Comparison of water quality thresholds for macroinvertebrates in two Mediterranean catchments quantified by the inferential techniques TITAN and HEA

Abstract

Catchment managers face complex decisions to balance between competing targets such as water quantity and quality, land use and biodiversity. To balance these targets satisfactorily, decisions need to be grounded in reliable information. The present study aimed to identify potential water quality drivers that may impact the sustainability of the River Torrens and the Onkaparinga River catchments in South Australia by means of the Threshold Indicator Taxa Analysis (TITAN) and the Hybrid Evolutionary Algorithm (HEA). Fourteen years of data comprising abundance for 518 macroinvertebrate taxa and water quality variables were examined to determine possible stimulating and inhibiting effects of water quality conditions on macroinvertebrates at catchment and stream site level. When applied to the Torrens (n = 262) and Onkaparinga (n = 227), TITAN identified positive (z+) and negative (z−) responding species for both catchments and detected community change points in response to total nitrogen (Torrens: z+ = 305 µg L−1, z− = 277 µg L−1; Onkaparinga: z+ = 420 µg L−1, z− = 397 µg L−1), total phosphorus (Torrens: z+ and z− = 22 µg L−1; Onkaparinga: z+ = 73 µg L−1, z− = 71 µg L−1) and electrical conductivity (Torrens: z+ = 931 µS cm−1, z− = 407 µS cm−1; Onkaparinga: z+ = 951 µS cm−1, z− = 428 µS cm−1). These results suggested similar change points along electrical conductivity (EC) and total nitrogen (TN) gradients in both catchments, but three-times higher change points for total phosphorus (TP) in the Onkaparinga River catchment than in the Torrens catchment.Since TITAN identified the species Austrochiltonia australis as responding positively to TP, TN and EC increases, and Paramerina spp. as responding negatively, these taxa were selected for modelling by HEA. In contrast to TITAN, HEA used averaged data from 37 stream sites in each of the two catchments. Resulting models for A. australis revealed much higher change points for TN, TP and EC in the Torrens catchment compared to the TITAN results. Models for Paramerina spp. suggested twice as high thresholds for TN and EC in Onkaparinga as compared to the Torrens. Whilst TN and TP thresholds specified by HEA fall within the range identified by TITAN, thresholds for EC exceeded significantly those identified by TITAN.Overall, the study has shown that the nature of the data used in the two threshold approaches is likely to have resulted in differences in the derived thresholds. Importantly, despite the differences observed in thresholds based either on a catchment or the modelling technique, overall results indicate that South Australian water quality guidelines for freshwater systems are too high.