John Alroy

Coverage‐based rarefaction (CBR) is a high‐profile tool for assessing biodiversity that provides relative species richness estimates by leveraging the Good‐Turing index u to interpolate. In contrast to alternatives such as the Shannon and Simpson indices, CBR's main appeal is providing values in units of species. CBR is tested against a series of other biodiversity measures. Data are both simulated and empirical, in the latter case drawn from an eclectic global database of terrestrial organisms. Various challenges are presented. First, species counts are simulated under the geometric Weibull (GW), Poisson log normal, and discretised Weibull abundance distributions. CBR and six other diversity estimators are then computed. Second, diversity estimates are computed for species inventories and then recomputed after excluding the single most common species in each one, which stands in for random inclusion of high counts in replicate samples. Third, randomly selected pairs of inventories are either 1) analysed separately with richness estimates summed, or 2) combined and only then analysed. Fourth, estimates are compared for randomly paired inventories: each pair must include an identical or highly similar number of singletons (= species sampled once). On average, fitting GW in simulation consistently returns an accurate and precise estimate of richness. GW returns much the same values regardless of how the empirical data are treated. CBR often overestimates by a large margin when dominants are excluded, underestimates by a large margin when data are combined, and nearly randomises values for singleton‐matched inventory pairs. In other words, CBR does not respond predictably to variation in species richness and cannot reconstruct it when the data have strong internal structure. Because better options are available, it is neither wise nor necessary to standardise by coverage in order to estimate richness.

Aim: Understanding patterns and drivers of biodiversity is a primary focus of ecological research. While all biodiversity data contain biases, most recent studies have used range maps derived from experts or species distribution models to test for controls on observed species richness patterns. Less attention has been given to other data sources, such as occurrence records from museum specimens or community science initiatives, which are heavily biased by sampling effort. However, this problem can be resolved using richness extrapolators to estimate ‘true’ richness from counts of observations, which allows for undetected or undescribed species to be considered. Location: Australia. Time Period: Recent. Major Taxa Studied: Amphibians. Methods: We apply four richness extrapolation methods while evaluating their performance. We analyse occurrence records and compare how richness estimated from range maps, museum specimens and community science affects support for biodiversity hypotheses. These are compared using spatial autoregressive models at two resolutions. We also include sensitivity analyses using zero‐richness cells and human population density. Results: The geometric series index (GSI) distribution performed best on the museum occurrence datasets and estimated higher richness than using range maps. Fitting the geometric Weibull (GW) distribution yielded similar performance. We find that occurrence‐ and range map‐based models show support for the energy and habitat heterogeneity hypotheses, although models explain much less variation in occurrence estimates. Including zero‐richness cells sometimes improved model performance. However, extrapolating richness from community science occurrences failed to detect missing richness or recover the latitudinal diversity gradient present in the other two datasets. Main Conclusions: Richness estimated from different data sources may support different biodiversity hypotheses, and these differences may be influenced by underlying biases. We conclude that occurrence records can be informative for testing biodiversity hypotheses when biases are appropriately considered, and that future studies should take advantage of multiple data types when available.

Cambridge Prisms: Extinction was launched to provide an open, synthetic forum for research on the patterns, processes and consequences of extinction. Four years on, the release of the journal’s first impact factor offers a useful moment to look back at what has begun to take shape and to look ahead to what it should now become. The early content shows a young journal whose first visible scholarly impact has been driven largely by synthesis. That was deliberate: a review-led beginning helped define the journal’s intellectual map, draw together conservation biology, palaeontology, modelling, policy, social science and the humanities, and create pathways for more primary and data-driven work. The next stage should build on that foundation through more empirical research and through Extinction Accounts , a new article format designed to analyse individual extinctions.

There is much contention over the causes and correlates of megafaunal extinctions at the end of the Pleistocene. A major role for human impact such as hunting has been discussed widely. If correct, the overkill hypothesis explains not only why large mammals in general were highly prone to extinction but suggests that extinction may have been selective within large mammals. Among other things, it has been argued that extinct large mammals tended to be large and have small brains. Here we test these hypotheses using a comprehensive global dataset of 22 ecological and life history traits mapped to 120 living and 14 extinct carnivore species. The data document occurrences within 260 distinct fossil assemblages that span the Late Pleistocene and Holocene. To address collinearity and phylogenetic autocorrelation, we first perform least‐squares orthogonalisation of the predictor variables and then use phylogenetic comparative methods to carry out regressions. Only basal metabolic rate and diurnality are robust predictors of extinction, even after accounting for phylogenetic and trait uncertainty. Furthermore, we show that living carnivores with high metabolic rates are more likely to be threatened and address the implications for conservation and the current extinction crisis.

Many of the world's megafaunal species went extinct during the late Quaternary, leading to dramatic reductions in community and ecosystem functioning. While the nature and severity of the extinctions are well documented on global and continental scales, less is known about local‐scale impacts. We quantified the biomass and energy use of 292 pre‐extinction and 360 post‐extinction fossil assemblages from around the globe to determine effects on large mammal communities. Assemblage energy use was calculated from metabolic rates obtained for 562 individual species, and was compared to species richness along with indicators of taphonomy, archaeology, and biogeography, using three analytical methods: least‐squares orthogonalization regression, spatial autoregression, and linear mixed effects model analysis. Globally, total biomass and energy use are greatly reduced in post‐extinction assemblages. Human‐accumulated assemblages are further homogenised post‐extinction due to their high abundances of domesticated species. The presence of domesticates greatly altered the biomass and energy use of assemblages post‐extinction, producing strong energetic variation across continents that differs considerably to pre‐extinction patterns. This fundamental anthropogenic alteration of communities further exacerbated the impacts of Pleistocene extinctions, even in less severely impacted regions. The results show how human activities have altered mammalian communities for many thousands of years.

