rarely account for variation in detection probability, affecting the power of data to detect trends in occupancy. We explore the power of occupancy models informed by two locally‐informed methods (interviews and daily diaries) to detect changes in occupancy for 14 mammal species hunted for wild meat within a community forest in Cameroon. This is the first study to use the formula developed by to compare power between locally‐informed methods and camera traps, and identify the monitoring strategies best suited to different species. Comparable effort is required between the three methods to detect 50% as 80% change in occupancy, except where occupancy is less than 0.13 (diary data), 0.03 (camera), or 0.6 (interviews). Overall, where occupancy <0.54, 200 sites and four repeat visits were required to detect at least a 30% change in occupancy. Achieving power to detect any level of change useful for conservation planning is often not viable for projects with small budgets and for species with very low detection rates. However, some species of conservation importance (e.g., gorilla, chimpanzee) are better detected and as such could be monitored using data collected in collaboration with local communities.

Quantifying the relationship between prey density, livestock and illegal killing of leopards

AbstractMany large mammalian carnivores are facing population declines due to illegal killing (e.g. shooting) and habitat modification (e.g. livestock farming). Illegal killing occurs cryptically and hence is difficult to detect. However, reducing illegal killing requires a solid understanding of its magnitude and underlying drivers, while accounting for the imperfect detection of illegal killing events. Despite the importance of illegal killing of large carnivores in comparison with other causes of mortality, its relationship with potential drivers such as livestock density and wild prey abundance is rarely described.Using ranger‐collected data (2007–2019) of leopard killing events and data on covariates (livestock density, wild prey abundance, road length, protected area size, elevation) across Iran, we applied a single‐visit N‐mixture model to jointly model variation in detection probability and expected annualized number of leopard killing events.Over the study period, we estimated 428 leopard mortalities (95% CI 184 to 1,014), which was 45% larger than the observed number. Expected intensity of leopard killing was positively related to protected area size, livestock density and wild prey abundance. Detection of leopard killing was higher in areas with more developed road networks.Synthesis and applications. Ranger‐based monitoring data on poaching of carnivores are cost effective, but traditional analysis does not take into account imperfect detection. We show that innovative statistics (single‐visit N‐mixture modelling) can reliably quantify poaching events and address their drivers, at large geographical scales. We used the example of the Persian leopard across Iran, but our approach is also applicable to understand killing dynamics of other species. Results suggest that a high frequency of leopard killing is likely to occur in areas with >100 livestock per km2and >450 individuals of wild prey per km2. This highlights the need for improved management of livestock grazing and effective measures around high‐risk protected areas to mitigate human–leopard conflict and reduce killing of leopards.

Monitoring Occupancy of Bats with Acoustic Data: Power and Sample Size Recommendations

Bats are difficult to study due to their nocturnal, cryptic, and highly vagile nature. Ongoing advances in acoustic recording hardware and call classification software have made species detection and activity monitoring more feasible. Our objectives were to determine the effort necessary to monitor bat assemblages using an occupancy framework and acoustic data and to provide guidelines for researchers interested in developing similar monitoring programs. We collected data at 2 study areas in South Texas from June through September in 2015, 2016, and 2017. We used Pettersson D500X Mk II real-time full-spectrum detectors and classified sound files using SonoBat bat call analysis software. We attempted to collect data during 2 visits to individual sites, with up to 5 consecutive nights per visit each year. We estimated occupancy rates for each species in each study area using occupancy models in Program MARK and included terms to define trends in detection probability through the season. Over the 3 years of our study, we sampled 106 sites with 803 sampling nights and classified a total of 2880 sound files to 7 species. Data sets for 6 of the species supported models indicating that detection probability varied throughout our sampling period. Our results generally indicate that sample sizes between 10 and 20 sites would be required to detect declines in occupancy of 50% over 25 years using 10 nights per site with a starting occupancy rate of 0.70. Detecting declines of 30% in 10 years may require >75 sampling sites. Finally, our analysis shows that recognizing seasonal variation in detection probability, and then timing surveys accordingly, can greatly reduce sample size requirements.

