• aquaculture;
  • cormorants;
  • distribution;
  • Phalacrocroax auritus;
  • predation;
  • selectivity


Estimating the catfish aquaculture production losses that can be attributed to double-crested cormorants (Phalacrocorax auritus) has proved problematic because knowledge of the distribution of cormorants on catfish aquaculture is lacking. We evaluated use versus availability of various production pond types and landscape-scale factors affecting the distribution of cormorants on channel catfish (Ictalurus punctatus) aquaculture facilities in Mississippi, USA. Cormorant distribution on aquaculture pond types indicated selection against brood-fish ponds, neutral selection on fingerling ponds, and selection for food-fish ponds (n = 29, Chesson's α = 0.19, 0.36, and 0.45, respectively). Modeled and validated correct classification rate (CCR) of general linear mixed models of cormorant occupancy of clusters of catfish ponds indicated seasonality of use and roost distance from aquaculture ponds was predictive (CCR = 81% and 71%, respectively). Modeled and validated ordinal models of levels of use (low, moderate, high) were less predictive (CCR = 67% and 59%, respectively). However, predictability within use levels for the validation data set was mixed, ranging from 0.19 to 0.86. Odds ratios indicate both higher risk of occupancy and levels of use over the period February–April relative to October. Management efforts targeted during this time frame will have the greatest impact in reducing depredation losses. The majority of cormorants occurred on food-fish ponds. Consequently estimates of economic loss should be revisited and refined based on distributional information provided in this study. Published 2012. This article is a U.S. Government work and is in the public domain in the USA.