Ressources numériques en sciences humaines et sociales OpenEdition Nos plateformes OpenEdition Books OpenEdition Journals Hypothèses Calenda Bibliothèques OpenEdition Freemium Suivez-nous

Testing the purported increases in tree mortality across the world’s tropical forests/ Tester les supposées augmentations de la mortalité des arbres dans les forêts tropicales du monde

Résidence de recherche organisée par Daniel Zuleta du 15 au 27 septembre 2025.

 

Participants

Daniel ZULETA (Department of Biological and Environmental Sciences, University of Gothenburg – Göteborg, Suède), David BAUMAN (AMAP, Université de Montpellier, IRD, CIRAD, CNRS, INRAE – Montpellier, France)

Résumé

A standing dead tree in the 25-ha Amacayacu Forest Dynamics Plot, Colombian Amazon (credit: Daniel Zuleta).

L’augmentation de la mortalité des arbres a été largement documentée dans les forêts tempérées, mais les preuves issues des régions tropicales demeurent fragmentaires et souvent centrées sur des événements extrêmes ponctuels. Comprendre si la mortalité des forêts tropicales est réellement en hausse ou non constitue une lacune de connaissance cruciale à combler, car la mortalité des arbres dans ces écosystèmes représente l’une des principales sources de carbone atmosphérique provenant des écosystèmes terrestres. De plus, parvenir à prédire avec précision la durée de résidence future du carbone dans la biomasse forestière tropicale demeure l’une des principales sources d’incertitude dans les modèles globaux de prévision du climat futur.

Pour combler cette lacune, nous avons harmonisé des mesures de terrain issues de 25 grandes parcelles permanentes réparties à travers les régions tropicales des Amériques, d’Afrique, d’Asie et d’Océanie, couvrant 45 ans de suivi forestier détaillé. Lors d’une résidence de recherche à la Fondation des Treilles, du 15 au 27 septembre, nous – deux jeunes chercheurs – avons élaboré un nouveau protocole reproductible pour quantifier les tendances temporelles de la mortalité des arbres et identifier les espèces les plus touchées dans ces 25 parcelles.

Nous avons utilisé des inventaires répétés pour estimer la mortalité annuelle et les trajectoires spécifiques des espèces, en ajustant des modèles bayésiens hiérarchiques de survie dépendants de l’ontogénie des arbres, tenant compte de la structure des parcelles et du contexte régional. Les analyses préliminaires réalisées durant la résidence indiquent une hausse de la mortalité dans de nombreux sites tropicaux depuis les années 1980, avec une large proportion d’espèces présentant un risque de mortalité croissant.

Le projet fournira un code transparent, des synthèses à l’échelle des sites et des évaluations par groupes d’espèces, susceptibles d’éclairer les actions de conservation de la biodiversité et d’améliorer les projections de stabilité du carbone et de la biodiversité dans les forêts tropicales.

Abstract

Increasing tree mortality has been widely reported in temperate forests, but evidence from the tropics is patchy and often centred on one-time extreme events. Understanding if tropical forest mortality is indeed increasing or not is a timely knowledge gap to fill, as tree mortality in these ecosystems is among the major sources of carbon to the atmosphere from terrestrial ecosystems, and achieving accurate predictions of the future of tropical carbon residence time in the forest biomass remains one of the major sources of uncertainty in the global models forecasting the future of our climate.

To address this gap, we harmonised ground based measurements from 25 large permanent plots across the tropical Americas, Africa, Asia, and Oceania, spanning 45 years of detailed forest monitoring. During a research residency at the Fondation des Treilles from 15 to 27 September, we – two early career researchers – built a new and reproducible workflow to quantify temporal trends in tree death and to identify which species are most affected across those 25 plots. We used repeated censuses to estimate annual mortality and species-specific trajectories, using Bayesian size-dependent multilevel survival models that account for plot structure and regional context. Preliminary analyses achieved during the stay indicate rising mortality at many tropical sites since the 1980s, with a large share of species showing increasing risk. The project will deliver transparent code, site-level summaries, and species-group assessments that can guide biodiversity conservation and improve expectations for the stability of tropical forest carbon and biodiversity.