Variation in herpetofauna detection probabilities: implications for study design

Population monitoring is fundamental for informing management decisions aimed at reducing the rapid rate of global biodiversity decline. Herpetofauna are experiencing declines worldwide and include species that are challenging to monitor. Raw counts and associated metrics such as richness indices are common for monitoring populations of herpetofauna; however, these methods are susceptible to bias as they fail to account for varying detection probabilities. Our goal was to develop a program for efficiently monitoring herpetofauna in southern Texas. Our objectives were to (1) estimate detection probabilities in an occupancy modeling framework using trap arrays for a diverse group of herpetofauna and (2) to evaluate the relative effectiveness of funnel traps, pitfall traps, and cover boards. We collected data with 36 arrays at 2 study sites in 2015 and 2016, for 2105 array-days resulting in 4839 detections of 51 species. We modeled occupancy for 21 species and found support for the hypothesis that detection probability varied over our sampling duration for 10 species and with rainfall for 10 species. For herpetofauna in our study, we found 14 and 12 species were most efficiently captured with funnel traps and pitfall traps, respectively, and no species were most efficiently captured with cover boards. Our results show that using methods that do not account for variations in detection probability are highly subject to bias unless the likelihood of false absences is minimized with exceptionally long capture durations. For monitoring herpetofauna in southern Texas, we recommend using arrays with funnel and pitfall traps and an analytical method such as occupancy modeling that accounts for variation in detection.

Consequences of ignoring variable and spatially autocorrelated detection probability in spatial capture-recapture

ContextSpatial capture-recapture (SCR) models are increasingly popular for analyzing wildlife monitoring data. SCR can account for spatial heterogeneity in detection that arises from individual space use (detection kernel), variation in the sampling process, and the distribution of individuals (density). However, unexplained and unmodeled spatial heterogeneity in detectability may remain due to cryptic factors, both intrinsic and extrinsic to the study system. This is the case, for example, when covariates coding for variable effort and detection probability in general are incomplete or entirely lacking.ObjectivesWe identify how the magnitude and configuration of unmodeled, spatially variable detection probability influence SCR parameter estimates.MethodsWe simulated SCR data with spatially variable and autocorrelated detection probability. We then fitted an SCR model ignoring this variation to the simulated data and assessed the impact of model misspecification on inferences.ResultsHighly-autocorrelated spatial heterogeneity in detection probability (Moran’s I = 0.85–0.96), modulated by the magnitude of the unmodeled heterogeneity, can lead to pronounced negative bias (up to 65%, or about 44-fold decrease compared to the reference scenario), reduction in precision (249% or 2.5-fold) and coverage probability of the 95% credible intervals associated with abundance estimates to 0. Conversely, at low levels of spatial autocorrelation (median Moran’s I = 0), even severe unmodeled heterogeneity in detection probability did not lead to pronounced bias and only caused slight reductions in precision and coverage of abundance estimates.ConclusionsUnknown and unmodeled variation in detection probability is liable to be the norm, rather than the exception, in SCR studies. We encourage practitioners to consider the impact that spatial autocorrelation in detectability has on their inferences and urge the development of SCR methods that can take structured, unknown or partially unknown spatial variability in detection probability into account.

Estimating Abundance and Occupancy of Northern Barrens Tiger Beetles in an Isolated New York Population

Reliable monitoring of abundance of rare species across space and time is critical to determining their status and risk of extinction. For Cicindela patruela patruela (Notrthern Barrens Tiger Beetle), we determined whether mark–recapture estimates of abundance or site occupancy probability were sufficiently precise, unbiased, and efficient to be used for long-term monitoring of population status for this imperiled species in the Shawangunk Mountains, NY. Over 2 years, the estimated population size grew from 81 ± 5.7 to 109 ± 6.4 individuals. Consistent with a space-limited but potentially growing population, seasonal occupancy models (3-season model with year = season) indicated site-colonization probability remained constant, site extinction probability decreased, and detection probability increased among years—the latter indicating the need to account for variation in detection probability when monitoring population trends. Both occupancy and mark–recapture methods were promising for monitoring change in populations of Northern Barrens Tiger Beetle in this region.

