Combining functional traits, the environment and multiple surveys to understand semi‐arid tree distributions. (4th December 2018)

• Record Type: Journal Article
• Authors: Pollock, Laura J.; Kelly, Luke T.; Thomas, Freya M.; Soe, Paing; Morris, William K.; White, Matt; Vesk, Peter A.
• Editors: Bartha, Sándor
• Is Part Of: Journal of vegetation science. Volume 29:Number 6(2018:Nov.)
• Page Start: 967
• Page End: 977
• Publication Date: 2018-12-04
• Subjects: Australia -- Eucalyptus -- hierarchical multi‐species model -- leaf–height–seed scheme -- mallee -- semi‐arid ecosystem; Plant ecology -- Periodicals; Plant communities -- Periodicals; Plant populations -- Periodicals; 581.7
• DOI: 10.1111/jvs.12686
• Languages: English
• ISSNs: 1100-9233

Abstract

Abstract: Questions: Relationships between species, their functional traits and environmental gradients can now be more fully understood with trait‐based multi‐species distribution models (trait‐SDMs). However, general patterns are yet to emerge from founding studies using these models, which are mostly case studies at a single scale. Here, we address the generality of trait–environment relations by asking whether these relationships hold for different sampling schemes, environmental variables and species sets. Methods: We focus on the drought and fire‐resistant "mallee" eucalypts of a semi‐arid region of southeast Australia, which are likely to face new climates and disturbance regimes under global change. We use hierarchical regression modelling to test how trait–environment relationships change for two data sets representing an extensively collected, multipurpose data set and an intensively collected data set stratified along environmental gradients. Results: Three functional traits (specific leaf area, maximum height and seed mass) explained a substantial portion of the occurrence of species along soil, water and climatic gradients, with the relationship between seed mass and soil type robust across all tests. Other trait–environment relationships changed depending on study design and species set, with soil and substrate variables more important relative to climate (precipitation) for the intensively sampled survey. Remotely sensed variables were good surrogates for some field‐based measures (soil type), but not others (land form: dune or swale). In particular, airborne soil radiometric data show promise as a spatially continuous substitute for soil texture. Conclusions: Trait‐SDMs are a powerful tool for quantifying ecological interactions, but generalizations will only be possible when sample design, scale and environmental variables are carefully considered. We show that important ecological relationships can be diluted or missed entirely in broad scale trait–environment studies that rely on remotely sensed climate variables alone. Relationships that are robust to differences in study design, growth form and ecosystem (e.g., heavier seeds on sandy soil) are the most likely to reveal general ecological processes. Abstract : How general are functional trait‐environment relationships? We find some are (seed mass influences species responses to soil type across surveys, environmental variables, and species sets) and others aren't. However, even the inconsistencies are interesting in that trait‐edaphic relationships are stronger in an intensively sampled survey whereas trait‐climate relationships are stronger with a multi‐purpose survey.