Costs of detection bias in index�based population monitoring

Biological Conservation, 2013

Prior scientific knowledge inspires ecological research, hypotheses and debate but is rarely used explicitly to formulate predictive models. Bayesian statistics provide a formal way to include informative priors and evaluate their influence on parameter estimates. We use case studies of the influence of overabundant deer on bird species abundance in the Gulf Island, San Juan and Haida Gwaii archipelagos of western North America to demonstrate the utility of informative priors and Bayesian modelling to determine the consequences of overabundance. We found that by including informative priors about deer browsing impacts on bird species from a study undertaken in Haida Gwaii, the precision of estimates from a similar study undertaken in the Gulf and San Juan archipelagos could be significantly increased. Uncertainty about regional ecological impacts underpins many agencies failure to take management actions. We demonstrate here, that informative priors, when used logically and transparently, can be a highly cost effective way to increase understanding of ecological processes. In some cases, it may be the only way to inform decision-making when scarce resources limit support for long term field research or the threat is sufficiently great that immediate action is required. For several bird species examined here, the inclusion of informative priors strengthened the conclusion that their populations were negatively affected by changes in vegetation structure caused by deer browsing. Our findings suggest that deer browsing in these island archipelagos must be managed if the risk of local extinctions among native flora and fauna is to be avoided.

Journal of Ornithology

Effective integration of information in conservation decision making requires explicit consideration of conservation goals and objectives, specification of a range of potential actions, and the willingness to use information to assist in making and improving decisions. If these conditions are met, monitoring programs would benefit from the creative interplay between predictions, designs and models. Prediction, a basic component of scientific endeavor, is also a key component of scientific decision-making. Effective application of information to decision making typically requires integration of several types of information in a common framework, including “found” data and retrospective studies, innovative sampling designs, and the use of hierarchical data structures (e.g., demographic studies nested with occupancy sampling and analysis of community structure). Sampling designs can also include quasi-experimentation, in which confounding factors are accounted for by spatial or temporal “controls” or covariates. Hierarchical and state-space modeling, often most effectively performed in a Bayesian framework, provide a unified modeling structure for such designs and data. We illustrate these ideas with the problem of investigating and mitigating the effects of climate change on terrestrial birds in North America. Ecological theory and available data provide predictions about the impacts of global, regional, and local climate changes on avian communities. We outline a hierarchical sampling design providing for assessment and monitoring of key state variables. The design integrates broad-scale metrics such as species richness and turnover from existing monitoring programs, with directed monitoring using combinations of occupancy and capture–mark–recapture sampling to address specific questions of local extinction and colonization as well as demographic rates in relation to latitudinal and altitudinal gradients. The design would incorporate temporal, spatial, and management controls (presence/absence of selective interventions) as feasible, and manipulative experiments such as supplemental feeding to test specific predictions at specific sites. Data would be incorporated into a hierarchical modeling structure directed specifically at informing and updating the predictive management model.

Biological Conservation

Population monitoring traditionally relies on population counts, accounting or not for the issue of detectability. However, this approach does not permit to go into details on demographic processes. Therefore, Capture-Recapture (CR) surveys have become popular tools for scientists and practitioners willing to measure survival response to environmental change or conservation actions. However, CR surveys are expensive and their design is often driven by the available resources, without estimation about the level of precision they provide for detecting changes in survival, despite optimising resource allocation in wildlife monitoring is increasingly important. Investigating how CR surveys could be optimised by manipulating resource allocation among different design components is therefore critically needed. We have conducted a simulation experiment exploring the statistical power of a wide range of CR survey designs to detect changes in the survival rate of birds. CR surveys differ in terms of number of breeding pairs monitored, number of offspring and adults marked, resighting effort and survey duration. We compared open-nest (ON) and nest-box (NB) monitoring types, using medium-and long-lived model species. Increasing survey duration and number of pairs monitored increased statistical power. Long survey duration can provide accurate estimations for long-lived birds even for small population size (15 pairs). A costbenefit analysis revealed that for long-lived ON species, ringing as many chicks as possible appears as the most effective survey component, unless a technique for capturing breeding birds at low cost is available to compensate for reduced local recruitment. For medium-lived NB species, focusing the NB rounds at a period that maximises the chance to capture breeding females inside nest-boxes is more rewarding than ringing all chicks. We show that integrating economic costs is crucial when designing CR surveys and discuss ways to improve efficiency by reducing duration to a time scale compatible with management and conservation issues.

2009

Effective conservation decision-making requires robust estimates of population trends. It is often assumed that, as long as monitoring methods remain consistent over time, trends in relative abundance are valid proxies for actual abundance. However, if the bias and uncertainty of relative abundance estimates change over time, this can have a serious impact on the validity of monitoring programmes. We developed a simple model for the retrospective assessment of the likely error and bias in abundance estimates from aerial surveys of the saiga antelope. Due to dramatic reductions in group size and density, current estimates of abundance are probably substantially lower than the true population size, and the level of uncertainty surrounding these estimates precludes their use for monitoring trends. This has implications for the Government of Kazakhstan's ability to monitor progress towards their agreed conservation goals. The method is potentially widely applicable to species for which historical data on relative abundance and group size are available.

