Behavior and population dynamics of grass carp incrementally stocked for biological control

Grass carp Ctenopharyngodon idella have been stocked throughout the world due to their utility as a biological control. In the United States, the species has been used to successfully control invasive, aquatic weeds such as hydrilla Hydrilla verticillata. Despite the large body of research surrounding the use of grass carp, few studies have demonstrated widely applicable methods for evaluating the success of weed control based on grass carp behavior and population dynamics. Classic methods of biological control using grass carp often rely on a single, large stocking of fish. Few of these studies have demonstrated success in achieving intermediate levels of weed control. Managers would be better equipped to make decisions regarding stocking and maintenance grass carp populations with better information about behavior, survival, and population structure. Improved decision making could result in reduced cost and increased effectiveness of stocking. In order to examine current knowledge gaps for management, I investigated the movements and habitat use of grass carp, post-stocking survival, age-specific survival rates, and population dynamics of grass carp in Lake Gaston, North Carolina and Virginia.

I characterized relationships between grass carp behavior and environmental factors using radio-telemetry. The average rate of movement for grass carp in Lake Gaston was about 137 m/d. Rapid dispersal after stocking was followed by long periods of no movement. However, when time after stocking was held constant in models of behavior, fish moved about 200 m/d more in the second year after stocking than in the first year, and were found closer to shore. On average, grass carp were found about 40 m from shore in about 2.5-3.5 m of water, although mean depth of water at grass carp locations varied seasonally, being shallowest in summer and deepest in winter. Although depth of water at grass carp locations did not vary by stocking location, Grass carp were found closer to shorelines in the upper reservoir than in the lower reservoir. I found significant relationships between grass carp behavior and hydrological processes such as lake elevation and dam releases in the reservoir, as well as with other environmental factors such as water temperature, photoperiod, and weather conditions. The results of this study should be useful in better understanding how behavior can affect management decisions. Specifically, grass carp behavior appears to change with age and environmental conditions within large reservoir systems. Future research should focus more closely on the effects of large-scale flow dynamics on grass carp behavior.

I estimated age-1 survival of grass carp from mark-recapture models designed for radio-tagged animals, and characterized relationships between age-1 survival and factors under the control of management, such as stocking locations and size at stocking. . According to the most-plausible model developed in this study, survival of age-1 grass carp in Lake Gaston varied throughout the year, and the probability of an individual grass carp surviving to the end of its first year (±SE) was 0.57(±0.10). According to the second-most-plausible model developed in this study, grass carp survival varied between stocking locations, and was twice as high in the upper reservoir (0.87±0.09) than in the lower reservoir (0.43±0.11). The differences in survival between stocking locations suggest that the cost-effectiveness of grass carp stocking could be improved by focusing stocking efforts in specific regions of Lake Gaston. Furthermore, none of the models developed in this study that incorporated the effects of size (length and weight) or condition factor accounted for a meaningful amount of the total model weights. These results suggest that costs of grass carp stocking could be reduced in Lake Gaston by using a smaller minimum size (352 mm, TL) than is commonly referred to in the literature (450 mm, TL).

I used grass carp collected by bowfishers in Lake Gaston to characterize the age, growth, and survival of grass carp in the system. From these data, I characterized relationships between fish population dynamics and annual hydrilla coverage. Grass carp collected from Lake Gaston ranged in age 1-16 years. Growth of grass carp in Gaston was described by the von Bertalanffy growth function as Lt = 1297(1-e -0.1352 (t+1.52)). I estimated mortality from the von Bertalanffy growth parameters using methods based on growth, temperature, and age; and with each mortality estimate I estimated population size and standing biomass of grass carp. Use of age-specific mortality rates produced lower estimates of grass carp numbers and standing biomass in Lake Gaston than did the use of a single, instantaneous mortality rate for all ages. I determined that growth of grass carp slowed considerably after the fourth year and that slowed growth, in combination with changes in mortality, resulted in a decrease in the amount of hydrilla controlled by a given cohort after four years in Lake Gaston. This phenomenon resulted in an approximately linear relationship between the biomass of grass carp at year i and hectares of hydrilla at year i+3. Based on this relationship, I predicted that the biomass of grass carp necessary to reduce hydrilla coverage to the target level of 120 ha in Lake Gaston is about 91,184 kg (±38,146 kg) and that the current biomass of grass carp in Lake Gaston is about 108,073 kg (±3,609 kg). I conclude that grass carp biomass is at or near levels that should reduce hydrilla coverage to 120 ha between 2013 and 2018. This research provides an effective means for synthesis of information that is critical to understanding sterile, triploid grass carp populations when assumptions of other methods cannot be met. The results of this study should be of immediate utility to hydrilla management efforts in Lake Gaston and other systems. Furthermore, the age-specific mortality rates developed in this study should be useful as starting values for grass carp management in similar systems.