A cloudy forecast for species distribution models: Predictive uncertainties abound for California birds after a century of climate and land-use change.

Correlative species distribution models are widely used to quantify past shifts in ranges or communities, and to predict future outcomes under ongoing global change. Practitioners confront a wide range of potentially plausible models for ecological dynamics, but most specific applications only consider a narrow set. Here, we clarify that certain model structures can embed restrictive assumptions about key sources of forecast uncertainty into an analysis. To evaluate forecast uncertainties and our ability to explain community change, we fit and compared 39 candidate multi- or joint species occupancy models to avian incidence data collected at 320 sites across California during the early 20th century and resurveyed a century later. We found massive (>20,000 LOOIC) differences in within-time information criterion across models. Poorer fitting models omitting multivariate random effects predicted less variation in species richness changes and smaller contemporary communities, with considerable variation in predicted spatial patterns in richness changes across models. The top models suggested avian environmental associations changed across time, contemporary avian occupancy was influenced by previous site-specific occupancy states, and that both latent site variables and species associations with these variables also varied over time. Collectively, our results recapitulate that simplified model assumptions not only impact predictive fit but may mask important sources of forecast uncertainty and mischaracterize the current state of system understanding when seeking to describe or project community responses to global change. We recommend that researchers seeking to make long-term forecasts prioritize characterizing forecast uncertainty over seeking to present a single best guess. To do so reliably, we urge practitioners to employ models capable of characterizing the key sources of forecast uncertainty, where predictors, parameters and random effects may vary over time or further interact with previous occurrence states.

Concordant and opposing effects of climate and land-use change on avian assemblages in California's most transformed landscapes.

Climate and land-use change could exhibit concordant effects that favor or disfavor the same species, which would amplify their impacts, or species may respond to each threat in a divergent manner, causing opposing effects that moderate their impacts in isolation. We used early 20th century surveys of birds conducted by Joseph Grinnell paired with modern resurveys and land-use change reconstructed from historic maps to examine avian change in Los Angeles and California's Central Valley (and their surrounding foothills). Occupancy and species richness declined greatly in Los Angeles from urbanization, strong warming (+1.8°C), and drying (-77.2 millimeters) but remained stable in the Central Valley, despite large-scale agricultural development, average warming (+0.9°C), and increased precipitation (+11.2 millimeters). While climate was the main driver of species distributions a century ago, the combined impacts of land-use and climate change drove temporal changes in occupancy, with similar numbers of species experiencing concordant and opposing effects.

A flexible and efficient Bayesian implementation of point process models for spatial capture-recapture data.

Spatial capture-recapture (SCR) is now routinely used for estimating abundance and density of wildlife populations. A standard SCR model includes sub-models for the distribution of individual activity centers (ACs) and for individual detections conditional on the locations of these ACs. Both sub-models can be expressed as point processes taking place in continuous space, but there is a lack of accessible and efficient tools to fit such models in a Bayesian paradigm. Here, we describe a set of custom functions and distributions to achieve this. Our work allows for more efficient model fitting with spatial covariates on population density, offers the option to fit SCR models using the semi-complete data likelihood (SCDL) approach instead of data augmentation, and better reflects the spatially continuous detection process in SCR studies that use area searches. In addition, the SCDL approach is more efficient than data augmentation for simple SCR models while losing its advantages for more complicated models that account for spatial variation in either population density or detection. We present the model formulation, test it with simulations, quantify computational efficiency gains, and conclude with a real-life example using non-invasive genetic sampling data for an elusive large carnivore, the wolverine (Gulo gulo) in Norway.

Estimating spatially variable and density-dependent survival using open-population spatial capture-recapture models.

Open-population spatial capture-recapture (OPSCR) models use the spatial information contained in individual detections collected over multiple consecutive occasions to estimate not only occasion-specific density, but also demographic parameters. OPSCR models can also estimate spatial variation in vital rates, but such models are neither widely used nor thoroughly tested. We developed a Bayesian OPSCR model that not only accounts for spatial variation in survival using spatial covariates but also estimates local density-dependent effects on survival within a unified framework. Using simulations, we show that OPSCR models provide sound inferences on the effect of spatial covariates on survival, including multiple competing sources of mortality, each with potentially different spatial determinants. Estimation of local density-dependent survival was possible but required more data due to the greater complexity of the model. Not accounting for spatial heterogeneity in survival led to up to 10% positive bias in abundance estimates. We provide an empirical demonstration of the model by estimating the effect of country and density on cause-specific mortality of female wolverines (Gulo gulo) in central Sweden and Norway. The ability to make population-level inferences on spatial variation in survival is an essential step toward a fully spatially explicit OPSCR model capable of disentangling the role of multiple spatial drivers of population dynamics.