Using PIT-tags and portable antennas for quantification of fish movement and survival in streams under different environmental conditions.

While PIT-tag tracking using mobile antennas is being increasingly used to study fish movement and survival in streams, little is known about the limitations of the method, especially over longer periods of time and under different environmental settings. We used 6 years of data combining tagging, mobile antenna tracking and recaptures of Salmo trutta in multiple small streams in the Lake Lucerne drainage area in Switzerland to evaluate the relative importance of different environmental and intrinsic factors affecting the efficiency of the method. Our study system and experimental design allowed us to accurately verify the continuous presence and survival of recaptured fish in the stream after tracking, which meant that we could estimate detection probability with high confidence. The mean detection probability of tagged trout was 43%, but we found that fish length had a strong negative effect on detection probability, especially in males. Multivariate axes of stream environmental features did not predict efficiency but stream width alone was significantly positively correlated with efficiency. Additionally, stream temperature when tracking had a positive effect on fish detectability. Tag loss at recapture was globally rare (<8%) but common in large postspawn females (>30%). Based on the escape response of fish after detection, we could estimate the proportion of ghost tags, which reached a plateau of around 80% 2 years after tagging. We finally showed that our models of tag loss, fish detection and escape response are needed to interpret detection events. Our results highlight that individual variation in detection probability and tag loss is high and has to be considered for analysis.

Probabilities of detecting submersed aquatic vegetation species using a rake method may vary with biomass

Levels of submersed aquatic vegetation (SAV) are commonly assessed using a modified garden rake. However, the utility of the rake sampling method relative to methods that are typically viewed as more definitive (and expensive) such as snorkeling and coring remains a matter of debate. This study explores whether probabilities of species detections for four SAV species varied among sampling units in a rake-biomass study and, if so, whether such variation reflected variation in species abundance. Variation in detection probabilities, when unaddressed, may yield biased estimators of percent frequency of occurrence (“occupancy”) and of occurrence-habitat associations. Biomass-driven variation in detection probabilities is important because such variation may not be explainable using covariates typically measured when sampling using the rake method. This study found substantial among-unit variation in detection probabilities, with majorities of that variation on the logit or modeling scale being associated with biomass but not with the non-biomass covariates substrate type, water depth and day of study. The study closes by exploring sampling protocols and modeling methods that may yield improved SAV occupancy estimates.

Size Matters: the Influence of Trap and Mesh Size on Turtle Captures

ABSTRACTMethods used in wildlife ecology can influence population‐ and community‐level estimates, such as species richness, sex ratio, age and size structure, occupancy and detection probabilities, and community composition. Various trapping and sampling biases exist for freshwater turtles including bait and trap choice and survey technique. To date, no study has investigated the influence of hoop net and mesh size on various population‐ and community‐level estimates. Here, we use detection models to determine if trap and mesh size influence detection probability of nine species of freshwater turtles over 3 consecutive years (2016–2018) in west Tennessee. Additionally, we use multivariate models to determine if freshwater turtle community composition was influenced by hoop trap and mesh sizes. Our results indicated that there was a bias related to mesh size in detection probabilities and community composition. Smaller mesh sized traps were better at detecting smaller‐bodied turtle species, which then changed community species richness but not catch‐per‐unit‐effort estimates (i.e., abundance). Additionally, larger mesh sized traps were better at detecting common snapping turtles (Chelydra serpentina), which supports previous research. Our results suggest that researchers should account for the variation in detection probabilities by mesh size when conducting mark‐recapture and occupancy analyses. Moreover, erroneous inferences about population trends and changes in diversity within turtle communities through time could cause managers to misidentify population declines and conservation value of a site. © 2021 The Wildlife Society.

Evaluating a herpetofaunal monitoring program in the Rolling Plains of Texas.

