George Melendez Wright

Abstract Beavers have occurred in North America since at least 7 Ma, but relatively little is known about their distribution across the continent. We modeled distributions of beavers in the late Pliocene (3.3 Ma), Pleistocene (130 ka and 21 ka), and recent Holocene (1970 to 2000) to understand their dispersal across North America, predict future distributions and predict their possible response to future climate and habitat changes. Occurrence data for Castor canadensis were derived from the Global Biodiversity Information Facility. Those data were used with both modern (1970 to 2000) and modeled future (EC-Earth-Veg 2081 to 2100) bioclimatic variables from WorldClim as well as past (Pliocene Marine Isotope Stage M2, Pleistocene Last Interglacial, and Pleistocene Last Glacial Maximum) bioclimatic variables from PaleoClim to model beaver distributions through time. Fossil locality points for Castor extracted from the New and Old Worlds Database of Fossil Mammals (NOW), NEOTOMA Paleoecology Database, and Paleobiology Database were overlain on past projection models to use as validation points. Models were run using MaxEnt with post-processing in ArcGIS. Accuracy for the 5 models ranged between 59.6% and 60.2%. Results for the present model (1970 to 2000) showed habitat suitability in areas beavers inhabit today. During the Pliocene MIS M2 cooling event (3.3 Ma) and Pleistocene Last Glacial Maximum (21 ka), habitat suitability shifted further south into Mexico and peninsular Florida and away from more periglacial northern regions. During the Last Interglacial period (130 ka) and modeled future (2081 to 2100) EC-Earth-Veg 2081 to 2100, habitat suitability was higher in coastal and central regions in North America and lower in southern regions compared to their present distribution. Distributions were most affected by precipitation seasonality, isothermality, and mean annual temperature. High variability in seasonal precipitation and temperatures is likely to influence surface water availability, vegetation type, and riparian vegetation composition, which consequently may reduce available food resources and habitat for beavers. Observed shifts during warmer periods may indicate areas in the late Miocene that facilitated dispersal into North America. Future models using other predicted climatic scenarios and shared socioeconomic pathways may provide better resolution of potential future shifts in beaver distribution with best- and worst-case climate scenarios, thereby permitting at-risk areas to be prioritized for conservation in the face of climate change.

Understanding and mitigating the effects of anthropogenic climate change on species distributions requires the ability to track range shifts over time. This is particularly true for species occupying high-latitude regions, which are experiencing more extreme climate change than the rest of the world. In North America, the geographic ranges of many mammals reach their northernmost extent in Alaska, positioning this region at the leading edge of climate-induced distribution change. Over a decade has elapsed since the publication of the last spatial assessments of terrestrial mammals in the state. We compared public occurrence records against commonly referenced range maps to evaluate potential extralimital records and develop repeatable baseline range maps. We compared occurrence records from the Global Biodiversity Information Facility for 61 terrestrial mammal species native to mainland Alaska against a variety of range estimates (International Union for Conservation of Nature, Alaska Gap Analysis Project, and the published literature). We mapped extralimital records and calculated proportions of occurrences encompassed by range extents, measured mean direction and distance to prior range margins, evaluated predictive accuracy of published species models, and highlighted observations on federal lands in Alaska. Range comparisons identified 6,848 extralimital records for 39 of 61 (63.9%) terrestrial mainland Alaskan species. On average, 95.5% of Alaska Gap Analysis Project occurrence records and ranges were deemed accurate (i.e., > 90.0% correct) for 31 of 37 species, but overestimated extents for 13 species. The International Union for Conservation of Nature range maps encompassed 68.1% of occurrence records and were > 90% accurate for 17 of 39 species. Extralimital records represent either improved sampling and digitization or actual geographic range expansions. Here we provide new data-driven range maps, update standards for the archiving of museum-quality locational records and offer recommendations for mapping range changes for monitoring and conservation.

Species occurrence-based climate niche models (CNMs) serve as valuable tools for predicting the future ranges of species’ suitable habitats, aiding the development of climate change adaptation strategies. However, these models do not address an essential aspect - productivity, which holds economic significance for timber production and ecological importance for carbon sequestration and ecosystem services. In this study, we investigated the potential to extend the CNMs to predict species productivity under various climate conditions. Lodgepole pine (Pinus contorta Dougl. ex Loud.) and Douglas-fir (Pseudotsuga menziesii Franco.) were selected as our model species due to their comprehensive range-wide occurrence data and measurement of site productivity. To achieve this, we compared and optimized the performance of four individual modeling algorithms (Random Forest (RF), Maxent, Generalized Boosted Models (GBM), and Generalized Additive Model (GAM)) in reflecting site productivity by evaluating the effect of spatial filtering, and the ratio of presence to absence (p/a ratio) observations. Additionally, we applied a binning process to capture the overarching trend of climatic effects while minimizing the impact of other factors. We observed consistency in optimal performance across both species when using the unfiltered data and a 1:1.5 p/a ratio, which could potentially be extended to other species. Among the modeling algorithms explored, we selected the ensemble model combining RF and Maxent as the final model to predict the range-wide site productivity for both species. The predicted range-wide site productivity was validated with an independent dataset for each species and yielded promising results (R2 above 0.7), affirming our model’s credibility. Our model introduced an innovative approach for predicting species productivity with high accuracy using only species occurrence data, and significantly advanced the application of CNMs. It provided crucial tools and insights for evaluating climate change's impact on productivity and holds a better potential for informed forest management and conservation decisions.

