You might, sometimes, have heard the phrase ‘everything is connected’. Maybe you are thinking about computers and mobile phones, but in fact this statement is particularly true in nature. For instance, we know that species are not isolated entities, instead they are part of communities in which multiple different species are interacting with each other. Some of these interspecific interactions are cooperative and positive for all interacting partners, and are called mutualistic interactions. Virtually all species on Earth are involved in one or more mutualistic interactions. Specifically, the interactions between plants and their pollinators may be some of the most studied ones, as nearly 85% of plants rely on animals for pollination service. In the last 20 years the study of pollination interactions using network analysis has been a hot topic in ecology. Networks have proven to be a useful tool to unravel patterns in plant-pollinator interactions at the whole community level. Usually, almost all plant-pollinator networks are constructed at the species-level (species-based networks), i.e. nodes in the network are plant and animal species and links represent the interactions occurring between them (e.g. flower visits). However, species are composed of populations of individuals and those individuals are the true actors establishing interactions in nature. Even more interesting is the fact that conspecific individuals are phenotypically and behaviourally diverse with respect to, e.g. size, sex, age, and social status, which also might imply that their foraging decisions become different. Most ecological networks studied to date have not considered this intraspecific variation in interactions, despite the importance of individual variation within natural populations addressed in the theory of evolution by natural selection. For that reason, moving from species-based networks to individual-based networks, to disentangle a process, which can be defined as network downscaling, is probably one of the major challenges right now in ecological network research.

 

 

In an attempt to fill this gap of knowledge, we got the idea of downscaling an entire pollination network to the individual level for the pollinator subset and explore network patterns at both interacting scales: species and individuals. This was possible with the study of pollen loads of insect individuals. Insect flower visitors in two mountain shrub communities from Mallorca (Balearic Islands) were captured, and later in the laboratory, pollen carried by each one was identified and quantified under the microscope. It was a highly time consuming and difficult task, but it paid well off as it provided a record of the flowering species visited by each individual pollinator over time. Data revealed that generalized species in the plant-pollinator network are composed of specialized and idiosyncratic individuals. The high heterogeneity in individual foraging behaviour and the high individual specialization of pollinators are obviously hidden in traditional species-based networks, and thus determine differences in several topological properties between species-based and individual-based networks. Particularly, the modular structure – a broadly described pattern in pollination networks which consists of densely connected groups or cliques of nodes with sparse connections to other groups– is not consistent across networks at the two scales. We found that modularity increases when downscaling networks to the individual level, and we confirmed this result using different modularity detection algorithms. In contrast to the view of modules as a set of taxonomically related species or species with convergent morphological traits in species-based networks, modules in individual-based networks are groups of functionally different pollinators distantly related but with overlapping pollen niches. Thus, interestingly, conspecific individuals are distributed in different modules. Modules showed to have a strong phenological component, and attributes related to the phenophase of plants and individuals even determined the topological roles of nodes in the network. Only when downscaling to the individual level it was possible to detect a dynamical interaction switching within-species and a module turnover throughout the flowering season, thus modules of individuals assembled and disassembled over time.

 

In conclusion, findings reported in our study, “Increasing modularity when downscaling networks from species to individuals”  (Tur et al.) highlight that network patterns differed across the individuals and the species scales, because much within-species variation exists. This implies that it is not always possible to deduce structure at one hierarchical level from information about structure at an adjacent level. Combining the study of networks at both scales offers the possibility of uncovering important properties and processes, which might influence network stability, dynamics and the outcomes of interactions.

 

By Christina Tur