Juvenile salmon growth across different habitat types in Central Valley, CA
Dudley, Peter (2022), Juvenile salmon growth across different habitat types in Central Valley, CA, Dryad, Dataset, https://doi.org/10.7291/D1S68V
The size of an organism is an important factor for a variety of physiological and ecological processes. For fishes, larger size can increase long-term survival and provide a population-level benefit. Therefore, threatened and endangered species management often focuses on supporting high-quality habitat that provides growth opportunities. There are numerous habitat characteristics that can affect growth including food availability, temperature, and habitat complexity. Understanding how growth responds to habitat characteristics of different quality is the first step in determining what could lead to increased growth and potentially increased individual survival. We use Bayesian techniques to determine which of the differing methods is the best for modeling the effects of habitat type and temperature on growth. To apply this method, we gather data from previous studies of the growth benefits of differing habitats and temperature regimes on Chinook salmon on the Sacramento River, CA, USA. We find a consistent growth benefit of floodplain rearing across multiple studies and show that a Ratkowsky model is the best for modeling this growth data. This information can specifically help managers model and protect the endangered and threatened Chinook populations in this system and more generally understand fish growth across differing habitat types.
We conducted searches on Web-of-Science for articles that matched the following term set for either the title or topic: (delta OR bay OR river OR floodplain) AND (Chinook) AND (feeding OR growth OR consumption OR size) AND (Sacramento OR California OR "Central Valley"). We then filtered the 253 resulting articles based on title and abstract and finally selected those which contained data on fish growth rates, temperature, and habitat type. We also flagged articles that did not present these data, but which indicated that researchers had collected data of those types. We contacted the authors of these articles requesting their data, and authors of other studies to find reports or unpublished data. This effort resulted in 10 studies with useable data (Sommer et al. 2001, 2020; Jeffres et al. 2008, 2020; Katz et al. 2014, 2019; Jeffres 2017; Takata et al. 2017; Cordoleani et al. 2020, 2021). While most of the studies used caged fish, one included data from tagged free swimming fish (Takata et al. 2017). These studies covered two habitat types; lower velocity, higher temperature, more eutrophic floodplain habitat and faster velocity, lower temperature, more oligotrophic riverine habitat. As there were no data for very high or low temperatures (field data range 10.2–16.4 °C), we used lab data for temperatures > 23 °C (Brett et al. 1982; Yanke 2006) and temperatures < 6 °C (Stauffer 1973) to anchor the fits close to the minimum and maximum growth temperatures. As feeding conditions in the floodplain are likely closer to the ad libitum conditions of the lab, we only used the high temperature lab data to assist in the fitting of the floodplain data.
From each study, we extracted the start and end date of the experiment, the start mass or length, the end mass or length, the growth rate, the habitat type (floodplain or riverine), and the mean temperature over the experimental window. We chose to use the mean temperature over the whole window because in some reports finer scale data were not available. This reduced the noisiness of the data and allows management-models with course time resolutions to use our results. For studies that only reported mass, we converted mass to length (length [cm] = (mass [g] / 0.0189)0.34. We chose to model length as it is often the metric used in management, measured in the field when processing a large number of captured fish, and used in management models. These parameters are based on averages from literature (Petrusso and Hayes 2001; MacFarlane and Norton 2002; Kimmerer et al. 2005; Chapman et al. 2013; Michel et al. 2013).
R programing languague
Bureau of Reclamation