Multi-scale structure of ecological communities

点击翻译按钮获取中文摘要

Variation partitioning, combined with multi-scale PCNM analysis, allows researchers to differentiate the resuls of the complementary processes of environmental control (niche-based) and community dynamics (neutral processes). My long-term research objective is to test hypotheses about these processes in ecosystems. This is of great importance because quantitative changes in beta diversity, or changes in its spatial organization, indicate natural or anthropogenic changes in ecosystems. During my career, I learned to combine community ecology and statistics, creating new numerical methods that proved to be of prime importance for the progress of community ecology. The new avenues that I will explore during the next 5-year cycle are described in the Proposal. That work can only be done by teams of statistically-inclined scientists who understand the primary ecological questions and are able to relate to researchers in statistics and computer sciences.Fundamental ecology: we will analyze ecosystems for the relative importance of broad-scale physical processes versus local environmental conditions. At finer scales, we will assess the relative importance of neutral versus niche-based processes. PCNM analysis will be applied to multivariate ecological time series. It will also be extended to test the importance of historical events in shaping present-day communities, providing clues about the effect of ongoing anthropogenic changes on communities. Fourth-corner analysis will be generalized to make its results richer in ecological information. Fourth-corner analysis will be incorporated into a new form of variation partitioning for the study of partial relationships between species traits and environmental characteristics while controlling for the effect of well-understood environmental variables. Methodological developments: we will develop new statistics for model selection in the study of multivariate regression trees (MRT). We will also develop a new test for the space-time interaction in multi-species surveys without replication.  We will extend our single-beam sonar methodology for seafloor mapping and our SonarX R-language library to multi-beam sonars, which provide more information-rich sonar returns.