About

Walter Jetz is a Professor of Ecology and Evolutionary Biology and a Professor of Environmental Studies at Yale University. His work is interdisciplinary, combining elements of biogeography, community ecology, landscape ecology, macroecology, global change ecology, evolution, comparative biology, biodiversity informatics, and conservation. He primarily uses terrestrial vertebrates and plants as study systems. His research aims to integrate across scales of geography and ecological organization, from global to local assemblages, to better understand ecological and evolutionary processes.

Research topics

• Computer Science
• Biology
• Geography
• Ecology
• Environmental resource management
• Environmental science
• Environmental planning
• Artificial Intelligence
• Evolutionary biology
• Environmental protection
• Remote sensing
• Medicine
• Cartography
• Aerospace engineering
• Engineering
• World Wide Web
• Virology
• Physical geography
• Data science

Selected publications

• Measuring range-wide species population changes to support global conservation monitoring and decision-making

  2026-03-10

  articleOpen access1st authorCorresponding

  Global conservation organizations invest heavily in protected areas and management actions to support biodiversity outcomes, often with a particular focus on target species. However, they require more flexible, efficient means to monitor the impact of these interventions at scale. Lack of responsive tools that can regularly track progress and impact of conservation actions can limit effective resourcing and planning for future work. While sufficient in situ data may exist for specific regions, and species, conservation organizations require standardized measurements and aggregate metrics to support decision-making across their global portfolio and countries of activity. This presented work addresses these challenges by leveraging the advanced and proven species-level models and metrics of Map of Life (MOL). This effort utilizes NASA-supported remote sensing workflows and the digital infrastructure that enables calculations of Species Habitat Scores and Species Habitat Indices – formally adopted indicators of Goal A of the UN Global Biodiversity Framework. These indicators rely on species-habitat association calculations, historically developed from expert-informed range maps validated against a single independent source of species presence data. To enhance the accuracy and reliability of these measures, we will implement emerging modelling techniques that integrate diverse data streams, including citizen science records as well as visual, acoustic, and GPS tracking data. Combining species occurrence information from multiple data sources substantially improves model predictions and resulting conservation insights, but these approaches have yet to be widely adopted by conservation practitioners for large-scale decision-making.We show results for implementing these models for mammal species based on camera trap data and citizen science data, with all routines designed to be easily extended to additional taxa. By integrating multi-source datasets with NASA and other Earth Observation products that capture fine-scale climatic and environmental conditions, we can provide annual, range-wide estimates of suitable habitat, habitat connectivity, protection adequacy, and estimate total population size trends for target species. Our advanced species distribution and occupancy modeling approaches enables us to offer both high-resolution maps and species-level trends alongside method-associated uncertainty that will directly support tracking progress towards 30x30 goals and guide adaptive management interventions.

  PublisherDOI

• Extreme weather shrinks estimated range boundaries and alters biodiversity predictions

  2026-03-10

  articleOpen accessSenior author

  Anthropogenic climate change has been linked to rapid changes in ecological patterns and processes, including species re-distributions and phenological shifts, and is threatening one out of six species with extinction. While most work to date has explored the ecological consequences of gradually rising mean temperatures, the influence of increasingly frequent and intense extreme weather for wildlife species and biodiversity patterns has been largely underexplored but is increasingly becoming appreciated. Many organisms are physiologically constrained by their upper and lower thermal limits and thus may be more likely to be pushed past their physiological limits by an extreme weather event than gradually shifting mean conditions. Species might be especially sensitive to extreme weather at the edges of their geographic ranges, where they are often already living near their physiological limits. Thus, understanding how the incidence of extreme weather events limits the boundaries of species distributions is critical for accurate ecological forecasts and better conservation outcomes under rapidly accelerating climate change. However, the influence of climatic variability and extreme weather is often ignored in favor of climatic means when estimating distributional and richness patterns. Here we use hundreds of millions of citizen science bird observations from 2004-2024 and high-resolution extreme weather risk maps to explore how climatic variability and extreme weather risk alters summer and winter distributions and biodiversity patterns for 540 North American species. We find that species distribution models accounting for historical extreme weather risk performed better at predicting richness and species’ presence or absence across ~250 sites. These models predicted much narrower geographic distributions than traditional models relying on only climatic means, with range truncation observed primarily at the range edges. Additionally, we observed these effects in both seasons, though they were particularly strong in winter. Richness estimates were substantially lower when extreme weather was accounted for, especially in the US southwest and central plains, regions highly prone to extreme heat, cold and drought. Our results suggest that more mechanistically informed biodiversity predictions that account for extreme weather are critical for appreciating and reliably predicting shifting biodiversity distributions.