Assessing niche evolution remains an open question and an actively developing area of study. The family Ursidae consists of eight extant species for which, despite being the most studied family of carnivores, little is known about the influence of climate on their evolutionary history and diversification. We evaluated their evolutionary patterns based on a combined phylogeography and niche modeling approach. We used complete mitogenomes, estimated divergence times, generated ecological niche models and applied a phyloclimatic model to determine the species evolutionary and diversification patterns associated with their respective environmental niches. We inferred the family evolutionary path along the environmental conditions of maximum temperature and minimum precipitation, from around 20 million years ago to the present. Our findings show that the phyloclimatic niches of the bear species occupy most of the environmental space available on the planet, except for the most extreme warm conditions, in accordance with the wide geographic distribution of Ursidae. Moreover, some species exhibit broader environmental niches than others, and in some cases, they explore precipitation axes more extensively than temperature axes or vice versa, suggesting that not all species are equally adaptable to these variables. We were able to elucidate potential patterns of niche conservatism and evolution, as well as niche overlapping, suggesting interspecific competitive exclusion between some of the bear species. We present valuable insights into the ecological and evolutionary processes driving the diversification and distribution of the Ursidae. Our approach also provides essential information for guiding effective conservation strategies, particularly in terms of distribution limits in the face of climate change.

Species climate niche models (CNMs) have been widely used for assessing climate change impact, developing conservation strategies and guiding assisted migration for adaptation to future climates. However, the CNMs built based on species occurrence data only reflect the species’ realized niche, which can overestimate the potential loss of suitable habitat of existing forests and underestimate the potential of assisted migration to mitigate climate change. In this study, we explored building a fundamental climate niche model using widely available species occurrence data with two important forest tree species, lodgepole pine (Pinus contorta Dougl. ex Loud.) and Douglas-fir (Pseudotsuga menziesii Franco.), which were introduced to many countries worldwide. We first compared and optimized three individual modeling techniques and their ensemble by adjusting the ratio of presence to absence (p/a) observations using an innovative approach to predict the realized climate niche of the two species. We then extended the realized climate niches to their fundamental niches by determining a new cut-off threshold based on species occurrence data beyond the native distributions. We found that the ensemble model comprising Random Forest and Maxent had the best performance and identified a common cut-off threshold of 0.3 for predicting the fundamental climate niches of the two species, which is likely applicable to other species. We then predicted the fundamental climate niches of the two species under current and future climate conditions. Our study demonstrated a novel approach for predicting species’ fundamental climate niche with high accuracy using only species occurrence data, including both presence and absence data points. It provided a new tool for assessing climate change impact on the future loss of existing forests and implementing assisted migration for better adapting to future climates.

The Qinghai-Tibetan Plateau (QTP) harbors abundant and diverse plant life owing to its high habitat heterogeneity. However, the distribution pattern of biodiversity hotspots and their conservation status remain unclear. Based on 148,283 high-resolution occurrence coordinates of 13,450 seed plants, w…

During the Quaternary, large climate oscillations impacted the distribution and demography of species globally. Two approaches have played a major role in reconstructing changes through time: Bayesian Skyline Plots (BSPs), which reconstruct population fluctuations based on genetic data, and Species …

Context: Large carnivores are crucial to ecosystem functioning, as they enhance the biodiversity of the native communities in which they live. However, most large carnivores are threatened with extinction resulting from human persecution, habitat encroachment, and the loss of habitat connectivity. …

The New Zealand flora has a high proportion of endemic species but has been invaded by almost the same number of non-native plant species. To support management of invasive plant species, we provide an updated inventory of New Zealand’s naturalised flora and compare it with the native flora to ident…

In multiple regression Y ~ β0 + β1X1 + β2X2 + β3X1 X2 + ɛ., the interaction term is quantified as the product of X1 and X2. We developed fractional-power interaction regression (FPIR), using βX1M X2N as the interaction term. The rationale of FPIR is that the slopes of Y-X1 regression along the X2 gr…