Comparing N-mixture models and GLMMs for relative abundance estimation in a citizen science dataset.

To analyze species count data when detection is imperfect, ecologists need models to estimate relative abundance in the presence of unknown sources of heterogeneity. Two candidate models are generalized linear mixed models (GLMMs) and hierarchical N-mixture models. GLMMs are computationally robust but do not explicitly separate detection from abundance patterns. N-mixture models separately estimate detection and abundance via a latent state but are sensitive to violations in assumptions and subject to practical estimation issues. When one can assume that detection is not systematically confounded with ecological patterns of interest, these two models can be viewed as sharing a heuristic framework for relative abundance estimation. Model selection can then determine which predicts observed counts best, for example by AIC. We compared four N-mixture model variants and two GLMM variants for predicting bird counts in local subsets of a citizen science dataset, eBird, based on model selection and goodness-of-fit measures. We found that both GLMMs and N-mixture models-especially N-mixtures with beta-binomial detection submodels-were supported in a moderate number of datasets, suggesting that both tools are useful and that relative fit is context-dependent. We provide faster software implementations of N-mixture likelihood calculations and a reparameterization to interpret unstable estimates for N-mixture models.

Identifying engaging bird species and traits with community science observations.

Identifying rates at which birders engage with different species can inform the impact and efficacy of conservation outreach and the scientific use of community-collected biodiversity data. Species that are thought to be "charismatic" are often prioritized in conservation, and previous researchers have used sociological experiments and digital records to estimate charisma indirectly. In this study, we take advantage of community science efforts as another record of human engagement with animals that can reveal observer biases directly, which are in part driven by observer preference. We apply a multistage analysis to ask whether opportunistic birders contributing to iNaturalist engage more with larger, more colorful, and rarer birds relative to a baseline approximated from eBird contributors. We find that body mass, color contrast, and range size all predict overrepresentation in the opportunistic dataset. We also find evidence that, across 472 modeled species, 52 species are significantly overreported and 158 are significantly underreported, indicating a wide variety of species-specific effects. Understanding which birds are highly engaging can aid conservationists in creating impactful outreach materials and engaging new naturalists. The quantified differences between two prominent community science efforts may also be of use for researchers leveraging the data from one or both of them to answer scientific questions of interest.

Building integral projection models with nonindependent vital rates.

Population dynamics are functions of several demographic processes including survival, reproduction, somatic growth, and maturation. The rates or probabilities for these processes can vary by time, by location, and by individual. These processes can co-vary and interact to varying degrees, e.g., an animal can only reproduce when it is in a particular maturation state. Population dynamics models that treat the processes as independent may yield somewhat biased or imprecise parameter estimates, as well as predictions of population abundances or densities. However, commonly used integral projection models (IPMs) typically assume independence across these demographic processes. We examine several approaches for modelling between process dependence in IPMs and include cases where the processes co-vary as a function of time (temporal variation), co-vary within each individual (individual heterogeneity), and combinations of these (temporal variation and individual heterogeneity). We compare our methods to conventional IPMs, which treat vital rates independent, using simulations and a case study of Soay sheep (Ovis aries). In particular, our results indicate that correlation between vital rates can moderately affect variability of some population-level statistics. Therefore, including such dependent structures is generally advisable when fitting IPMs to ascertain whether or not such between vital rate dependencies exist, which in turn can have subsequent impact on population management or life-history evolution.

To rarefy or not to rarefy: robustness and efficiency trade-offs of rarefying microbiome data.

