Functional trait variation along a hydrological gradient and trait-based predictions of the composition of a wetland plant community

Predicting the assembly of plant communities is considered the Holy Grail of functional ecology and has never been more important as we head into an era of environmental change. Studying plant functional traits provides the best opportunity for understanding the community assembly processes that determine the abundance and distribution of plant species. Plant functional traits provide information on the direct physiological adaptations of plants to various environmental conditions. The assembly of plant communities is driven by filtering processes that select for or against certain functional traits and a plant can only be present within a community if it contains the functional traits necessary to germinate, survive and compete in the environment of the community. An understanding of how functional traits are filtered by the environment and biotic interactions provides the foundations for predictive community assembly models. However, the understanding of how functional traits are filtered along hydrological gradients is poor for the majority of functional traits. In this thesis I aimed to identify how plant functional traits respond to variation in soil hydrology in the presence and absence of grazing and determine whether a trait-based model of environmental filtering could predict the composition of an ephemeral wetland plant community. To achieve these aims, I performed a survey of the plant community in an ephemeral wetland in grazed and ungrazed transects. The survey was conducted along a hydrological gradient that was split into an elevation gradient above the flood line and a flooding gradient below the flood line. I measured nine root, leaf and shoot traits on 885 plant samples collected during the community survey and investigated the response of community-weighted and individual-level traits along the hydrological gradient using Generalized Additive Models (GAMs) and Generalized Linear Models (GLMs). To determine whether a trait-based model of environmental filtering could predict the composition of the plant community, I incorporated the individual-level trait relationships into a community assembly model known as Traitspace and predicted the relative abundance of the 23 most common species found within the Kettlehole. Community-weighted root traits were more strongly related to the hydrological gradient than aboveground traits. Root aerenchyma increased as the number of days submerged increased while root dry matter content (RDMC), specific root length (SRL), root tissue density (RTD) and root branching intensity (RBI) decreased with increasing flooding. Community-weighted specific leaf area (SLA) and root tissue density were more closely related to the elevation gradient than any other traits. SLA decreased as elevation above the flood line increased while RTD increased with elevation. The relationships between individual-level traits and the elevation and flooding gradients were far weaker than community-weighted traits but showed similar trends in the directions of trait responses along the gradients. Grazing reduced the community-weighted trait values of all traits except SLA and aboveground dry matter content (AGDMC) along the flooding gradient and had little effect on the trait values observed along the elevation gradient. Using environmental filtering of functional traits, the Traitspace model was able to predict the distribution and abundance of a number of key species within the wetland plant community but was unable to accurately predict the abundance and distribution of other species. The accuracy of the Traitspace model was best when all nine functional traits were used to produce predictions of species abundances but SLA and height were recognised as the two traits to provide the most predictive ability. The strong filtering of root traits illustrate the important information that root traits provide in terms of the adaptations of plants in environments with varied soil hydrology. Easily measurable aboveground traits are often favoured in functional ecology but these results highlight the importance of measuring root traits in trait-based research. Traitspace has the ability to predict the abundance and distribution of some species within a wetland plant community using environmental filtering of functional traits. Community assembly at small spatial scales is mostly driven by biotic interactions rather than environmental filtering which may limit the power of current predictive models. The inclusion of biotic interactions into trait-based models will improve the predictions of community composition at small spatial scales in wetland ecosystems and help us to progress towards achieving the goal of accurately predicting the composition of plant communities