Applying the mean free-path length model to juvenile Chinook salmon migrating in the Sacramento River, California

Abstract

Population vital rates, such as stage-specific survival, are influenced by individual behavior and movement patterns. Yet few methods exist to incorporate behavior into predator-prey models, omitting a potentially important source of variability in population dynamics. Here were combine results from an acoustic telemetry study of juvenile Chinook salmon (Oncorhynchus tshawytscha) with an existing predator prey model, called the mean free-path length model, originally presented in Anderson et al. (2005). The model describes the probability of predator-prey encounters as a function of the predator density and the movement patterns of predators and prey. Greater predator densities and greater variation in movement vectors should result in higher predator-prey encounter rates, and lower survival for the prey. Fitting this model to data provides insight into mechanisms of mortality for migrating fishes. Here we estimate model parameters for two flow conditions in the Sacramento River, California, examining the importance of natural environmental variation in shaping encounters and prey survival. Survival estimates were similar between the high and low flow conditions, yet travel time was slower at lower flows. The model estimates of mean free-path length were lower when compared to those estimated in the Snake River system, corresponding with lower survival. We discuss the value of model parameters estimated from telemetry data in providing a tool for forecasting population-level responses to structural or hydrodynamic modifications in large river systems, and we explore how the XT model can provide insight into nonlinear and threshold-like responses of migratory fish survival to flow.

Appendices

Appendix 1

Six groups of acoustically tagged Chinook salmon were released into the Sacramento River, California across two study years. Fish tracks were truncated if movement patterns indicated predation of tagged smolt; detections after initiation of a prolonged upstream movement (>1 km) were removed. Long-term environmental data obtained from monitoring gauges located above study reach, termed ‘Colusa’ (39.214°N, −122.000°W), and within study reach below confluence with Feather River, termed ‘Verona’ (38.774°N, − 121.598°W). Release groups were classified into a high flow group if the mean discharge at Verona upon release was >400 m³/s. Water temperature was recorded at Verona.

Water Year	Release group	Flow group	Release date	Mean discharge (m3/s) @ Colusa	Mean discharge (m3/s) @ Verona	Mean water temp. (°C)	N release	N alive at Gate 1	N truncated tracks
2013	1	High	12/20/2012	331.3	662.6	8.2	94	76	0
	2	High	1/10/2013	288.8	552.2	7.9	100	99	0
	3	High	1/30/2013	269.0	484.2	9.1	100	97	1
	4	Low	3/6/2013	181.8	305.8	12.3	100	94	2
	5	Low	3/27/2013	181.2	331.3	15.0	108	107	15
	6	Low	3/28/2013	179.0	322.8	15.7	102	100	27
2014	1	High	2/12/2014	258.8	628.6	12.3	150	148	31
	2	High	2/13/2014	201.0	484.2	12.9	150	146	37
	3	Low	2/25/2014	111.3	119.6	14.3	150	148	22
	4	Low	2/26/2014	111.6	119.1	14.0	150	143	12

Appendix 2

Survival of telemetered juvenile Chinook salmon (Table 3), and detection efficiency of acoustic receiver gates (Table 4), and associated standard errors. Salmon were released in multiple groups 6.2 km upstream of the initial receiver gate (Gate 1). Boldface indicates detection efficiency below 0.90.