Does diversity beget diversity? The case of host specificity of vascular epiphytes

热带雨林的树冠上栖息着数量庞大但数量未知的动植物物种。这种物种丰富度的进化和维持需要深入的研究。草食性昆虫的多样性与寄主树种的多样性有直接关系。这同样适用于种类丰富的维管附生植物群，即维管植物生长在树上而不寄生于它们。然而，为了验证这一假设，有必要量化该群体的宿主特异性。尽管对这一主题进行了许多研究，但我们离这一目标还很遥远。目前的情况是，附生植物通常能够在森林中的任何树种中定植，但观察到的附生植物丰度在宿主物种之间的分布往往不均匀。然而，由于一些方法学问题，这些观察到的偏差并不是宿主特异性的明确证明：在分析分布数据时，可能会产生统计伪影，例如，忽略附生菌的聚集性。此外，几乎没有任何明确可解释的研究方法直接关注附生植物的个体适合度。最后，由于热带雨林中附生植物和树种的数量惊人，通常应用的植物区系方法不足以根据一般规则预测它们之间的联系。拟议项目的目标是解决所有这些问题。子项目1借助计算机模型研究同种附生植物空间自相关对宿主特异性测试结果的影响。为了改进对经验数据集的分析，将开发一个考虑这种影响的零模型。子项目2（一个系统设计的野外研究）比较不同树种间附生植物的丰度和种类组成。相对较大的样本量将允许可靠地检测分布偏差。子项目3借助实验操作研究宿主在适合度参数（发芽、存活率和生长率）水平上的特异性。最后，子项目4从以往研究的重点物种相互作用层面出发，转向附生植物与树木性状的相互作用（性状匹配）。长期目标是机械地解释分布模式，并为性状的相关性发展一般规则。项目成功的决定性先决条件是选择巴拿马的现场。从之前的一个项目中已经有两个临时复制的数据集，描述了0.9公顷森林地块内约20,000种附生植物的空间分布。将这些数据集的分布模式与拟议项目的结果进行比较，将有助于评估这些数据集对宿主特异性研究的适用性。

The canopy of tropical rainforests hosts an enormous, still unknown number of animal and plant species. The evolution and maintenance of this species richness is subject to intensive research. For the diversity of herbivorous insects, a direct relationship with the diversity of their host tree species has been postulated. The same might apply for the very species-rich group of vascular epiphytes, i.e. vascular plants growing on trees without parasitizing them. However, to test this hypothesis it is necessary to quantify the host specificity of this group. In spite of the many studies on this topic we are still far from this goal. State of the art is that epiphytes are usually capable of colonizing any tree species in a forest but that the observed epiphyte abundances are often unequally distributed among host species. However, due to a number of methodological issues such observed biases are not an unambiguous demonstration of host specificity: When analyzing distributional data statistically artefacts might arise from, e.g., disregarding the clumped occurrence of epiphytes. Moreover, there are barely any unambiguously interpretable study approaches that focus directly on the individual fitness of epiphytes. Finally, given the daunting number of epiphyte and tree species in tropical rainforests, the usually applied floristic approach is inadequate to predict associations on the basis of general rules. The objective of the proposed project is to address all these issues. Subproject 1 investigates, with the aid of a computer model, the effect of spatial autocorrelation of conspecific epiphytes on the results of host specificity tests. To improve the analysis of empirical datasets, a null model that considers this effect will be developed. Subproject 2 (a systematically designed field study) compares epiphyte abundances and species composition among tree species. Comparably large sample sizes will permit to reliably detect distributional biases. Subproject 3 investigates host specificity on the level of fitness parameters (germination, survival and growth rates) with the aid of experimental manipulations. Finally, subproject 4 leaves the level of species interactions, which have been the focus of previous studies, and turns onto the interaction of epiphyte and tree traits (trait matching). The long-term goal is to explain distributional patterns mechanistically and to develop general rules for the correlation of traits. A decisive precondition for the success of the project is the choice of the field site in Panama. There are already two temporally replicated datasets from a previous project, describing the spatial distribution of ca. 20,000 epiphytes within a 0.9 ha forest plot. Comparing the distributional patterns from these datasets to the results of the proposed project will allow to evaluate how suitable such datasets are for the investigation of host specificity.