Tuning stochastic matrix models with hydrologic data to predict the population dynamics of a riverine fish

Abstract

We developed stochastic matrix models to evaluate the effects of hydrologicalteration and variable mortality on the population dynamics of a lotic fish in a regulated riversystem. Models were applied to a representative lotic fish species, the flathead catfish(Pylodictis olivaris), for which two populations were examined: a native population from aregulated reach of the Coosa River (Alabama, USA) and an introduced population from anunregulated section of the Ocmulgee River (Georgia, USA). Size-classified matrix models wereconstructed for both populations, and residuals from catch-curve regressions were used asindices of year class strength ( i.e., recruitment). A multiple regression model indicated thatrecruitment of flathead catfish in the Coosa River was positively related to the frequency ofspring pulses between 283 and 566 m3/s. For the Ocmulgee River population, multipleregression models indicated that year class strength was negatively related to mean Marchdischarge and positively related to June low flow. When the Coosa population was modeled toexperience five consecutive years of favorable hydrologic conditions during a 50-yearprojection period, it exhibited a substantial spike in size and increased at an overall 0.2%annual rate. When modeled to experience five years of unfavorable hydrologic conditions, theCoosa population initially exhibited a decrease in size but later stabilized and increased at a 0.4% annual rate following the decline. When the Ocmulgee River population was modeled toexperience five years of favorable conditions, it exhibited a substantial spike in size andincreased at an overall 0.4% annual rate. After the Ocmulgee population experienced five yearsof unfavorable conditions, a sharp decline in population size was predicted. However, thepopulation quickly recovered, with population size increasing at a 0.3% annual rate followingthe decline. In general, stochastic population growth in the Ocmulgee River was more erraticand variable than population growth in the Coosa River. We encourage ecologists to developsimilar models for other lotic species, particularly in regulated river systems. Successfulmanagement of fish populations in regulated systems requires that we are able to predict howhydrology affects recruitment and will ultimately influence the population dynamics of fishes.