Report

Why mortality matters

Forests do not decline only by growing more slowly; they also change when more trees die. Because mortality can spike abruptly, for example during droughts, heat waves, hurricanes, or pest outbreaks, these spikes can erase decades of biomass gains – i.e. the result of tree growth – and reshape forest communities in a short time. These changes directly influence climate regulation through trees’ crucial contribution to the water cycle, through carbon storage, biodiversity, food and medicine sources, as well as cultural values. Yet, despite its importance, tree mortality is difficult to study. It is relatively rare in any given year (trees are intrinsically long-living organisms, which translates into low mortality rates), often caused by multiple interacting factors, and historically has been measured at local scales that are hard to compare across regions. Tropical forests add a further challenge. Thousands of species coexist, and each can respond differently to environmental stresses. While some studies suggest that tree deaths are increasing in the Amazon and parts of Oceania, others show little change in Central Africa, while some regions remain poorly studied, leaving key questions unresolved. Are tropical forests experiencing increasing mortality rates? And if so, which species are most vulnerable? Addressing these questions is crucial for predicting how forests will respond to a warming planet and for guiding effective conservation and climate strategies. This was the main goal of our research residency at the Fondation des Treilles.

On the shoulders of giants

To address these challenges, we drew on long-term permanent plots that are part of the Forest Global Earth Observatory (ForestGEO) network. Before arriving at Les Treilles, we had secured access to data from 25 large-scale tropical plots across the Americas, Africa, Asia, and Oceania. Thanks to the dedication of hundreds of our collaborators who established and remeasured these forests, often under demanding field conditions, – and who will all coauthor the scientific paper that will come out of this project -, we assembled an unprecedented dataset of approximately 10 million tree-by-census observations spanning about 2,200 species. Each plot includes repeated measurements of tree diameter, species identity, and precise spatial location, providing the foundation needed to evaluate how tree mortality rates have changed through time and which species are most affected.

Two weeks, big steps

The quiet, focused time at the Fondation des Treilles allowed us to do in two weeks what might otherwise have taken months or years of scattered video calls. We first agreed on a conceptual framework for analyzing mortality trends and for separating plot-level changes from species-level responses. We then implemented a quality-assurance and quality-control protocol to harmonize species names, standardize measurement intervals, and correct potential measurement errors. In parallel, we built a reproducible analytical pipeline, from data cleaning to model fitting. This included simulating data to test different versions of complex survival models, to test their respective capacity to achieve the aimed outputs with precision and accuracy in a controlled setting. About half of our stay was partly spent on this daunting task, which provided us with a robust and powerful statistical model that we could then use on the real datasets. Our Bayesian ontogenetic survival models were then used to assess species-specific changes in mortality over time across the studied plots. This analytical pipeline is designed to be transparent, extensible to new sites, and shareable with our collaborators and the wider scientific community.

Preliminary results

By the end of the residency, preliminary analyses revealed a variety of  temporal changes in tree mortality among the many tropical sites and species since the 1980s, with varying magnitudes of change among species, sites, and tropical regions. Further testing the robustness of those results and better characterising mortality changes as well as the proportions of species and sites showing mortality increases and decreases will be key next steps on which we will continue working in the coming months. These results will be essential to definal potential declines in biomass and carbon storage, or shifts in species composition that could reduce or maintain resilience to future extremes. The key objective of the retreat at the Fondation des Treilles was (1) the harmonization of the 25 large datasets, (2) the generation of a working statistical Bayesian survival model, (3) the rigorous validation of this model against controlled simulated data, (4) and the writing of a manuscript outline. We achieved all those objectives, and had the bonus to already have some preliminary results, however the major step of result obtention will come in the coming months, thanks to the most difficult phase of the work that consisted in the above first three steps.

Which species are at risk

In addition to the above-defined next steps, understanding which species drive the observed changes in mortality is our next step. We will focus exclusively on species-level responses using the plot data already assembled. With Bayesian multilevel models, we will estimate species-specific mortality trends through time while accounting for repeated censuses within plots and differences among regions. We will identify vulnerable functional groups, highlight species that appear resilient, and generate testable hypotheses about the ecological mechanisms behind the patterns.

What comes next

The residency accelerated our collaboration with site managers and scientists across the tropics. Additional plots are now being prepared for inclusion, expanding the geographic and environmental coverage of our analysis. Over the coming months we will: finalize the harmonized dataset, complete the hierarchical models and sensitivity analyses, and produce the manuscript to be submitted to a scientific journal.


OpenEdition vous propose de citer ce billet de la manière suivante :
ldiebold (27 octobre 2025). Testing the purported increases in tree mortality across the world’s tropical forests/ Tester les supposées augmentations de la mortalité des arbres dans les forêts tropicales du monde. Les carnets de la Fondation des Treilles. Consulté le 14 novembre 2025 à l’adresse https://doi.org/10.58079/151lx