Shevtsov Jane, Rael Rosalyn
Division of Life Sciences, UCLA, Los Angeles, CA, United States of America.
Tulane-Xavier Center for Bioenvironmental Research, Tulane University, New Orleans, LA, United States of America.
PLoS One. 2015 Oct 5;10(10):e0137829. doi: 10.1371/journal.pone.0137829. eCollection 2015.
Indirect interactions between species have long been of interest to ecologists. One such interaction type takes place when energy or materials flow via one or more intermediate species between two species with a direct predator-prey relationship. Previous work has shown that, although each such flow is small, their great number makes them important in ecosystems. A new network analysis method, dynamic environ approximation, was used to quantify the fraction of energy flowing from prey to predator over paths of length greater than 1 (flow indirectness or FI) in a commonly studied food web model. Web structure was created using the niche model and dynamics followed the Yodzis-Innes model. The effect of food web size (10 to 40 species) and connectance (0.1 to 0.48) on FI was examined. For each of 250 model realizations run for each pair of size and connectance values, the FI of every predator-prey interaction in the model was computed and then averaged over the whole network. A classification and regression tree (CART) analysis was then used to find the best predictors of FI. The mean FI of the model food webs is 0.092, with a standard deviation of 0.0279. It tends to increase with system size but peaks at intermediate connectance levels. Of 27 potential predictor variables, only five (mean path length, dominant eigenvalue of the adjacency matrix, connectance, mean trophic level and fraction of species belonging to intermediate trophic levels) were selected by the CART algorithm as best accounting for variation in the data; mean path length and the dominant eigenvalue of the adjacency matrix were dominant.
物种之间的间接相互作用长期以来一直是生态学家感兴趣的领域。当能量或物质通过一个或多个中间物种在具有直接捕食关系的两个物种之间流动时，就会发生这样一种相互作用类型。先前的研究表明，尽管每一次这样的流动都很小，但它们的数量众多，使其在生态系统中具有重要意义。一种新的网络分析方法——动态环境近似法，被用于在一个常用的食物网模型中，量化能量从猎物流向捕食者且路径长度大于1的比例（流动间接性或FI）。食物网结构使用生态位模型创建，动态变化遵循约齐斯 - 英尼斯模型。研究了食物网规模（10至40个物种）和连通度（0.1至0.48）对FI的影响。对于每一对规模和连通度值运行的250个模型实现中的每一个，计算模型中每个捕食者 - 猎物相互作用的FI，然后在整个网络上进行平均。然后使用分类与回归树（CART）分析来找到FI的最佳预测因子。模型食物网的平均FI为0.092，标准差为0.0279。它倾向于随着系统规模的增加而增加，但在中间连通度水平达到峰值。在27个潜在的预测变量中，只有五个（平均路径长度、邻接矩阵的主导特征值、连通度、平均营养级以及属于中间营养级的物种比例）被CART算法选为最能解释数据变化的变量；平均路径长度和邻接矩阵的主导特征值起主导作用。
