Prof. Marti Anderson
Institute of Information and Mathematical Sciences, Massey University, Albany Campus
Title:
Turning environmental models around: biological data can be used to predict the physical health of ecosystems
Time:
Thu, 28 May, 1pm
Venue:
Bldg 104, OR1
Abstract:
Environmental agencies often have monitoring data consisting of counts of species abundances from multiple sites through time at regional scales. A key challenge for managers is to follow temporal trajectories in community structure for individual sites and to identify changes in direction that specifically coincide with expectations under scenarios of impact. There are, however, multiple potential types of environmental stressors (different possible "directions" of impact in multivariate space), so how can these be characterised? For example, estuarine intertidal soft-sediment systems from multiple estuaries across the Auckland region are currently being monitored by the Auckland Regional Council (ARC). Two important potential stressors in these environments with increases in urbanisation of surrounding catchments include: (i) heavy metals (such as Cu, Pb and Zn) from run-off and (ii) increased inputs of fine terrigenous sediments, resulting in increased "muddiness". Multivariate models and software (now actively used by the ARC) have been developed that allow new observations to be positioned and assessed (at any given point in time) along a specific gradient of "ecosystem health" for each of these different types of impact. Furthermore, the gradients can be characterised quantitatively in terms of individual species' responses (which tend to be non-linear, heterogeneous, zero-inflated and non-symmetric) using quantile regression splines. These models greatly aid the interpretation of multivariate community-based models and also coincide directly with the ecological concept of limiting factors acting as constraints on the distribution and abundance of organisms.
