Selection of river flow indices for the assessment of hydroecological change
journal contributionposted on 04.09.2013, 14:06 by Wendy A. Monk, Paul WoodPaul Wood, David M. Hannah, Douglas A. Wilson
A wide range of ‘ecologically relevant’ hydrological indices (variables) have been identified as potential drivers of riverine communities. Recently, concerns have been expressed regarding index redundancy (i.e. similar patterns of variance) across the host of hydrological descriptors on offer to researchers and water resource managers. Some guiding principles are required to aid selection of the most statistically defensible and meaningful river flow indices for hydroecological analysis. In this short communication, we investigate the utility of a principal components analysis (PCA)-based method that identifies 25 hydrological variables to characterise the major modes of statistical variation in 201 hydrological indices for 83 rivers across England and Wales. The emergent variables, and all 201 hydrological variables, are used to develop regression models [for the whole data set and three river flow regime shape (i.e. annual hydrograph form) classes] for an 11-year macroinvertebrate community dataset (i.e. LIFE scores). The same ‘best’ models are produced using the PCA-based method and all 201 hydrological variables for two of the three river flow regime groups. However, weaker models are yielded by the PCA-based method for the remaining (flashy) river flow regime class and the whole data set (all 83 rivers). Thus, it is important to exercise caution when employing data reduction/ index redundancy approaches, as they may reject variables of ecological significance due to the assumption that the statistically dominant sources of hydrological variability are the principal drivers of, perhaps more subtle (sensitive), hydroecological associations.
- Social Sciences
- Geography and Environment