We evaluated the herpetofaunal monitoring program on a state-owned property in the Rolling Plains ecoregion of Texas to determine whether the current study design and survey effort are meeting the following monitoring objectives: (1) inventory herpetofaunal diversity on the property, (2) improve understanding of herpetofaunal habitat preferences, and (3) track herpetofaunal response to land management activities. We used the iChao1 species richness estimator to evaluate the sample completeness of the current dataset, which was collected intermittently from 2004 to 2016. Species richness was evaluated by site and year as well as pooled across sites and years. Observed species richness across all sites and years was 23 species, and estimated richness was 23.999 (23.090-34.052, 95% CI). Observed species richness for each year was less than the cumulative total, with 20 species observed in 2004, 6 species observed in 2013, 11 species observed in 2014, and 19 species observed in 2016. The estimated species richness for individual years of sampling ranged from 6.250 to 39.320 species. Our analysis suggests that inadequate sampling design and effort prevents all three objectives from being addressed using the existing dataset. Necessary improvements to the monitoring program to meet all objectives include increasing the spatial and temporal scale of the monitoring effort and utilizing a random sampling protocol. Monitoring methods that address variation in detection probability, such as occupancy and mark-recapture, will also need to play a role. Our findings show that monitoring programs require substantial investment in materials and personnel to obtain datasets sufficient for rigorous biodiversity monitoring.

Change in global transmission rates of COVID-19 through May 6 2020.

We analyzed COVID-19 data through May 6th, 2020 using a partially observed Markov process. Our method uses a hybrid deterministic and stochastic formalism that allows for time variable transmission rates and detection probabilities. The model was fit using iterated particle filtering to case count and death count time series from 55 countries. We found evidence for a shrinking epidemic in 30 of the 55 examined countries. Of those 30 countries, 27 have significant evidence for subcritical transmission rates, although the decline in new cases is relatively slow compared to the initial growth rates. Generally, the transmission rates in Europe were lower than in the Americas and Asia. This suggests that global scale social distancing efforts to slow the spread of COVID-19 are effective although they need to be strengthened in many regions and maintained in others to avoid further resurgence of COVID-19. The slow decline also suggests alternative strategies to control the virus are needed before social distancing efforts are partially relaxed.

Assessing Population Viability of Black Bears using Spatial Capture‐Recapture Models

ABSTRACTThe Central Georgia Bear Population (CGP) is the least abundant and most isolated of Georgia's 3 American black bear (Ursus americanus) populations. Beginning in 2011, changes to regulations governing harvest of the CGP resulted in an increase in female bear harvest, creating concern that future harvest could be an important influence on population viability. Hence, our objective was to assess viability of the CGP under various levels of female mortality. During 2012–2016, we used barbed‐wire hair snares to collect bear hair samples from within the range of the CGP in Georgia, USA. We used microsatellite genotyping to identify individual bears and created robust‐design, spatial detection histories for all female bears detected. We fit open population spatial capture‐recapture (SCR) models to the detection histories in a Bayesian framework. We used the Widely Applicable Information Criterion (WAIC) to rank models that varied with respect to sources of variation in detection probability, survival, and per capita recruitment, and used the model with the lowest WAIC to forecast dynamics of the CGP 50 years into the future under various levels of female mortality. We assessed the 50‐year extinction probability under a continuation of mortality levels documented during 2012–2016, and under incremental increases in female mortality above this baseline. The top model included density‐dependent per capita recruitment, annual variation in detection probability, and a trap‐level behavioral response. Abundance increased from 106 (95% CI = 86–132) females in 2012 to 136 (95% CI = 113–161) females in 2013 and remained relatively stable thereafter. Annual female survival was 0.75 (95% CI = 0.69–0.82) and did not vary among years. The per capita recruitment rate decreased over time as density increased, and was 0.49 (95% CI = 0.33–0.66) during the first time interval and 0.29 (95% CI = 0.20–0.38) during the final time interval. Annual growth rate () was 1.28 (95% CI = 1.07–1.52) between 2012 and 2013 but decreased throughout the study, ending at 1.04 (95% CI = 0.93–1.17). Forecasts indicated continuation of the female mortality levels experienced from 2012–2016 were sustainable over 50 years, with the estimated extinction risk being &lt;0.001%. Increasing annual harvest by 5 females introduced a negligible increase in the 50‐year probability of extinction, but harvesting an additional 10 females/year caused extinction risk to rise to 1.15%. We recommend that harvest regulations are structured such that mortality rates remain at current levels or do not increase by more than an annual average of 5 females above levels observed during our study. Furthermore, we recommend that managers continue to monitor the population so that harvest regulations and population models can be refined over time. © 2020 The Wildlife Society.

