Pollination depends on relationships. A flowering plant must encounter an animal capable of moving its pollen, while the pollinator must find flowers that provide usable rewards. Some of these relationships are broad and flexible, whereas others are highly specialised. A new global analysis suggests that the geography of this specialisation is considerably more complicated than a simple tropical-to-temperate gradient.
Published in Nature Ecology & Evolution on 30 September 2026, the study brings together more than 3,400 quantitative plant-pollinator networks containing more than 110,000 recorded interactions. Rather than finding that ecological partnerships become steadily more specialised towards the Equator, the researchers found distinct patterns for plants, pollinators and entire interaction networks. Climate, particularly temperature and precipitation, often explained the structure of these partnerships better than latitude itself.
A global question with no simple latitudinal answer
Ecologists have long been interested in whether interactions between species change predictably from the poles to the tropics. Tropical ecosystems contain exceptional biodiversity, which could create more opportunities for specialised relationships. At the same time, high diversity may also allow organisms to interact with many alternative partners. Earlier studies have therefore produced competing expectations about how specialisation should vary with latitude.
Sailee P. Sakhalkar and an international team of researchers addressed this problem at an unusually large scale. Their database combined quantitative networks assembled from ecological studies around the world. A network records which flowering plants interact with which pollinators and, crucially, how frequently those interactions occur. This allowed the researchers to examine specialisation at the level of the whole network as well as separately for plants and pollinators.
The study considered more than 3,400 networks and over 110,000 plant-pollinator interactions. The researchers analysed broad geographic patterns while also accounting for environmental variables and differences among functional groups. They used generalised additive modelling approaches that can detect curved and non-linear relationships rather than forcing the data into a simple straight-line trend.
Specialisation did not simply rise towards the tropics
The central result challenges the idea of a universal latitudinal gradient. Across the global dataset, there was no consistent monotonic increase or decrease in specialisation as ecological networks approached the Equator.
Instead, network-level specialisation and pollinator specialisation peaked at low northern latitudes around the tropical-subtropical boundary. Plant specialisation followed a different pattern and showed hemispheric asymmetry. In other words, asking whether pollination becomes more specialised in the tropics produces an incomplete answer because the result depends on which side of the interaction is being measured and where in the world the community occurs.
This distinction matters. A network can appear relatively specialised because pollinators concentrate their visits on a limited subset of plants even when the plants themselves receive visits from a broader range of animals. Treating plants and pollinators as though they necessarily respond to geography in the same way can therefore hide important ecological structure.
Climate often explained more than latitude
When the researchers compared possible predictors, climatic variables were most often better supported than latitude, biodiversity or environmental productivity. Mean annual temperature was particularly important at the network level. Overall network specialisation declined as mean annual temperature increased, although the direction and strength of relationships differed substantially among pollinator groups.
Precipitation produced an even more nuanced pattern. Plant specialisation was highest at intermediate levels of precipitation and lower at the wettest sites. Pollinator specialisation showed a contrasting tendency, becoming lower at intermediate precipitation in the broader analyses, often through interactions with temperature.
Birds and insects were especially informative because their precipitation-related patterns differed strongly. That divergence makes biological sense. Pollinator groups differ in physiology, mobility, activity periods, floral preferences and sensitivity to weather. A hummingbird, bee, moth and hoverfly may all transfer pollen, but they do not experience temperature and rainfall in equivalent ways.
Why this matters under climate change
The findings shift attention from latitude as a geographic shorthand towards the environmental mechanisms that may actually shape ecological relationships. Latitude correlates with many aspects of climate, but organisms respond to conditions such as temperature and rainfall rather than to degrees on a map.
That distinction becomes increasingly important as climate changes. Rising temperatures and altered rainfall can change where species live, when flowers open and when pollinators are active. Even if both members of a partnership remain present in a region, changes in timing or abundance can reduce the probability that they encounter one another.
Specialised communities may face particular risks when environmental change removes an important partner. A plant visited by many interchangeable pollinators may retain some reproductive opportunities if one species declines. A plant dependent on only a small set of effective pollinators has fewer alternatives. The same logic applies to pollinators that rely heavily on particular floral resources.
The new results do not imply that specialised networks will universally fare worse. Instead, they suggest that vulnerability is likely to be uneven. Because different pollinator groups show different climatic relationships, future ecological change may reorganise interaction networks in ways that cannot be predicted from latitude alone.
A major synthesis still carries limitations
The scale of the dataset is a major strength, but global ecological syntheses also inherit limitations from the studies they combine. Sampling intensity, observation methods, seasons and taxonomic resolution can vary among source networks. Geographic research effort is also uneven, meaning that some regions and ecological communities are represented more densely than others.
The study evaluates observed interaction structure, not the future survival of individual species. An association between climate and specialisation therefore does not establish that a particular amount of warming or rainfall change will cause a predictable loss of pollination. Climate can also operate indirectly by changing vegetation, flowering phenology, species distributions and community composition.
The authors additionally note that raw interaction matrices are not all publicly available because some remain part of ongoing projects, although the modelling data and complete R code used for the analysis have been deposited in Zenodo. This improves reproducibility of the reported modelling while preserving restrictions attached to some contributing datasets.
A more detailed map of ecological dependence
The study’s broader contribution is to replace a simple global rule with a more realistic ecological picture. Plant-pollinator specialisation is geographically structured, but not along one universal north-to-south gradient. Plants and pollinators can show different patterns, and climatic conditions frequently provide more explanatory power than latitude alone.
For conservation, this means that protecting pollination systems may require attention not only to how many species remain in an ecosystem, but also to who interacts with whom and how dependent those relationships are. As temperature and precipitation regimes change, the architecture of ecological partnerships may shift differently across regions and taxonomic groups.
Source Information
Study: Global patterns in plant-pollinator specialization
Authors: Sailee P. Sakhalkar and colleagues
Journal: Nature Ecology & Evolution
Published: 30 September 2026
DOI: 10.1038/s41559-026-03170-7
