Homing in on factor interactions

Posted on March 27, 2017

In most cases the dynamics of ecological systems are best explained as responses to a number of interacting environmental factors, rather than to a single variable. Such interactions can be complex, and although they may have major implications for management, they’re typically very poorly understood. A team of Australian researchers has proposed a method for detecting and assessing interactions between ecological factors, using as an example the responses of river red gum (Eucalyptus camaldulensis) to a range of environmental variables. While noting that some interactions can be positive and some negative, the researchers also acknowledge the existence of “qualitative” interactions, in which the effect of a given factor switches from positive to negative (or vice versa) depending on the value of another factor. Drawing on Bayesian network modelling, their approach can be used with small datasets and easily allows for the exploration of different types of interactions and management scenarios. An example study was based on data collected from 37 floodplain wetlands in the Condamine catchment of the Murray–Darling Basin, in Queensland. Tree abundance and tree condition were used as measures of the responses of red gums to six hydrological variables and four land use variables. Depending on whether or not factor interactions were considered, the perceived consequences of different management actions changed dramatically – for instance, the negative impact of grazing on red gums was stronger in areas where groundwater was relatively deep. Qualitative interactions were also detected: for example, frequent wetland inundation had a positive influence on tree vigour in wetlands minimally impacted by weirs, but a negative influence in wetlands close to weirs. Qualitative interactions may be especially important in highly altered landscapes: for instance, in lightly-managed floodplains, hydrologic connectivity has a strong positive effect on the distribution of species, but in intensively managed landscapes, such as urban and irrigated agricultural areas, high connectivity may have a negative effect by spreading invasive species and pollutants.

Reference: Kath, J.M. et al. 2016. Using a Bayesian network model to assess ecological responses to hydrological factor interactions Ecohydrology 9, 9–18. http://onlinelibrary.wiley.com/doi/10.1002/eco.1597/abstract