Spatially explicit power analysis for detecting occupancy trends for multiple species.

Assessing the statistical power to detect changes in wildlife populations is a crucial yet often overlooked step when designing and evaluating monitoring programs. Here, we developed a simulation framework to perform spatially explicit statistical power analysis of biological monitoring programs for detecting temporal trends in occupancy for multiple species. Using raster layers representing the spatial variation in current occupancy and species-level detectability for one or multiple observation methods, our framework simulates changes in occupancy over space and time, with the capacity to explicitly model stochastic disturbances at monitoring sites (i.e., dynamic landscapes). Once users specify the number and location of sites, the frequency and duration of surveys, and the type of detection method(s) for each species, our framework estimates power to detect occupancy trends, both across the landscape and/or within nested management units. As a case study, we evaluated the power of a long-term monitoring program to detect trends in occupancy for 136 species (83 birds, 33 reptiles, and 20 mammals) across and within Kakadu, Litchfield, and Nitmiluk National Parks in northern Australia. We assumed continuation of an original monitoring design implemented since 1996, with the addition of camera trapping. As expected, power to detect trends was sensitive to the direction and magnitude of the change in occupancy, detectability, initial occupancy levels, and the rarity of species. Our simulations suggest that monitoring has at least an 80% chance at detecting a 50% decline in occupancy for 22% of the modeled species across the three parks over the next 15 yr. Monitoring is more likely to detect increasing occupancy trends, with at least an 80% chance at detecting a 50% increase in 87% of species. The addition of camera-trapping increased average power to detect a 50% decline in mammals compared with using only live trapping by 63%. We provide a flexible tool that can help decision-makers design and evaluate monitoring programs for hundreds of species at a time in a range of ecological settings, while explicitly considering the distribution of species and alternative sampling methods.

Correction: Occupancy and detectability modelling of vertebrates in northern Australia using multiple sampling methods.

[This corrects the article DOI: 10.1371/journal.pone.0203304.].

Occupancy and detectability modelling of vertebrates in northern Australia using multiple sampling methods.

Understanding where species occur and how difficult they are to detect during surveys is crucial for designing and evaluating monitoring programs, and has broader applications for conservation planning and management. In this study, we modelled occupancy and the effectiveness of six sampling methods at detecting vertebrates across the Top End of northern Australia. We fitted occupancy-detection models to 136 species (83 birds, 33 reptiles, 20 mammals) of 242 recorded during surveys of 333 sites in eight conservation reserves between 2011 and 2016. For modelled species, mean occupancy was highly variable: birds and reptiles ranged from 0.01-0.81 and 0.01-0.49, respectively, whereas mammal occupancy was lower, ranging from 0.02-0.30. Of the 11 environmental covariates considered as potential predictors of occupancy, topographic ruggedness, elevation, maximum temperature, and fire frequency were retained more readily in the top models. Using these models, we predicted species occupancy across the Top End of northern Australia (293,017 km2) and generated species richness maps for each species group. For mammals and reptiles, high richness was associated with rugged terrain, while bird richness was highest in coastal lowland woodlands. On average, detectability of diurnal birds was higher per day of surveys (0.33 ± 0.09) compared with nocturnal birds per night of spotlighting (0.13 ± 0.06). Detectability of reptiles was similar per day/night of pit trapping (0.30 ± 0.09) as per night of spotlighting (0.29 ± 0.11). On average, mammals were highly detectable using motion-sensor cameras for a week (0.36 ± 0.06), with exception of smaller-bodied species. One night of Elliott trapping (0.20 ± 0.06) and spotlighting (0.19 ± 0.06) was more effective at detecting mammals than cage (0.08 ± 0.03) and pit trapping (0.05 ± 0.04). Our estimates of species occupancy and detectability will help inform decisions about how best to redesign a long-running vertebrate monitoring program in the Top End of northern Australia.

Optimal timing of biodiversity offsetting for metapopulations.