We analysed fossil mammal assemblages from over 350 Late Pleistocene and Holocene sites worldwide to test whether human activities, such as agriculture, domestication and intensified land use, restructured global patterns of mammal co-occurrence. Using presence–absence data, we contrasted a novel iterative ‘chase clustering’ method, which is compositionally driven, against a traditional spatially constrained Ward’s clustering approach. Both methods recovered continental-scale groupings in the Pleistocene, consistent with known biogeographic boundaries. Holocene land use and domestication reconfigured these historical patterns, creating novel assemblages independent of previous biogeographic constraints. Faunal turnover at the local scale varied substantially across regions, being especially pronounced in the Americas, whereas other areas showed relative stability. Even moderate expansion of domesticates altered how communities grouped, highlighting their disproportionate ecological influence. Our findings demonstrate that human-driven niche modification, beyond earlier megafaunal extinctions, profoundly reshaped mammal communities on a global scale. Recognizing these anthropogenic legacies provides essential context for anticipating how current and future human pressures might further transform biodiversity.

Ecological interactions influence which species can coexist locally, but assessing the effects of interactions on species distributions at landscape to regional scales has proven challenging. Here, we present a new statistical method to address this question at the assemblage level. Our method, which we call mixed-effects co-occurrence modelling, entails fitting pairwise species co-occurrence data to generalized linear mixed models using Fisher’s non-central hypergeometric distribution. We use this method to examine the effects of phylogenetic relatedness, dietary-niche overlap (a proxy for potential food competition) and human-induced habitat alteration on pairwise co-occurrence patterns for bat and bird assemblages that span most of the Neotropics. For both assemblages, average pairwise co-occurrence increased with phylogenetic relatedness, indicating that phylogenetic niche conservatism contributes to environmental filtering at broad spatial scales. After controlling for phylogeny, variance in co-occurrence tended to be highest for species pairs in the same dietary guilds, suggesting varied responses to food competition. Effects of habitat alteration were relatively weak and inconsistent, though our analysis precluded identifying effects that were phylogenetically structured. Overall, our findings indicate that phylogenetic relationships among species pairs are instrumental in determining patterns of species co-occurrence and thereby influence how biological interactions play out at broad spatial scales.

1. Over the decades, Barro Colorado Island (BCI) has incubated major conceptual and empirical advances in ecology. For example, the theoretical zero-sum multinomial (ZSM), log series, and Poisson log normal (PLN) models have been used to describe abundance distributions and quantify diversity in the BCI 50 ha forest plot. 2. A new model called the compound exponential-geometric series (CEGS) better predicts distributions and quantifies alpha and beta diversity in 1 ha subplots. Distributions are analysed by fitting the four models to each of the subplot's data sets and then predicting counts in adjacent subplots. The PLN performs second-best while the ZSM and log series have almost no predictive power. Meanwhile, only CEGS detects plausible species-area relationships. 3. CEGS assumes nothing specifically about regional speciation rates, density dependence, or competition. It results when recruitment rates are on a per-species basis, whereas death rates are per-capita. It is very much not neutral because it posits that each species has a randomly allocated underlying abundance. It could be interpreted as implying that ecological traits govern demographic rates and in turn structure stem counts. A series of new predictions follows from the model. 4. The model deserves further testing with ecological data, including other forest plot inventories. 5. Synthesis: Ecologists often advocate so-called neutral models in which tree species have identical recruitment and mortality rates, but the actual data for the best-studied tree plot in the world suggest high variation in rates. When a better model is used, internally consistent and sensible species richness estimates are obtained.

The idea that diversity is a combination of species richness and the so‐called “evenness” of count distributions is a bedrock concept in ecology. Researchers often compute stand‐alone evenness indices. They also examine Hill numbers related to Shannon's H and Simpson's D because these metrics balance richness and “evenness” to various degrees. But evenness is an operationally problematic abstraction, not a thing out in the world. Evenness indices and Hill numbers in empirical data are overly sensitive to the abundance of dominant species, poorly replicable within communities, highly variable among similar communities, and a weak indicator of latitudinal biodiversity trends. They are inconsistently related to the parameters of key models that might underlie count distributions, and they vary highly in simulation even when these model parameters do not vary. Ecologists would benefit by instead determining which real distributions fit which theoretical models and using estimated parameters to understand community structure and assembly.

Coverage-based rarefaction (CBR) is a high-profile tool for assessing biodiversity that provides relative species richness estimates. It leverages the Good-Turing index u to interpolate expected richness given a specified level of frequency distribution coverage. In contrast to alternatives such as the Shannon and Simpson indices, CBR’s main appeal is providing values in units of species. CBR is tested against a series of other biodiversity measures. Data are both simulated and empirical, in the latter case drawn from an eclectic global database of terrestrial organisms. First, species counts are simulated under three underlying abundance distributions: the compound exponential-geometric series (CEGS), Poisson log normal, and discretised Weibull. CBR and five other diversity estimators are then computed. Second, diversity estimates are computed for species inventories and then recomputed after excluding the single most common species in each one. Third, randomly selected pairs of inventories are either (1) analysed separately with richness estimates summed, or (2) combined and only then analysed. On average, fitting CEGS in simulation consistently returns an accurate and precise estimate of richness. CBR yields little signal. CEGS returns much the same values regardless of whether empirical data sets include or exclude dominants and regardless of whether they are combined or analysed separately. CBR often overestimates by a large margin when dominants are excluded and underestimates by a large margin when data are combined. CBR does not respond predictably to variation in species richness and cannot reconstruct it when the data have strong internal structure. CBR’s usefulness as a biodiversity indicator is unclear.