Individual-based analysis of interactions between plants: A statistical modelling approach applied to banana and cacao in heterogeneous multistrata agroecosystems in Talamanca, Costa Rica

Understanding how the spatial organization of diversified plant communities alters their performance is an important step in designing and managing diversified agroecosystems. The high level of spatial heterogeneity in tropical agroforests makes this task challenging. In 19 agroforestry plots in Talamanca (Costa Rica), we analyzed the effect of the structure of the plant community in the neighborhood of each individual cacao tree and banana plant on their growth and yield parameters. We developed an individual-based analysis in two steps. First, we selected without a priori the distance at which the number of neighboring plants of a given functional group (banana plants, cacao trees, fruit trees, or wood trees) best explained the proportion of attainable yield (PAY) of cacao and banana plants. In a second step, we tested the significance of the abundances of the four groups of plants in a complete model that predicted the PAY of banana and cacao plants. The abundance of neighboring plants did not increase banana PAY except in the case of other banana plants, suggesting that banana plants yield better when aggregated. All other groups of plants tended to reduce both banana and cacao PAY. In the case of wood trees, these trends were not significant. Interestingly, our results suggests that it is possible to associate banana plants and cocoa trees to moderate densities of other plants without reducing their yield. The two complete linear models predicted about 60 % of the variance of the average response of the PAY to the neighboring plant assemblage. We suggest that in the future, it would be important to differentiate processes (resources partitioning, pest and diseases) inside our statistical approach. While requiring much more data, it could be useful to address the effect of cultural practices.