Avian Conservation and Ecology, 2019

An important metric for many aspects of species conservation planning and risk assessment is an estimate of total population size. For landbirds breeding in North America, Partners in Flight (PIF) generates global, continental, and regional population size estimates. These estimates are an important component of the PIF species assessment process, but have also been used by others for a range of applications. The PIF population size estimates are primarily calculated using a formula designed to extrapolate bird counts recorded by the North American Breeding Bird Survey (BBS) to regional population estimates. The extrapolation formula includes multiple assumptions and sources of uncertainty, but there were previously no attempts to quantify this uncertainty in the published population size estimates aside from a categorical data quality score. Using a Monte Carlo approach, we propagated the main sources of uncertainty arising from individual components of the model through to the final estimation of landbird population sizes. This approach results in distributions of population size estimates rather than point estimates. We found the width of uncertainty of population size estimates to be generally narrower than the order-of-magnitude distances between the population size score categories PIF uses in the species assessment process, suggesting confidence in the categorical ranking used by PIF. Our approach provides a means to identify species whose uncertainty bounds span more than one categorical rank, which was not previously possible with the data quality scores. Although there is still room for additional improvements to the estimation of avian population sizes and uncertainty, particularly with respect to replacing categorical model components with empirical estimates, our estimates of population size distributions have broader utility to a range of conservation planning and risk assessment activities relying on avian population size estimates.

2005

Planning for bird conservation has become increasingly reliant on remote sensing, geographical information systems, and, especially, models used to predict the occurrence of bird species as well as their density and demographics. We address the role of such tools by contrasting two models used in bird conservation. One, the Mallard (Anas platyrhynchos) productivity model, is very detailed, mechanistic, and based on an enormous body of research. The Mallard model has been extensively used with success to guide management efforts for Mallards and certain other species of ducks. The other model, the concept of Bird Conservation Areas, is simpler, less mechanistic, and less wellgrounded in research. This concept proposes that large patches of suitable habitat in a proper landscape will be adequate to maintain populations of birds. The Bird Conservation Area concept recently has been evaluated in the northern tallgrass prairie, where its fundamental assumptions have been found not to hold consistently. We argue that a more comprehensive understanding of the biology of individual species, and how they respond to habitat features, will be essential before we can use remotely sensed information and geographic information system products with confidence.

Oryx, 2010

H e i k o U . W i t t m e r , P a u l o C o r t i , C r i s t i Á n S a u c e d o and J o s É L u i s G a l a z Abstract Considerable efforts have been invested in recent years to improve methods for both data collection and analyses required for population monitoring. Where historical or current estimates of population size are not adjusted for detection probabilities they may be too inaccurate to provide meaningful estimates of trends and thus monitoring methods need to be adapted. Here, we use data from the Endangered huemul deer Hippocamelus bisulcus to outline a framework to develop accurate robust estimates of detection probabilities that can be incorporated into new surveys in a cost-effective way and applied to existing survey data sets. In particular, by retroactively estimating detection probabilities for surveys of huemul, we show that current survey methods for huemul are inadequate to determine population trends reliably. Based on these results we propose a new monitoring method for the huemul and discuss the importance of estimating accuracies of historical survey data to ensure that changes in the abundance of the species reflect real population trends and are not an artefact of variation over time in the accuracy of survey data.

2000

Biological Conservation, 2011

Prioritizing management actions for wildlife conservation is a difficult task due to the large number of problems relative to available conservation resources and uncertainty about the benefits arising from numerous potential management actions. In this study we use a cost-efficiency protocol to evaluate and prioritize eight different management actions for waterbird community in wetlands throughout southeastern Spain. The protocol generated an action priority ranking based on the costs and predicted benefits of the actions in terms of waterbird carrying capacity. Action prioritization outcomes were also evaluated using population viability analysis models for two of the study species. Removal of dead bird carcasses to prevent disease outbreaks was identified as the most cost efficient action. Removing lead pellets from the sediment was the least efficient strategy. Our approach highlights the role of detailed risk assessment as a form of quality control on the simpler prioritization protocols. We recommend a twostep prioritization protocol based on (i) a rapid, usually simpler prioritization approach for the bulk of species or values being managed, and (ii) a more sophisticated risk assessment for a subset of the species of interest for which detailed risk assessments are tractable. This process strikes a balance between sophistication and practicality.

Selection of a modeling approach is an important step in the conservation planning process, but little guidance is available. We compared two statistical and three theoretical habitat modeling approaches representing those currently being used for avian conservation planning at landscape and regional scales: hierarchical spatial count (HSC), classification and regression tree (CRT), habitat suitability index (HSI), forest structure database (FS), and habitat association database (HA). We focused our comparison on models for five priority forest-breeding species in the Central Hardwoods Bird Conservation Region: Acadian Flycatcher, Cerulean Warbler, Prairie Warbler, Red-headed Woodpecker, and Worm-eating Warbler. Lacking complete knowledge on the distribution and abundance of each species with which we could illuminate differences between approaches and provide strong grounds for recommending one approach over another, we used two approaches to compare models: rank correlations among model outputs and comparison of spatial correspondence. In general, rank correlations were significantly positive among models for each species, indicating general agreement among the models. Worm-eating Warblers had the highest pairwise correlations, all of which were significant (P , 0.05). Red-headed Woodpeckers had the lowest agreement among models, suggesting greater uncertainty in the relative conservation value of areas within the region. We assessed model uncertainty by mapping the spatial congruence in priorities (i.e., top ranks) resulting from each model for each species and calculating the coefficient of variation across model ranks for each location. This allowed identification of areas more likely to be good targets of conservation effort for a species, those areas that were least likely, and those in between where uncertainty is higher and thus conservation action incorporates more risk. Based on our results, models developed independently for the same purpose (conservation planning for a particular species in a particular geography) yield different answers and thus different conservation strategies. We assert that using only one habitat model (even if validated) as the foundation of a conservation plan is risky. Using multiple models (i.e., ensemble prediction) can reduce uncertainty and increase efficacy of conservation action when models corroborate one another and increase understanding of the system when they do not.