To determine effects of land use and management practices on wild turkey {Meleagris gallapovo) populations, managers need reliable, low-cost methods to estimate absolute and relative densities. Therefore, we evaluated estimation ability of 4 models using wild turkey capture data from 1986-1993 and summer bait site observations from 1990-1993 in Kemper County, Mississippi. Capture-markrecapture models performed worse (i.e., higher coefficients of variance) than capture-mark-resight models. Estimates from 1 resighting model were biased from non-homogenous capture probabilities in most intervals (4 of 6) for both sexes. Estimates from a second resighting model required restriction of marked population to active radio-equipped hens. This model performed well in 1990 and 1991, but likely overestimated population size in 1992 and 1993 when marked/sighted samples were only 9 and 4 individuals. A 3-stage resighting-based sampling design was suggested for turkey census. Minimum counts of individuals, from concomitant bait site monitoring, agreed with estimates for hens, but were less consistent for gobblers. More precise estimates for longer periods are required to adequately assess reliability of minimum counts as population indices. Density estimates of hens were between 0.95 and 3.21 hens/km2 and were generally higher than those from more "traditional" habitats.
