Cold tolerance of red drum (Sciaenops ocellatus) and thermal-refuge technology to protect this species from cold-kill in aquaculture ponds

Abstract

The need to protect red drum in aquaculture ponds from cold-kill led to the development of thermalrefuge technology for overwintering these fish. Successive versions of an experimental thermal refuge were installed and operated in two adjacent red drum culture ponds at Redfish Unlimited, Palacios, Texas, during the winters of 1991-92 and 1992-93. Although red drum in the ponds at Palacios were not threatened by cold during either of these mild winters, the thermodynamic efficiency of the two versions of the refuge could be tested and compared, both with each other and with the prototype refuge used in the winter of 1990-91. The 1992-93 refuge design, which featured an inflated, dome-like cover, was the most effective in terms of cost, ease of maintenance, and resistance to heat loss. A mathematical model, based on Newton's law of cooling, was used to simulate refuge thermodynamics. The best-fitting model had exponential rate constants ranging from 0.01 1 to 0.026 min-' for the 1991-92 version and 0.004 to 0.006 min-' for the 1992-93 version (r > 0.99). A second phase of the research focused on cold tolerance of red drum in ponds at Palacios, from 6 February to 2 April 1993. Values of the 24-h lower lethal temperature of fish sampled from refuge-equipped and non-refuge-equipped ponds were estimated via lethal-cold bioassays. Regression of probit-transformed percent survival at 24 h yielded estimates of mean lower lethal temperature which rose from 0.7 IC on 6 February to 6.6 IC on 2 April. Within dates lower lethal temperatures of fish taken from refuge-equipped and non-refugeequipped ponds did not differ; however, there was a significant difference among sample dates. Values of estimated acclimation temperature for these red drum ranged from 13.5 IC on 6 February to 17.7 IC on 2 April. Acclimation temperatures were inferred by exponentially filtering the pondtemperature time series so that the filtered values had maximum linear correlation (r = 0.93) with the lower-lethal-temperature series; the requisite filter had an exponential rate constant of 0.044 day-1.