How short are we on the goal line of estimating insect-plant interaction in a hyperdiverse tropical environment?

Abstract

In ecology, dealing with sample design is a constant challenge, especially in interaction networks, where most studies suffer from undersampling. Different sampling methods are used in interaction networks, from continuous observations over time to observations at the peak of flowering of plants. However, species richness and abundance vary crucially over time. Therefore, there is significant uncertainty about how far we are from complete sampling. This study used data from a biweekly sampling of insect-plant interactions in a highly diverse biome over an entire year to quantify this uncertainty by identifying the required effort to achieve sampling completeness. We also extended this investigation to each of the four seasons. We used accumulation curves based on abundance data to estimate the sampling deficiency. In addition, we fit nonlinear models to estimate the sampling effort required to achieve completeness based on six network-level metrics. Subsequently, we sought to identify the factors responsible for the difference in sampling completeness. We found that the sampling effort carried out was sufficient to record most animals and blooming plants throughout all seasons. As for interactions, a substantial increase in sampling effort would be required to achieve sampling completeness, especially in fall and spring. Furthermore, we have seen that the dissimilarity of interactions between seasons is the main factor that influences the completeness of the sampling. Finally, we observed that the main factor influencing the dissimilarity of interactions was the links turnover and that such alterations possibly occur due to changes in species composition, especially of plants.