Biodiversity offsetting schemes permit habitat destruction, provided that losses are compensated by gains elsewhere. While hundreds of offsetting schemes are used around the globe, the optimal timing of habitat creation in such projects is poorly understood. Here, we developed a spatially explicit metapopulation model for a single species subject to a habitat compensation scheme. Managers could compensate for destruction of a patch by creating a new patch either before, at the time of, or after patch loss. Delaying patch creation is intuitively detrimental to species persistence, but allowed managers to invest financial compensation, accrue interest, and create a larger patch at a later date. Using stochastic dynamic programming, we found the optimal timing of patch creation that maximizes the number of patches occupied at the end of a 50-yr habitat compensation scheme when a patch is destroyed after 10 yr. Two case studies were developed for Australian species subject to habitat loss but with very different traits: the endangered growling grass frog (Litoria raniformis) and the critically endangered Mount Lofty Ranges Southern Emu-wren (Spititurus malachurus intermedius). Our results show that adding a patch either before or well after habitat destruction can be optimal, depending on the occupancy state of the metapopulation, the interest rate, the area of the destroyed patch and metapopulation parameters of the focal species. Generally, it was better to delay patch creation when the interest rate was high, when the species had a relatively high colonization rate, when the patch nearest the new patch was occupied, and when the destroyed patch was small. Our framework can be applied to single-species metapopulations subject to habitat loss, and demonstrates that considering the timing of habitat compensation could improve the effectiveness of offsetting schemes.

Learning about colonization when managing metapopulations under an adaptive management framework.

Adaptive management is a framework for resolving key uncertainties while managing complex ecological systems. Its use has been prominent in fisheries research and wildlife harvesting; however, its application to other areas of environmental management remains somewhat limited. Indeed, adaptive management has not been used to guide and inform metapopulation restoration, despite considerable uncertainty surrounding such actions. In this study, we determined how best to learn about the colonization rate when managing metapopulations under an adaptive management framework. We developed a mainland-island metapopulation model based on the threatened bay checkerspot butterfly (Euphydryas editha bayensis) and assessed three management approaches: adding new patches, adding area to existing patches, and doing nothing. Using stochastic dynamic programming, we found the optimal passive and active adaptive management strategies by monitoring colonization of vacant patches. Under a passive adaptive strategy, increasing patch area was best when the expected colonization rate was below a threshold; otherwise, adding new patches was optimal. Under an active adaptive strategy, it was best to add patches only when we were reasonably confident that the colonization rate was high. This research provides a framework for managing mainland-island metapopulations in the face of uncertainty while learning about the dynamics of these complex systems.

Abiotic and biotic interactions determine whether increased colonization is beneficial or detrimental to metapopulation management.

Increasing the colonization rate of metapopulations can improve persistence, but can also increase exposure to threats. To make good decisions, managers must understand whether increased colonization is beneficial or detrimental to metapopulation persistence. While a number of studies have examined interactions between metapopulations, colonization, and threats, they have assumed that threat dynamics respond linearly to changes in colonization. Here, we determined when to increase colonization while explicitly accounting for non-linear dependencies between a metapopulation and its threats. We developed patch occupancy metapopulation models for species susceptible to abiotic, generalist, and specialist threats and modeled the total derivative of the equilibrium proportion of patches occupied by each metapopulation with respect to the colonization rate. By using the total derivative, we developed a rule for determining when to increase metapopulation colonization. This rule was applied to a simulated metapopulation where the dynamics of each threat responded to increased colonization following a power function. Before modifying colonization, we show that managers must understand: (1) whether a metapopulation is susceptible to a threat; (2) the type of threat acting on a metapopulation; (3) which component of threat dynamics might depend on colonization, and; (4) the likely response of a threat-dependent variable to changes in colonization. The sensitivity of management decisions to these interactions increases uncertainty in conservation planning decisions.

The sensitivity of population viability analysis to uncertainty about habitat requirements: implications for the management of the endangered southern brown bandicoot.

Whenever population viability analysis (PVA) models are built to help guide decisions about the management of rare and threatened species, an important component of model building is the specification of a habitat model describing how a species is related to landscape or bioclimatic variables. Model-selection uncertainty may arise because there is often a great deal of ambiguity about which habitat model structure best approximates the true underlying biological processes. The standard approach to incorporate habitat models into PVA is to assume the best habitat model is correct, ignoring habitat-model uncertainty and alternative model structures that may lead to quantitatively different conclusions and management recommendations. Here we provide the first detailed examination of the influence of habitat-model uncertainty on the ranking of management scenarios from a PVA model. We evaluated and ranked 6 management scenarios for the endangered southern brown bandicoot (Isoodon obesulus) with PVA models, each derived from plausible competing habitat models developed with logistic regression. The ranking of management scenarios was sensitive to the choice of the habitat model used in PVA predictions. Our results demonstrate the need to incorporate methods into PVA that better account for model uncertainty and highlight the sensitivity of PVA to decisions made during model building. We recommend that researchers search for and consider a range of habitat models when undertaking model-based decision making and suggest that routine sensitivity analyses should be expanded to include an analysis of the impact of habitat-model uncertainty and assumptions.