The study, published in Nature Ecology & Evolution, reveals that neither network-level specialization nor the specialization of individual pollinators increases or decreases monotonically toward the Equator. Instead, both peak at low northern latitudes, near the boundary between the tropics and subtropics, and fall away on either side. Plant specialization tells a different story altogether: it is hemispherically asymmetric, meaning that the pattern of how choosy plants are about their visitors differs fundamentally between the Northern and Southern Hemispheres. This asymmetry echoes earlier regional findings, such as work in South Africa suggesting that specialization increases with latitude only in the Southern Hemisphere, but the new analysis places such observations within a truly global framework for the first time.

Technically, the achievement rests on an unprecedented harmonization of field data. Quantitative interaction networks record not merely whether a pollinator visits a flower, but how often, allowing ecologists to compute indices of specialization that account for the diversity of available partners. The team compiled networks contributed by more than a hundred researchers spanning every continent, calculated network-level metrics such as H2′ and species-level metrics such as d′ for plants and pollinators separately, and then modeled how these values change across latitude, climate, biodiversity and productivity gradients. Hierarchical generalized additive models allowed the researchers to capture nonlinear relationships, a crucial capability given that the patterns they uncovered are anything but straight lines.

Perhaps the most consequential finding concerns climate. When the team compared the explanatory power of latitude against that of climate variables, biodiversity measures and environmental productivity, climate most often emerged as the best-supported predictor of specialization. Network-level specialization declined with increasing mean annual temperature, though with strong differences among pollinator groups. This temperature relationship is particularly striking because it runs counter to the intuitive expectation that warmer, more biodiverse tropical communities should host the most specialized interactions. Instead, the hottest sites tend to host more generalized networks, a pattern that may reflect the sheer abundance and overlap of flowering species in hyperdiverse tropical lowlands, where any given pollinator can afford to be promiscuous.

Precipitation, meanwhile, exerts effects that are distinctly nonlinear and, remarkably, opposite in direction for plants and pollinators. Plant specialization peaks at intermediate levels of rainfall but drops at the wettest sites, whereas pollinator specialization is lowest at intermediate precipitation, often in interaction with temperature. The authors suggest plausible mechanisms rooted in the physics and phenology of tropical rain. Heavy rainfall can physically disrupt pollinator activity, favoring flowers with protective architectures and potentially diluting the reliability of any single visitor. Seasonal rainfall regimes in tropical forests also drive strong turnover in flower-visiting insects, particularly moths and butterflies, which may reshape who meets whom across the year and thus the specialization recorded in any single sampling window.

One of the study’s most vivid results is the stark divergence among pollinator groups, especially the contrasting precipitation-related patterns between birds and insects. Bird pollinators, such as hummingbirds and sunbirds, respond to moisture gradients differently from bees, flies, butterflies and beetles, reflecting profound differences in their physiology, mobility and foraging strategies. Nectar-feeding birds can fly between raindrops and travel long distances, while many insects are confined to narrow activity windows and are highly sensitive to desiccation or washout. The finding underscores a growing recognition in network ecology that treating ‘pollinators’ as a single functional category obscures the very variation that determines how ecosystems respond to environmental change.

The new analysis also settles a long-running scientific dispute. In 2012, a study in Current Biology reported that specialization in mutualistic networks decreases toward tropical latitudes, a conclusion later challenged as a ‘zombie idea’ by researchers who argued that the evidence for stronger and more specialized interactions in the tropics was weak. Subsequent studies of cacti, honey bees and specific regional floras produced a patchwork of supporting and contradicting results, partly because small datasets are vulnerable to sampling artifacts and to the strong influence of network size on measured specialization. By assembling thousands of networks under a standardized analytical protocol, and by explicitly accounting for sampling completeness, the new study provides the statistical power needed to see the true shape of the global pattern, which turns out to be a peak rather than a slope.

The implications for conservation are sobering. Pollination underpins both wild biodiversity and a large share of global food production, and the structure of pollination networks influences how resilient these services are to disturbance. Specialized interactions are generally considered more vulnerable: if a plant depends on a narrow set of pollinators, the loss of those partners can cascade through the ecosystem. Because the new study shows that specialization is concentrated in particular climatic zones, notably the low northern latitudes near the tropical–subtropical boundary, and is governed by temperature and precipitation in taxon-specific ways, the effects of future climate change on pollination networks are likely to be geographically and biologically uneven. Warming may push some regions past thresholds where network structure shifts abruptly, while altered rainfall regimes could reorganize plant and pollinator specialization in opposite directions within the same community.

The researchers have made their modeling data and complete R code openly available through Zenodo, enabling other teams to scrutinize and extend the analysis, although raw interaction matrices remain restricted because many are part of ongoing projects. That transparency matters, because the study’s central message is methodological as much as ecological: global patterns in species interactions cannot be inferred from a handful of well-studied sites, and the choice of metric, the size of the network and the completeness of sampling all leave fingerprints on the results. By confronting those challenges at planetary scale, the team has replaced a tidy latitudinal dogma with a richer, more complicated map, one in which climate, hemisphere and evolutionary history of each pollinator lineage jointly determine how tightly woven the fabric of pollination really is.

Subject of Research: Global geographic and climatic patterns of specialization in plant–pollinator interaction networks

Article Title: Global patterns in plant–pollinator specialization

Article References: Sakhalkar, S. P., Blüthgen, N., Burkle, L. A., CaraDonna, P., Dalsgaard, B., Dormann, C. F., Kaiser-Bunbury, C. N., Knight, T. M., Ollerton, J., Resasco, J., Schleuning, M., Vázquez, D. P., Afagwu, R. N., Alarcón, R., Amorim, F. W., Amorim, M. D., Anýž, D., Arroyo-Correa, B., Artamendi, M., … Tropek, R. (2026). Global patterns in plant–pollinator specialization. Nature Ecology & Evolution. https://doi.org/10.1038/s41559-026-03170-7

Image Credits: AI Generated

DOI: 10.1038/s41559-026-03170-7

Keywords: plant–pollinator networks, ecological specialization, latitudinal gradient, biogeography, climate, pollination, ecological networks, biodiversity, precipitation, temperature, community ecology, conservation