  PublisherDOI

• Self-explaining Deep Learning-based Species Distribution Models

  2026-03-10

  articleOpen accessSenior author

  Many fundamental processes in ecology are nowadays modelled with ever-increasing complexity, also thanks to advancements in data science. Species distribution modelling is no exception to this trend, and in particular deep learning-based models have steadily been maturing in recent years, promising high prediction performance for many species at large scales. Yet, a fundamental desire in ecology is not just to predict, but understand, observed processes, both environmental and model-intrinsic. Deep learning models are often described as black boxes due to their complexity, and hence have a notorious reputation of being unsuitable for either. However, recent years have seen great advancements in both unravelling and more explicitly quantifying the decision process of deep neural networks.In this work, we explore the potential of a deep learning-based species distribution model (SDM) that explains itself by design. The model achieves this via a learned top-K sampling scheme with attention mechanisms on the environmental covariates it receives. In detail, the model is forced to select a subset of user-definable size (K) of covariates that is as useful as possible for the prediction of species encounter likelihoods at each data point. Within this sampling scheme, covariates are either available or not (and not modulated as in regular attention mechanisms), and unlike in post-hoc explainability methods, no auxiliary model is required to explain the SDM's decision process. The result are per-covariate importance scores that are as trustworthy as possible.We evaluate our model on a set of around 830,000 observations for 356 mammal species, sampled over North America, comparing prediction performances and investigating obtained covariate importances. We find that our sampling scheme does result in highly consistent covariate combinations across runs, and further see plausible correlations with the environmental configuration across the continent. We further investigate correspondence with post-hoc explainability methods and find improvable agreement, highlighting the challenges in explainability for machine learning models in general, and deep learning SDMs in particular.

  PublisherDOI

• A theoretical framework for scaling ecological niches from individuals to species

  Proceedings of the National Academy of Sciences · 2025-08-29 · 4 citations

  articleOpen accessSenior authorCorresponding

  The niche is a key concept that unifies ecology and evolutionary biology. However, empirical and theoretical treatments of the niche are mostly performed at the species level, neglecting individuals as important units of ecological and evolutionary processes. So far, a formal mathematical link between individual-level niches and higher organismal-level niches has been lacking, hampering the unification of ecological theories and more accurate forecasts of biodiversity change. To fill in this gap, we propose a bottom-up approach to derive population and higher organismal-level niches from individual niches. We demonstrate the power of our framework by showing that 1) the statistical properties of higher organismal-level niches (e.g., niche breadth, skewness, etc.) can be partitioned into individual contributions and 2) the species-level niche shifts can be estimated by tracing the responses of individuals. By using individual-level GPS (Global Positioning System) tracking data from three different species, we show that climate change could have contrasting consequences on population-level niche shift depending on individual niche compositions. Our method paves the way for a unifying niche theory and enables mechanistic assessments of organism-environment relationships across organismal scales.

  PublisherDOI

• Anthropocene Imperilment of Ancient Diversity and Evolutionary Potential in Terrestrial Vertebrates

  Research Square · 2025-09-25

  preprintOpen accessSenior author
  PublisherOA PDFDOI

• Extreme weather risk shrinks range size estimates and alters biodiversity predictions

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-10-14 · 1 citations

  preprintOpen accessSenior author

  Abstract Extreme weather events, including heat waves, cold snaps, and droughts, are increasing in frequency and intensity with expected but little understood consequences for biodiversity. Extreme weather events can push organisms beyond their physiological thermal or hydric tolerances and thus limit where they can persist, affecting their geographic distributions. Species might be especially sensitive to extreme weather at the edges of their geographic ranges, where they are often already living near their physiological limits. However, the influence of climatic variability and extreme weather is often ignored in favor of climatic means when estimating distributional and richness patterns. Here we link hundreds of millions of citizen science bird observations from 2004-2024 to high-resolution extreme weather risk maps to explore how climatic variability and extreme weather risk alters summer and winter distributions and biodiversity patterns for 535 North American species. We find that species distribution models accounting for historical extreme weather risk performed better at predicting richness and the presence of individual species across 220 well-surveyed sites. Models incorporating extreme weather predicted narrower geographic distributions than models relying on only climatic means, with species’ ranges shrinking an average of 6% in summer and 10% in winter and range truncation observed at the range edges. These effects were observed in both seasons but were particularly strong in winter, a time with greater short-term weather variability than summer. Richness estimates were substantially lower when extreme weather was accounted for, especially in the US southwest and central plains (up to 30-40 fewer species), regions highly prone to extreme heat, cold and drought. Our results suggest that more mechanistically informed biodiversity predictions that account for extreme weather are critical for reliably predicting shifting distributional and biodiversity patterns.

  PublisherOA PDFDOI

• D-PLACE dataset derived from Kreft and Jetz 2007 'Global patterns and determinants of vascular plant diversity'

  Zenodo (CERN European Organization for Nuclear Research) · 2025-11-13

  datasetOpen accessSenior author

  Cite the source of the dataset as: Kreft H, Jetz W. Global patterns and determinants of vascular plant diversity. Proc Natl Acad Sci. 2007;104: 5925–5930.

