PARENT SESSION
Oral Session #71: Animal Population Ecology. Presiding: J. Dooley.
Thursday, August 9, 2001. 1:00 PM to 4:15 PM. Hall of Ideas J.
Cyclicity in a non-boreal galliform? Time-series analyses of Northern Bobwhite abundance.
Thogmartin, Wayne1, Woolf, Alan1, 1
ABSTRACT- Population cycling in boreal galliforms such as Ruffed Grouse (Bonasa umbellus) is well- established, whereas in non-boreal birds it is less common. Previous research suggested abundance of some northerly populations of Northern Bobwhite (Colinus virginianus) cycled with an 8-yr period, but this phenomenon is not universal within the species. For a population to cycle repeatedly, abundance in any one year must be correlated to abundance in some previous year. Few published data assess extent (in years) and strength of this correlative effect in bobwhite. We re-examined cyclicity in quail populations using time-series techniques. We studied temporal patterns of variability in bobwhite abundance in Illinois with 33 time series derived from North American Breeding Bird Survey call counts collected between 1973 and 1997 (n = 25 yrs). Lag plots and plots of the autocorrelation and partial autocorrelation functions identified the cyclic component, whereas autoregressive integrated moving average (ARIMA) models described the density dependent structure. In general, we found bobwhite populations showed a weak (n = 20) or no tendency (n = 13) to regular fluctuations. Cycling populations possessed a period of either 7 years in the southern part or 17 years in the central part of the state. This difference in period was reflected in the temporal structuring of the time series as the ARIMA models for the 17-yr group were more complex, possessing moving average and second order autoregressive coefficents. We posit environmental forcing as the mechanism entraining bobwhite population dynamics in Illinois. Severe historical winter weather in the central portion of the species range led to a reduction in the correlation of abundance between consecutive years. First order autoregression coefficient was 0.52 (SD = 0.47) for the 17-yr group versus 0.78 ± 0.14 for the 7-yr group. This reduction in correlation stretched the frequency within the central time series. However, it is still not clear why 13 populations did not cycle. Lag correlation and ARIMA coefficients were virtually identical between the 7-yr and non-cycling groups. The underlying cycle (signal) may have been obscured by great variation in abundance (noise), preventing detection of periodicity.
KEY WORDS: ARIMA, density dependence, periodicity, quail