Reproductive success of Eastern Bluebirds (Sialia sialis) varies with the timing and severity of drought

Drought affects avian communities in complex ways. We used our own and citizen science-generated reproductive data acquired through The Cornell Lab of Ornithology’s NestWatch Program, combined with drought and vegetation indices obtained from governmental agencies, to determine drought effects on Eastern Bluebird (Sialia sialis L.) reproduction across their North American breeding range for the years 2006–2013. Our results demonstrate that some aspects of bluebird reproductive success vary with the timing and severity of drought. Clutch size was unaffected by any level of drought at the time of clutch initiation or during the 30 to 60 days prior to clutch initiation. Hatching and fledging rates decreased as drought severity increased. Drought conditions occurring at least 30 days prior to the date eggs should have hatched and 60 days prior to the date offspring should have fledged negatively affected reproduction. We also demonstrate the value of datasets generated by citizen scientists in combination with climate data for examining biotic responses at large temporal and spatial scales.


113
Our data collections were carried out in strict accordance with the Ornithological

183
In addition to the various fixed effects, we used random effects to account for sources of 184 spatial variability not captured by the fixed effects. We included county in which nests were 185 located and unique NestWatch user identification numbers nested within county as random 186 effects. These random effects accounted for random variability on a larger spatial scale (i.e., 187 county level) and the variability in skill level among the individuals that reported data to 188 NestWatch. Recognizing that these spatial random variability patterns varied by year, we nested 189 the above random effects within year. This allowed the model to reflect the spatial variability 190 that would not necessarily be the same from one year to the next.  (Table 3). None of the drought 263 status levels (D0-D4) had a significant effect on clutch size when compared with N (normal 264 rainfall, no drought) (Fig. 4).
We also tested whether a more complex random effects structure, with userid nested 281 within county and county nested within year, provided a better fit than a simpler one with just 282 year as a random effect. The LR test clearly favored the simpler random effects structure with 283 just year as a random factor (Χ 2 < 0.001, df = 2, P > 0.99). The results reported here (Table 3) 284 were based on using the simpler structure with year as a random factor.  Because latitude and longitude are typically significant indicators of clutch size, we 305 examined whether or not this trend was present in our data by removing the region variable from 306 the null model. Latitude (Z = 2.66, df = 1, P < 0.01) and longitude (Z = -2.02, df = 1, P < 0.05), 307 significantly affected clutch size when the region variable was not included.
308 Hatching rate 309 The modal number of hatching eggs was 4, followed by 5 and zero (Fig. 3). The 310 distribution was bimodal. The average number hatching was 3.23. The number of hatched young 311 declined from more than 3.3 to fewer than 3.0 hatching per nest with increasing drought status, 312 especially for the two most extreme categories (Fig. 4).

320
We report the model fitting results for hatch ratio for Models 1, 2, and 3 in 346 Hatch ratio was significantly affected by NDVI std., and the odds ratio of an egg hatching 347 increased by a factor of 1.14 with each unit increase.

348
In addition to the drought-status effects, hatch date had significant effects on hatching 349 ratios. For month, the odds of hatching decreased as the months progressed from March to 350 August; for a one-unit increase in month, the odds ratio of an egg hatching decreased by a factor 351 of 0.85. Only the South NADM climatic region was significant as an effect which suggested a 352 decrease in hatching ratios and odds of hatching (0.71) compared to the reference region.
353 Moreover, all interactions of hatch date and region were significant, but only for the Northeast 354 region were the odds of hatching increased (1.17).

355
A seasonal decline in hatching was evident across all regions (Fig. 5). Following a peak 356 in March and April, the mean number of eggs hatching decreased as the season progressed.
357 There was an increase in hatching within the South region during the final month of the 358 breeding season, but this was the case for only a very few observations.

360
The modal number of fledging young was 4, followed by zero and 5 (Fig. 3). Again, the 361 frequency distribution was bimodal, with more zero fledges than zero hatches. The average 362 number fledged was 2.93. The number of fledged young declined from more than 3.0 fledges per 363 nest to 2.4 fledges per nest with increasing drought status, especially for the two most extreme 364 categories (Fig. 4).
370 371 Reproductive data acquired over a 7-year period through NestWatch and from our own 526 study site, and drought data acquired from the NADM, provided the opportunity to examine 527 drought effects across the Eastern Bluebird North American breeding range. While smaller 528 studies conducted solely by experienced researchers yield more accurate data and have value for 529 exploring localized effects, the field time required for a broader study is prohibitive. For 530 example, reproductive data collected from our study site expended over 840 field hours on 531 average for each approximately 173-day field season. Data entry was also time intensive. The 532 more than 24,000 NestWatch records we received were easily imported into a database and then 533 scrubbed of questionable records using simple SQL queries. Consequently, our data set was 534 robust and efficient in terms of sample size and time expenditure, respectively.

536
Our results demonstrated distinct drought impacts on certain aspects of Eastern Bluebird 537 reproduction. We found a clear association between occurrence of extreme or exceptional 538 droughts during critical periods of the Eastern Bluebird nesting cycle and decreases in both 539 hatching and fledging success. We found especially striking that one of our models suggested 540 that drought status one and two months prior to expected hatching and fledging is more 541 important in explaining the effect of drought than during the month in which hatching or 542 fledging takes place.

543
Eastern Bluebirds are a species of least concern [29] and thus not likely to experience 544 widespread changes in population densities due to drought alone. We suggest, however, that 545 drought effects could be significantly more deleterious for species in decline, especially in light 546 of climate change. Studies based on citizen science-generated data for other species combined 547 with standardized climate date could provide additional support for this observation.