Fred H. Sklar
In the face of conflicting management scenarios, knowing the relative importance of each component of food availability is a precursor to understanding the effects of specific water management regimes on wading birds. Ongoing modeling efforts in south Florida such as the ATLSS program, integrate such information and provide predictive power for future management decisions. Currently, the biggest information gap limiting the wading bird model of ATLSS is foraging success as a function of prey availability. The South Florida Water Management District (SFWMD) is currently conducting a series of experiments aimed at determining the effects of water management on the use of foraging sites by wading birds. Site-use data are available immediately after each experiment and thus allow for quick analyses and write-up. However, also as part of those experiments, we recorded on film, foraging behavior of wading birds at feeding sites with known prey availabilities.
U.S. Department of Agriculture - Natural Resources Conservation Service (NRCS) Department of the Interior - U.S. Geological Survey Department of Commerce - National Oceanic and Atmospheric Administration (NOAA) Environmental Protection Agency (EPA) Smithsonian Institution - National Museum of Natural History (NMNH)
777 Glades Road
For all experiments, ponds were initially stocked at known fish densities; however, those densities decreased quickly as a result of bird predation. Thus, we treated fish density as a continuous variable, which we monitored regularly. We determined fish densities during at least 4 sampling periods based on 1-m2 throw-trap samples (Kushlan 1974). Sampling ceased in individual ponds if no stocked fish were captured in any of the throw-trap samples. Throw-trap samples were distributed evenly within a pond by dividing each pond into 16 10x8 m plots and conducting one throw-trap sample in each plot during each sampling period. A test of sampling efficiency indicated the numbers of fish captured in throw traps represents an average of 55% of the actual fish density in the experimental ponds. All fish density values reported are the number of fish captured in throw traps uncorrected for sampling efficiency. A linear regression model was fitted to the data from each pond such that the response was the number of fish/m2 transformed as y(ln +1). The predicted values were back-transformed to get estimated densities of fish for each pond each day of the experiment with the constraint that no predicted value could be less than zero. The estimated values were used in subsequent analyses of bird foraging success.
To measure wading bird foraging responses, we filmed feeding flocksfor 5-45 minutes from a vehicle with a Hi-8 mm video camera and 8-120 mm zoom lens. A pilot study indicated that filming from a parked vehicle with cloth-covered windows disturbed birds less than a portable blind. Selection of a flock to film was based on where the largest group of birds were foraging and whether data were lacking for a given treatment combination. Flocks were allowed several minutes to re-settle and resume feeding before filming was initiated. Filming was concluded if flock size changed by more than 25%. Our aim was to get 15 min of film on each bird (with a minimum time limit of 5 min) and a maximum of 15 birds per species per film session. If too few individuals in a film session met that criterion then the minimum time limit was reduced to 2 min. During feeding activities, birds would sometimes travel to the edge of a pond or leave the pond momentarily. Because of the sloped edge of the pond, water depths in those areas were less than treatment levels. Thus, we excluded from time-activity budgets any period where a bird was at the edge or outside of a pond. This criterion resulted in a further shortening of some time-activity budgets.
To eliminate the possibility of constructing time-activity budgets on the same individual more than once in a session, we did not use as focal birds, individuals of the focal species that appeared in view after the first focal bird left the screen. However, before a focal bird left the field of view, any new birds that appeared could have been used as focal birds. A foraging bout ended when a focal bird left the field of view, became obscured in a flock, or the film session ended (usually about 15 minutes). Capture rates were calculated as the number of prey consumed divided by the length of time (min) of the time-activity budget. Birds that never consumed a prey item were excluded from the analysis.
From each time-activity budget, we calculated mean prey-intake rate as the response variable. Descriptive statistics such as the mean and standard deviation are presented for each bird species at different treatment levels. This is the format most useful for incorporating parameter values into the ATLSS wading bird model (W. Wolff, Univ. of Miami, pers. commun.). For species of which we had adequate data to conduct statistical analyses, we determined the relative effects of the treatment variables on the response variables. Tests were conducted using PROC GLM in SAS version 6.12 for Windowsâ with water depth, fish size, and fish species as class variables and fish density as a continuous variable. We specified an initial full model containing main effects and interactions. Non-significant (p > 0.05) interactions indicated that a model was over-specified and contained more terms than necessary (Littel et al. 1991, Freund and Wilson 1993). In that case, we constructed a subsequent set of reduced models containing the main effects and significant interactions only.
777 Glades Road
777 Glades Road
U.S. Department of the Interior, U.S. Geological Survey, Center for
Coastal Geology
Comments and suggestions? Contact: Heather
Henkel - Webmaster
Generated by mp version 2.8.18 on Wed Jan 24 09:40:00 2007