Oceanic Drivers of Sei Whale Distribution in the North Atlantic

This study investigated the oceanic drivers of sei whale (Balaenoptera borealis) distribution in the central and eastern North Atlantic, and explored how distribution may have changed over almost three decades. Cetacean sightings data were available from Icelandic, Faroese and Norwegian surveys conducted throughout the central and eastern North Atlantic during summer between 1987 and 2015. Effective strip half width was estimated from the data to take account of variation in detection probability. Spatially-referenced environmental variables used as predictors in generalised additive models of sei whale relative density included: relief-related variables seabed depth, slope and aspect; monthly-varying physical oceanographic variables sea surface temperature (SST), mixed layer depth, bottom temperature, salinity, and sea surface height anomaly (SSH); and monthly-varying biological oceanographic variables chlorophyll-a concentration and primary productivity. Preliminary analysis considered which month (March-August) in the dynamic oceanographic variables explained most variability in sei whale density. Models including all variables (“full models”) could only be run for 1998-2015 because data for several variables were missing in earlier years. “Simple models" including only relief-related variables and SST were therefore run for 1987-89, and also for 1998-2015 for comparison. The best-fitting full model for 1998-2015 retained the covariates depth, May SST, May bottom temperature, July salinity, July SSH and July primary productivity. Of these, depth, May SST and July SSH were the strongest predictors of sei whale density. In the simple models for both 1987-89 and 1998-2015, depth (especially), May SST and seabed slope were the strongest predictors of sei whale density. The highest densities of sei whales were predicted in the Irminger Sea and over the Charles-Gibbs Fracture Zone; a pattern driven by large negative SSH, deep water (&gt;1500m) and polar-temperate SST (5-12oC). There was some inter-annual variability in predicted distribution and there appears to be a northward expansion in distribution consistent with prey species responding to ocean warming. The models could be used to predict future distribution of sei whales based on future environmental conditions predicted by climate models.

Seagrass canopies and the performance of acoustic telemetry: implications for the interpretation of fish movements

BackgroundAcoustic telemetry has been used with great success to quantify the movements of marine fishes in open habitats, however research has begun to focus on patterns of movement and habitat usage within more structurally complex habitats. To date, there has been no detailed assessment of the performance of acoustic telemetry within seagrass, which forms a crucial nursery and foraging habitat for many fish species globally. Information on the detection range of acoustic receivers within seagrass is essential to guide receiver array design, particularly positioning systems. Here, we compare detection ranges for transmitters (Vemco V7) within and above the seagrass to determine impacts on the performance of a Vemco Positioning System (VPS). We also investigate the influence of environmental conditions (i.e. wind, time of day, background noise, atmospheric pressure and depth) on detection probability.ResultsThe performance of the VPS declined dramatically when the transmitters were positioned within the seagrass (positional accuracy = 2.69 m, precision = 0.9 m, system efficiency (i.e. the proportion of successful positions) = 5.9%) compared to above the canopy (positional accuracy = 2.21 m, precision = 0.45 m, system efficiency = 30.9%). The reduction in VPS efficiency when transmitters were within seagrass was caused by a decline in the detection range of receivers (range of 50% detections) from 85 to 40 m, as this limited the ability of the three receivers to simultaneously detect transmissions. Additionally, no detections were recorded for the transmitters within seagrass at a distance greater than 150 m from the receiver. Increasing wind speed from 0 to 50 km h−1 correlated with a 15% reduction in detections while detection probability decreased from 0.8 during the day to 0.55 at night, due to higher in-band noise (69 kHz).ConclusionsOur findings demonstrate that tagged fish ensconced within seagrass are unlikely to be detected by receivers or positioned by a VPS. Further, we demonstrate that wind conditions and the time of day create temporal variation in detection probability. These findings highlight the need for telemetry studies to perform in situ range testing and consider how fish use vegetated habitats such as seagrasses when positioning receivers and interpreting data.