Search bioRxiv⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/2024.07.14.603429

Predicting future patterns of land cover from climate projections using machine learning

Stepinski, T.
View original source ↗Export metadata ↓

Abstract

Vegetation plays a crucial role in the Earths system, and its characteristics are strongly influenced by climate. Previous studies have investigated the climate-vegetation relationship, often attempting to predict vegetation types based on climate data. Many of them have utilized biome types as proxies for different vegetation forms. Biomes, although widely used, are not always optimal for this task. They are broadly defined, a priori linked to climate, and subject to change over time. This study proposes a novel approach by using the local composition of land cover (LC) categories as descriptors of vegetation types and examines the feasibility of modeling such compositions based on climate data. The investigation focuses on the Sahel region of Africa, which is tessellated into 5 x 5 km square tiles, serving as the basic units of analysis. The independent variable comprises a set of bioclimatic variables assigned to each tile, while the dependent variable consists of shares of each LC category within the tile. The modeling framework involves a set of n regressions, one for each LC category. The K-nearest neighbors (KNN) algorithm is employed to ensure that interdependently predicted shares sum up to 100%. The models performance is validated using 2010 data, where both climate and LC information are available. The mean absolute value of residuals ranges from 1% to 11%, depending on the LC category. Subsequently, future predictions of LC patterns are made for 2040, 2070, and 2100 using climate projections under IPCC scenarios 370 and 585. A novel visualization technique called synthetic landscape is introduced to visually compare the temporal sequence of predicted LC maps from 2010 to 2100 with similar sequences of biome maps and Koppen-Geiger climate type maps. This comparison highlights overall similarities across all sequences but also reveals some significant differences.

Explore related subjects

ecologyPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Stepinski, T.. 2024-07-17. Predicting future patterns of land cover from climate projections using machine learning. https://doi.org/10.1101/2024.07.14.603429

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Rapid estimation of insect biodiversity using AI assisted image analysis:a cost-effective tool for monitoring and ecological assessmen

Protection of natural areas and environmentally friendly management of anthropogenic landscapes require reliable tools for monitoring and comparing biodiversity. Bioindicator-based approaches can support large-scale ecological assessment, but insect monitoring remains constrained by the taxonomic expertise required for species identification. Here, we propose an AI-supported index for estimating the biodiversity of flying insects, based on the diversity of visual features of insect bodies. Flying insects were captured using yellow sticky traps across agricultural and natural habitats, seasons, crops, and sites in Israel's Mediterranean region. Traps were scanned at high resolution and uploaded to a dedicated image database. A machine learning model was used to detect and delineate insects, after which engineered visual features representing body size, color, and shape were extracted. Biodiversity was estimated from distances among feature vectors and compared with taxonomic diversity derived from expert identification of a representative validation set. The AI-derived diversity index correlated strongly with expert-based taxonomic diversity, supporting the hypothesis that morphological variation in insect images can provide a proxy for taxonomic diversity. Applying the index to the full dataset revealed a negative relationship between insect abundance and diversity, while diversity also varied across habitats and sampling seasons. Our results demonstrate the potential of AI-based image analysis to provide rapid and cost-effective assessments of insect biodiversity without requiring complete taxonomic identification. This approach may facilitate biodiversity monitoring and support timely ecological assessment and decision-making in sustainable agricultural and landscape management.

ecology↗
bioRxiv2026

Pervasive ethanol in the fruits and floral nectars consumed by tropical frugivores

Dietary ethanol has long been proposed as a selective force shaping the diets and physiology of tropical frugivores and nectarivores, yet this idea has rested on measurements from only a few species and sites. Here, we present the first global survey of ethanol in wild fruits and flowers (591 and 37 species, respectively, across 18 countries) and test ecological and phylogenetic predictors of its variation. We detected ethanol in the ripened fruits of 72% of angiosperm species (211 species, N [≥] 3 per species; 0.021% - 3.9% alcohol by volume (ABV)). Ethanol was also detected in the flowers of 12 species (0.037% - 0.73% ABV). Maximum fruit ethanol varied with temperature, humidity, and phylogenetic relatedness, and it was higher in ripe fruits collected from the forest floor relative to their canopy counterparts. We then extrapolated potential daily ethanol consumption for 10 primate species. Estimated maximum daily ethanol ingestion for primates ranged from negligible to 0.72 g/kg/day on average (equivalent to 5.1 international standard drinks, 10 g ethanol each, for a 70 kg human). Our findings show that ethanol is widespread in plant reproductive tissues and shaped by both phylogeny and environment, supporting the hypothesis that dietary ethanol has been a pervasive selective pressure on fruit- and nectar-feeding animals.

ecology↗
bioRxiv2026

Thermal Similarity Attenuates the Strength of Bergmanns Rule in Birds

Geographic body size variation, classically framed as Bergmann's rule, is a fundamental indicator of temperature-mediated morphological adaptation. However, empirical support for Bergmann's rule in birds remains highly unpredictable across lineages. We hypothesized that this variation is caused by differences in thermal similarity among species within the same genus, which may limit temperature driven changes in body size and be influenced by ecological factors. Using a global dataset comprising 5,715 bird species across 513 genera, we observed that thermal similarity was the strongest predictor of variation in the strength of Bergmann's rule among competing hypotheses. Among genera exhibiting significantly negative body mass temperature relationships, higher thermal similarity was consistently associated with less negative slopes, indicating weaker Bergmannian clines. Subgroup analyses indicated that the positive thermal similarity slope association was supported in dense and semi open habitats and among nonmigratory birds, but was not clearly supported in open habitats or among migratory birds. Our findings provide a novel explanation for variation in the strength of Bergmannian clines among avian lineages, showing that stronger morphological differentiation along thermal gradients is associated with greater thermal contrast. Consequently, our findings provide a basis for testing whether incorporating thermal similarity and ecological context can improve forecasts of avian morphological responses to ongoing climate change.

ecology↗