Abundance drives broad patterns of generalisation in plant–hummingbird pollination networks

Abundant pollinators are often more generalised than rare pollinators. This could be because abundance drives generalisation: neutral effects suggest that more abundant species will be more generalised simply because they have more chance encounters with potential interaction partners. On the other hand, generalisation could drive abundance, as generalised species could have a competitive advantage over specialists, being able to exploit a wider range of resources and gain a more balanced nutrient intake. Determining the direction of the abundance-generalisation relationship is therefore a ‘chicken-and-egg’ dilemma. Here we determine the direction of the relationship between abundance and generalisation in plant-hummingbird pollination networks sampled from a variety of locations across the Americas. For the first time we resolve the direction of the abundance-generalisation relationship using independent data on animal abundance. We find evidence that hummingbird pollinators are generalised because they are abundant, and little evidence that hummingbirds are abundant because they are generalised. Additionally, a null model analysis suggests this pattern is due to neutral processes: most patterns of species-level abundance and generalisation were well explained by a null model that assumed interaction neutrality. These results suggest that neutral processes play a key role in driving broad patterns of generalisation in animal pollinators across large spatial scales.DeclarationsFunding – BIS is supported by the Natural Environment Research Council as part of the Cambridge Earth System Science NERC DTP [NE/L002507/1]. JVB was funded by CERL - Engineer Research and Development Center. PKM was funded by the São Paulo Research Foundation (FAPESP grant #2015/21457-4). PAC was funded by the David Lack studentship from the British Ornithologists’ Union and Wolfson College, University of Oxford. CL was funded by the ESDEPED-UAT grant. MAM acknowledges the Consejo Nacional para Investigaciones Científicas y Tecnológicas (Costa Rica), German Academic Exchange Service and the research funding program ‘LOEWE-Landes-Offensive zur Entwicklung Wissenschaftlichö konomischer Exzellenz’ of Hesse’s Ministry of Higher Education, Research, and the Arts (Germany). ROP was funded by CONACyT (project 258364). MAR was supported by the State of São Paulo Research Foundation (FAPESP) within the BIOTA/FAPESP, The Biodiversity Institute Program (www.biota.org.br) and the ‘Parcelas Permanentes’ project, as well as by Coordenação de Pessoal de Nível Superior (CAPES), Fundo de Apoio ao Ensino e à Pesquisa (FAEP)/Funcamp/Unicamp and The Nature Conservancy (TNC) of Brazil. LCR was supported by CNPq and Capes. MS was funded by CNPq (grant #302781/2016-1). AMMG is supported through a Marie Skłodowska-Curie Individual Fellowship (H2020-MSCA-IF-2016-704409). LVD was supported by the Natural Environment Research Council (grants NE/K015419/1 and NE/N014472/1). AMMG, JS, CR and BD thank the Danish National Research Foundation for its support of the Center for Macroecology, Evolution and Climate (grant no. DNRF96). WJS is funded by Arcadia.