Comparing correlative and bioenergetics‐based habitat suitability models for drift‐feeding fishes. (28th June 2019)

• Record Type: Journal Article
• Authors: Naman, Sean M.; Rosenfeld, Jordan S.; Neuswanger, Jason R.; Enders, Eva C.; Eaton, Brett C.
• Is Part Of: Freshwater biology. Volume 64:Number 9(2019)
• Page Start: 1613
• Page End: 1626
• Publication Date: 2019-06-28
• Subjects: foraging -- instream flows -- mechanistic model -- PHABSIM -- Salmon; Freshwater biology -- Periodicals; Biologie d'eau douce -- Périodiques; 577.605
• DOI: 10.1111/fwb.13358
• Languages: English
• ISSNs: 0046-5070

Abstract

Abstract: Quantitative habitat suitability models (HSMs) are frequently used to inform the conservation and management of lotic organisms, often in the context of instream flow management. Correlative statistical models relating hydraulic variables to habitat preferences (habitat suitability curves based on use:availability ratios) are the most common form of HSM, but face significant criticism on the grounds that habitat preference may not reflect the fitness consequences of habitat use. Consequently, there has been a drive to develop mechanistic approaches that link habitat to direct correlates of fitness. Bioenergetic foraging models relating hydraulic conditions to energy balance are particularly well‐developed for drift‐feeding fishes (e.g. salmonids) and show promise as a more mechanistic approach to modelling suitability. However, these models are rarely validated empirically or quantitatively compared with correlative HSMs. We addressed these gaps by comparing the ability of a bioenergetics‐based HSM and two correlative HSMs (a traditional suitability index and a resource selection function) to predict density and growth of stream salmonids (juvenile steelhead, Oncorhynchus mykiss, and coastal cutthroat trout, Oncorhynchus clarki ). Suitability estimates differed between the approaches, with both correlative models predicting higher suitability relative to the bioenergetic model at shallow depths and low to intermediate velocities, but lower suitability as depth increased. The bioenergetic model explained over 90% of variation in trout growth, compared to c . 50% for the correlative model. The bioenergetic model was also better at predicting fish density; however, the improvement was less striking and a high proportion of variation remained unexplained by either method. Differences in suitability estimates between approaches probably reflect biotic interactions (e.g. territorial displacement or predation risk) that decouple realised habitat use from energetics‐based estimates of habitat quality. Results highlight fundamental differences between correlative HSMs, based on observed habitat use, and mechanistic HSMs, based on the physiology and behaviour of the focal taxa. They also suggest that mechanistic bioenergetics‐based models provide more rigorous estimates of habitat suitability for drift‐feeding stream fishes. The bioenergetics approach is readily accessible to instream flow practitioners because model predictions are expressed in terms of traditional habitat suitability curves.