Successional habitat filtering of rainforest trees is explained by potential growth more than by functional traits

Species along successional gradients differ in functional traits, which may function as environmental filters and also explain differences in growth rates. Alternatively, species might be filtered by growth rates and these are explained by differences in other traits. We explored the relationships between leaf and wood traits, growth and successional habitat, and asked if growth is related to habitat because both are driven by a similar set of traits or if growth is the main trait explaining habitat and relationships with other traits are indirect.
Tropical trees are often classified as second-growth (SG) or old-growth (OG) specialists, and generalists. Since succession is a continuous process and lumping species into three groups may not do justice to gradual differences, we tested if a continuous variable based on relative tree abundance in OG and SG forests might be a better predictor than classifying trees’ habitat preferences into categories.
We measured height growth of 47 trees species planted in a reforestation trial in Costa Rica and evaluated size-standardized potential growth during the first years when there was little shading among trees. Growth and habitat were related to wood density (WD), theoretical hydraulic conductivity (Kh), specific leaf area (SLA), leaf dry matter content (LDMC) and leaf nitrogen content (N).
Potential growth rates and Kh were significantly correlated with habitat measured on a continuous scale, while growth and LDMC differed among the three groups of habitat specialization. Growth was also correlated with WD and Kh.
Habitat specialization was correlated with the first principal component of the functional traits space. However, structural equation models suggest that the relationship with leaf and wood traits is mostly indirect and potential growth is the most important trait directly related to environmental filtering along a successional gradient. In our dataset, classifiying the successional status of species using the proportion of records from OG forests explains traits about as well as a classification into three groups, but provides more information about the species’ position along the successional gradient.

Methods
Tree height was measured to calculate height growth and height growth at 1m was modelled as described in the paper. Functional traits (branch wood density, theoretical hydraulic conductivity, specific leaf area, leaf dry matter content and leaf nitrogen per dry weight) are averages for 3-12 trees following standard protocols.

HScat is a classification of trees following Chazdon, R.L. et al. (2011) A novel statistical method for classifying habitat generalists and specialists. Ecology, 92, 1332-1343

HScat is the proportion of tree individuals that was recorded in old-growth forests / (old-growth + second-growth forests)

沿演替梯度（successional gradients）分布的物种在功能性状（functional traits）上存在差异，这些性状既可能作为环境过滤（environmental filters）因子，也可解释物种生长速率的差异。反之，物种也可能受生长速率的过滤，而生长速率的差异由其他性状的差异所决定。本研究探讨了叶片与木材性状、生长速率及演替生境之间的关联，并试图厘清：生长速率与生境的关联是由于二者受同一组性状共同调控，还是生长速率本身即为解释生境差异的核心性状，而其与其他性状的关联仅为间接关联。

热带树木通常被划分为次生林（second-growth, SG）专性种、原始林（old-growth, OG）专性种与广适种三类。然而演替是一个连续过程，将物种仅划分为三类可能无法充分体现其沿梯度的渐变差异。因此本研究测试了一种基于原始林与次生林中树木相对多度构建的连续变量，是否比将生境偏好划分为分类类群更适合作为预测因子。

本研究测定了哥斯达黎加（Costa Rica）某造林试验（reforestation trial）中47个树种的树高生长（height growth）情况，并评估了树木间遮荫较弱的前几年内，经体型标准化的潜在生长速率（size-standardized potential growth）。研究分析了生长速率与生境与以下性状的关联：木材密度（wood density, WD）、理论导水率（theoretical hydraulic conductivity, Kh）、比叶面积（specific leaf area, SLA）、叶片干物质含量（leaf dry matter content, LDMC）及叶片氮含量（leaf nitrogen content, N）。

潜在生长速率与理论导水率（Kh）与连续尺度下测定的生境存在显著相关性；而树高生长速率与叶片干物质含量（LDMC）则在三类生境特化类群间存在显著差异。此外，生长速率还与木材密度（WD）及理论导水率（Kh）存在相关性。

生境特化（habitat specialization）与功能性状空间（functional traits space）的第一主成分存在相关性。然而结构方程模型（structural equation models）结果显示，生境特化与叶片、木材性状之间的关联大多为间接关联，而潜在生长速率是与演替梯度上的环境过滤直接相关的最重要性状。在本数据集内，利用原始林记录占比划分物种演替地位的方式，对性状的解释能力与三分法分类大致相当，但能提供更多关于物种沿演替梯度分布位置的信息。

研究方法
通过测定树高计算树高生长速率，并按照论文所述方法对1米处的树高生长进行建模。功能性状（枝条木材密度、理论导水率、比叶面积、叶片干物质含量及单位干重的叶片氮含量）为3~12株个体的平均值，测定过程遵循标准实验规程。

HScat为依据Chazdon R.L.等（2011）提出的生境广适种与专性种分类新统计方法定义的树木分类指标，相关研究发表于《Ecology》92卷，1332-1343页。
HScat的取值为原始林记录的树木个体数占（原始林+次生林）总个体数的比例。