Distribution, population structure and habitat use of the endangered Saint Francis Satyr butterfly, Neonympha mitchellii francisci

American Midland Naturalist, The, April, 2008 by Daniel Kuefler, Nick M. Haddad, Stephen Hall, Brian Hudgens, Becky Bartel, Erich Hoffman

For each of the seven flight periods we monitored during 2002-05, we used publicly available environmental data from the National Climatic Data Center (http:// lwf.ncdc.noaa.gov/oa/ncdc.html) to estimate the beginning of the first flight period. Given the variation in the start of the first flight period, we presumed emergence was controlled not by photoperiod but by temperature. We used measures common to the agriculture literature (i.e., Vittum et al., 1965) that sum heat input into the environment, and started the measure on Mar. 1, when sedge species that are presumed hosts would begin active production. We summed "Growing Degree Days" (GDD), which accounts for minimum temperatures for larval development (50F) and maximum temperatures above which temperature does not increase growth rates (86F). GDD is calculated as the cumulative sum of the average of the maximum and minimum daily temperature minus 50 degrees Fahrenheit (where the maximum cannot be above 86 degrees and the minimum can not be below 50 degrees). When we plotted this value over time for 2003-2005, we found that we could accurately predict the time of emergence, with the first butterfly being observed around 984 GDD, and the peak of the emergence occurring at 1245 GDD (Fig. 2). Given these data, we can predict the first day of emergence and the start of the peak abundance 1 d. The second flight period is then best predicted by days from the first flight period (see above). These results can aid future monitoring efforts by focusing sampling regimes.

[FIGURE 2 OMITTED]

DEMOGRAPHIC VARIATION

In the second adult flight periods of 2002-2005, we marked a total of 1210 individual Neonympha mitchellii francisci in the four large subpopulation sites. Of those marked, we were able to determine the sex and age of 930 individuals. Model comparisons using these individuals indicate that, overall, parameter estimates vary by sex (Table 3). Males consistently have higher detectability. Furthermore, the model including sex classes was the best fit to MRR data for 2003 and very close to the best fit model in other years (within one criterion unit; see Burnham and Anderson, 2002). In the field, we observed behavioral differences whereby males actively search for mates and, hence, are more easily detected. Females, conversely, tend to be more sedentary and are more difficult to find. Males accounted for ~2/3 of all butterflies captured.

In 2003-04 model comparisons indicate that male survivorship tends to be lower than females, however this pattern did not hold for 2005 (Table 3). One explanation for increased male mortality is that males are exposed to more predation threats while searching for females (via spider webs or dragonflies). We found no support for differences in detectability by age, however oldest butterflies consistently show a trend toward lowest survivorship (Table 3).

POPULATION SIZES AND TRENDS

Between 2002 and 2005, the combined population estimates from MRR ranged from 502-1400, with a peak of 1400 in 2004. Subpopulation sizes totaled 49-739 adults per brood. The peak in 2004 may have been caused by weather conditions; however, we do not have a long enough time series of data to understand fully how weather affects population size. In 2005 we generated MRR estimates for both flight periods for A1, D1 and D3 (Fig. 3). We observed an increase in numbers between the first and second broods at sites Al and D3, which is consistent with the general increase between broods shown by modified Pollard-Yates counts (Table 2). One explanation for an increase in numbers between broods is that there is a much longer diapause between the second and first broods and thus greater possible larval mortality. The difference in population sizes between sites may be partially attributed to the size of each subpopulation site. Across sites, MRR estimates were positively correlated ([R.sup.2] = 0.95) with the area of wetland habitat at each site (Fig. 4). Even the four largest subpopulation sites are subject to frequent changes in habitat size, which can result in rapid fluctuations in population size. At site D1, the population responded favorably to the removal of a beaver dam and pond downstream in 2003. Uniformly higher counts after 2003 are attributable to a three-fold increase in wetland area. Differences in population sizes may also be due to habitat quality, as butterfly abundances are highly correlated with the abundance of potential host plants. Because there are only four sites, we cannot currently separate the effects of area and site quality on population size.