Selection of river flow indices for the assessment of hydroecological change

Abstract

• This is the pre-peer reviewed version of the article, which has been pub- Author for Correspondence:Wendy Monk, Canadian Rivers Institute, Department of Biology, Bag Service #45111, University of New Brunswick, Fredericton, New Brunswick, E3B 6E1, Canada. 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.3 IntroductionThe ecological importance of river flow regime variability is increasingly well recognised (e.g., Clausen and Biggs, 1997;Wood and Armitage, 2004); and a wide range of potentially 'ecologically relevant' hydrological indices have been identified (e.g. Olden and Poff, 2003). However, such hydroecological analysis is limited by a general lack of paired longterm hydrological and ecological time-series (Wood et al., 2001; Jackson and Füreder, 2006). The search for 'ecologically relevant' hydrological indices has been driven by the need to quantify variability in ecological communities and/or individual populations that may be sensitive to natural hydrological changes or anthropogenic modifications (Richter et al., 1996). Some concerns have been raised regarding the large number of potential hydrological predictors available, since significant redundancy (multicollinearity) exists between many variables (Olden and Poff, 2003). Consequently, some guiding principles are required to aid researchers and water resource managers select the most 'ecologically relevant' hydrological variable(s).Olden and Poff (2003) proposed a method using principal components analysis (PCA) for assessing redundancy between hydrological variables and identifying those indices which account for most variation in river flow regimes using long-term flow records for 420 locations across the continental USA. They suggested that the variables identified by this method may form the basis of future hydroecological analysis. However, to date, their redundancy methodology and the resulting variables have not been widely tested in terms of ecological prediction.The aim of this short communication is to provide the first test of the PCA-based approach proposed by Olden and Poff (2003) in association with ecological data, and to compare its 4 effectiveness against regression models developed using 201 potentially 'ecologically relevant' hydrological variables identified in previous research. Data and methodsHydrological and ecological data were employed for 83 sites in England and Wales ( Figu...

Macroinvertebrate community response to inter‐annual and regional river flow regime dynamics

“…Examination of the long-term LIFE scores for rivers at the national scale clearly indicated that the inter-annual pattern (1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) closely tracked, or corresponded with, the 'most significant hydrological descriptor' (lnSMED = specific median discharge; Figure 4a and Figure 5a) identified from 201 flow descriptors (Monk et al, 2006;Monk et al, 2007b;Olden and Poff, 2003). This flow index provides an annual measure of river flow (discharge) scaled by catchment area.…”

“…Redundancy methods have been suggested as a means of reducing the range of hydrological predictors of ecological response. However such an approach may overlook more subtle ecological associations (Monk et al, 2007b).…”

“…The hydrological variables were used to examine how macroinvertebrate communities responded to inter-annual flow regime variability at the national scale and for the long-term regime magnitude (RM) regions. The hydrological descriptors identified previously by Monk et al (2006;2007b) demonstrating the strongest statistical association with LIFE scores for the long-term RM classes are presented herein to explore the relationship between 'flow' and instream macroinvertebrates over the 11-year biological record. Hereafter these variables are described as the 'most significant hydrological descriptors'.…”

Six decades of changes in the riparian corridor of a Mediterranean river: a synthetic analysis based on historical data sources

“… Between the 1956 and 1976 images (pre-regulation and hydropower operation), river flows were divided into two sub-periods. For 1956For -1967 Eleven hydrological indices, which are relevant for the river ecosystem (Olden and Poff, 2003;Mathews and Richter, 2007;Monk et al, 2007), were estimated from the flow record. The selected indices were a sub-set of those described by Belmar et al (2011) to characterize Mediterranean rivers whilst avoiding redundancy.…”

“…Olden and Poff, 2003;Mathews and Richter, 2007;Monk et al, 2007;Belmar et al, 2011), and we evaluated 11 of these during our study (Table 4). Although our estimates of many of these indices were probably adversely affected by the low (monthly) resolution of the river flow time series, all of the indices show distinct changes over six decades and clearly correspond to the timing as well as the impacts of dam closure on the flow regime of the Mijares River.…”