  PublisherDOI

• Negative global-scale association between genetic diversity and speciation rates in mammals

  Nature Communications · 2025-02-20 · 6 citations

  articleOpen access

  Genetic diversity is critical for species evolution and their adaptability to global changes, while speciation rate is critical for explaining large-scale patterns of species richness. Exploring correlates of variation in genetic diversity and speciation rates across species is a major interest of evolutionary biologists, but these two questions have mostly been investigated independently. Here, we assess the relationship between intra-specific genetic diversity and speciation rate for 1897 mammal species (~one third of the total diversity) covering all mammalian orders. We find a negative association between mitochondrial genetic diversity and speciation rate across mammalian clades globally. This association is not accounted for by differences in the ecological attributes of species. Our findings suggest a systematic link between micro- and macroevolutionary processes that need to be better understood and considered when investigating determinants of either genetic diversity or speciation rates. Genetic diversity and speciation rate support adaptability and species richness patterns, respectively. Here, the authors find a negative association between mitochondrial genetic diversity and speciation rate in 1897 mammals that is not explained by ecological attributes.

  PublisherDOI

• Variability, drivers, and utility of genetic diversity-area relationships in terrestrial vertebrates

  2025-04-03

  preprintOpen access

  Maintaining genetic diversity within and among populations is critical for conservation and a prominent goal of the Kunming-Montreal Global Biodiversity Framework. However, direct estimates of genetic diversity are unavailable for most species, and time and resources are insufficient to fill these substantial data gaps and meet conservation target timelines. Robust, proxy-based predictions of genetic diversity loss would therefore be valuable for conserving genetic diversity for the many species lacking DNA-based data. We evaluated one such approach, the Genetic Diversity Area Relationship (GDAR), which describes the relationship between genetic diversity and the geographic area occupied by a species. We estimated differences in genetic diversity relative to the size of sample area using 55 previously published datasets from 51 species and found that GDARs were highly variable across species and strongly dependent on population structure. The mean change in allele count relative to area sampled across all species did not predict genetic diversity differences for individual species well. Traits correlated with population structure explained little variation in the GDAR. Our findings suggest that using a single GDAR is not appropriate to predict genetic diversity loss for individual species following area loss. Further work is needed to identify accurate methods to estimate species-specific levels of genetic diversity decline with area without genetic data. Although the GDAR remains valuable to highlight likely global patterns and scales of genetic diversity loss across many species, our results suggest it is currently too inaccurate for species-specific use.

  PublisherOA PDFDOI

• Global hotspots of butterfly diversity are threatened in a warming world

  Nature Ecology & Evolution · 2025-03-24 · 12 citations

  articleOpen accessSenior author

  Insects are in decline and threatened by climate change, yet lack of globally comprehensive information limits the understanding and management of this crisis. Here we uncover a strong concentration of butterfly diversity in rare and rapidly shrinking high-elevation climates. Integrating comprehensive phylogenetic and geographic range data for 12,119 species, we find that global centres of butterfly richness, range rarity and phylogenetic diversity are unusually concentrated in tropical and subtropical mountain systems. Two-thirds of the assessed species are primarily mountain dwelling and mountains hold 3.5 times more butterfly hotspots (top 5%) than lowlands. These hotspots only partially overlap with those of ants, terrestrial vertebrates and vascular plants (14-36%), while butterfly diversity is uniquely concentrated above 2,000 m elevation. We project that up to 64% of the temperature niche space of butterflies in tropical realms will erode by 2070, with the geographically restricted temperature conditions of mountains potentially turning these from refugia to traps for butterfly diversity. Our study identifies critical conservation priorities for butterflies and underscores the need for quantitative global assessments of at least select insect groups to help mitigate biodiversity loss in a rapidly warming world.

  PublisherOA PDFDOI

Recent grants

• Workshop: Activities supporting the international Biodiversity and Ecosystem Service Scenarios Network; March 6-10, 2016, Monte Verita, Ascona, Switzerland

  NSF · $113k · 2016–2019

• Collaborative Research: VertLife Terrestrial: A complete, global assembly of phylogenetic, trait, spatial and environment characteristics for a model clade

  NSF · $1.1M · 2014–2020

• Projecting the Effects of Environmental Change on the Geographic Ranges of Bird, Mammals, and Amphibians: A First Global Analysis

  NSF · $11k · 2010–2011

• Doctoral Dissertation Research: Measuring and Predicting Species Distribution Patterns Across Spatial Scales

  NSF · $12k · 2013–2015

• Collaborative Research: Map of Life: An infrastructure for integrating global species distribution knowledge

  NSF · $531k · 2010–2014

Frequent coauthors

• Robert Guralnick

  Florida Museum of Natural History

  204 shared

• Ajay Ranipeta

  Yale University

  170 shared

• Yanina V. Sica

  Yale University

  167 shared

• Stefan Pinkert

  Yale University

  164 shared

• Mélodie A. McGeoch

  Monash University

  160 shared

• Jennifer McGowan

  160 shared

• Matthew S. Rogan

  155 shared

• John Wieczorek

  152 shared