MOTIVATION: Microbiome datasets provide rich information about microbial communities. However, vast library size variations across samples present great challenges for proper statistical comparisons. To deal with these challenges, rarefaction is often used in practice as a normalization technique, although there has been debate whether rarefaction should ever be used. Conventional wisdom and previous work suggested that rarefaction should never be used in practice, arguing that rarefying microbiome data is statistically inadmissible. These discussions, however, have been confined to particular parametric models and simulation studies. RESULTS: We develop a semiparametric graphical model framework for grouped microbiome data and analyze in the context of differential abundance testing the statistical trade-offs of the rarefaction procedure, accounting for latent variations and measurement errors. Under the framework, it can be shown rarefaction guarantees that subsequent permutation tests properly control the Type I error. In addition, the loss in sensitivity from rarefaction is solely due to increased measurement error; if the underlying variation in microbial composition is large among samples, rarefaction might not hurt subsequent statistical inference much. We develop the rarefaction efficiency index (REI) as an indicator for efficiency loss and illustrate it with a dataset on the effect of storage conditions for microbiome data. Simulation studies based on real data demonstrate that the impact of rarefaction on sensitivity is negligible when overdispersion is prominent, while low REI corresponds to scenarios in which rarefying might substantially lower the statistical power. Whether to rarefy or not ultimately depends on assumptions of the data generating process and characteristics of the data. AVAILABILITY AND IMPLEMENTATION: Source codes are publicly available at https://github.com/jcyhong/rarefaction. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Multi-scale trend analysis of water quality using error propagation of generalized additive models.

Effective stewardship of ecosystems to sustain current ecological status or mitigate impacts requires nuanced understanding of how conditions have changed over time in response to anthropogenic pressures and natural variability. Detecting and appropriately characterizing changes requires accurate and flexible trend assessment methods that can be readily applied to environmental monitoring datasets. A key requirement is complete propagation of uncertainty through the analysis. However, this is difficult when there are mismatches between sampling frequency, period of record, and trends of interest. Here, we propose a novel application of generalized additive models (GAMs) for characterizing multi-decadal changes in water quality indicators and demonstrate its utility by analyzing a 30-year record of biweekly-to-monthly chlorophyll-a concentrations in the San Francisco Estuary. GAMs have shown promise in water quality trend analysis to separate long-term (i.e., annual or decadal) trends from seasonal variation. Our proposed methods estimate seasonal averages in a response variable with GAMs, extract uncertainty measures for the seasonal estimates, and then use the uncertainty measures with mixed-effects meta-analysis regression to quantify inter-annual trends that account for full propagation of error across methods. We first demonstrate that nearly identical descriptions of temporal changes can be obtained using different smoothing spline formulations of the original time series. We then extract seasonal averages and their standard errors for an a priori time period within each year from the GAM results. Finally, we demonstrate how across-year trends in seasonal averages can be modeled with mixed-effects meta-analysis regression that propagates uncertainties from the GAM fits to the across-year analysis. Overall, this approach leverages GAMs to smooth data with missing observations or varying sample effort across years to estimate seasonal averages and meta-analysis to estimate trends across years. Methods are provided in the wqtrends R package.

Estimating and forecasting spatial population dynamics of apex predators using transnational genetic monitoring.

The ongoing recovery of terrestrial large carnivores in North America and Europe is accompanied by intense controversy. On the one hand, reestablishment of large carnivores entails a recovery of their most important ecological role, predation. On the other hand, societies are struggling to relearn how to live with apex predators that kill livestock, compete for game species, and occasionally injure or kill people. Those responsible for managing these species and mitigating conflict often lack fundamental information due to a long-standing challenge in ecology: How do we draw robust population-level inferences for elusive animals spread over immense areas? Here we showcase the application of an effective tool for spatially explicit tracking and forecasting of wildlife population dynamics at scales that are relevant to management and conservation. We analyzed the world's largest dataset on carnivores comprising more than 35,000 noninvasively obtained DNA samples from over 6,000 individual brown bears (Ursus arctos), gray wolves (Canis lupus), and wolverines (Gulo gulo). Our analyses took into account that not all individuals are detected and, even if detected, their fates are not always known. We show unequivocal quantitative evidence of large carnivore recovery in northern Europe, juxtaposed with the finding that humans are the single-most important factor driving the dynamics of these apex predators. We present maps and forecasts of the spatiotemporal dynamics of large carnivore populations, transcending national boundaries and management regimes.