Cooperatively breeding species exhibit a range of social behaviors associated with different costs and benefits to group living, often in association with different environmental conditions. For example, recent phylogenetic studies have collectively shown that the evolution and distribution of cooperative breeding behavior are related to the environment. However, little is known about how environmental variation may drive differences in social systems across populations within species, and how the relationship between environmental conditions and sociality may differ across species. Here, we examine variation in social group size along a steep environmental gradient for two congeneric cooperatively breeding species of fairywrens (Maluridae) and show that they exhibit opposing ecogeographic patterns in social structure. Purple-backed Fairywrens, a species in which helpers increase group productivity, have larger groups in hot, dry environments and smaller groups in cool, wet environments. In contrast, Superb Fairywrens, a species with helpers that do not increase group productivity despite the presence of alloparental care, exhibit the opposite trend. We suggest differences in the costs and benefits of sociality contribute to these opposing ecogeographic patterns, demonstrating that comparisons of intraspecific patterns of social variation across species can provide insight into how ecology shapes social systems.

To aquire group size data, we performed two observational sampling transects along a rainfall and temperature gradient in south-eastern Australia, one over ten days in December 2018 (late breeding season) and the second over seven days in August 2019 (early breeding season). We visited local parks, conservation areas, and national parks, starting in coastal Victoria and extending north to inland New South Wales, passively observing fairywren social groups and recording their composition (group size and sex of individuals) at populations along the transect. Groups were only included in analyses (and this dataset) if the entire group was identified (though description of incomplete groups can be found on eBird via checklist identifies included in this dataset).

Climate variables were calculated from two gridded datasets available through the Australian Bureau of Meteorology. Rainfall grids have a resolution of 0.05 degrees (~5 km; Evans et al. 2020) and maximum temperature grids have a resolution of 0.025 degrees (~2.5 km; Australian Bureau of Meteorology 2021. For the long-term climate metrics, we used a window of grids from March 1988–August 2019, spanning 30 years of data. Within this time-window, a single year’s annual data was considered to span March–February, such that each year begins in the austral fall and ends with the austral summer. For each checklist location, we generated one average annual value and two measures of variation in both rainfall and temperature from the long-term climate grids (sensu Jetz & Rubenstein 2011). Average annual rainfall was calculated by summing rainfall totals for each month in a year, then averaging rainfall across years and log transforming the result. Rainfall variation within years was calculated by summing non-transformed seasonal rainfall totals (fall: March–May, winter: June–August, spring: September–November, summer: December–February) and calculating standard deviation across seasons within each year. We then averaged standard deviations across all years, providing an estimate of the seasonality of rainfall. Rainfall variation across years was determined by calculating seasonal standard deviation of the non-transformed average seasonal rainfall across all years, then averaging across all seasons. Temperature metrics were determined similarly. Average monthly maximum temperature was calculated by averaging the maximum monthly temperatures within a year, then averaging across years. Temperature variation within years was determined by calculating the standard deviation of average seasonal maximum temperatures within each year, then averaging the standard deviations across all years. Temperature variation across years was determined by calculating the seasonal standard deviation of average seasonal maximum temperatures across all years, then averaging across all seasons.

For the short-term climate metrics, we used gridded data from the calendar year preceding each breeding season we observed groups, spanning September of the prior year to August of the same year (i.e., September 2017–August 2018 for points collected in 2018 and September 2018–August 2019 for points collected in 2019).  From these data, we generated two rainfall values, total annual rainfall (the sum of monthly rainfall log-transformed) and monthly variation in rainfall (the standard deviation of non-transformed monthly rainfall), and two temperature values, average monthly maximum temperature and monthly variation in maximum temperature (the standard deviation of maximum monthly temperature).

Both long- and short-term climate variables were extracted from each raster for the corresponding latitude and longitude where each social group was observed. Because the limate metrics are closely related we performed principle components analysis on the 30-year and the one-year time windows separately. These metrics with reduced dimensions were used in subsequent analyses. However, all of the climate metrics as well as the principle component values are presented in the dataset.

Citations:

• Australian Bureau of Meteorology. 2021. Long range weather and climate. Available at: [http://www.bom.gov.au/climate/]. Last accessed 7 Dec 2021.
• Evans A, Jones D, Smalley R, Lellyett S. 2020. An enhanced gridded rainfall analysis scheme for Australia. Bureau of Meteorology Research Report. No. 41.
• Jetz W, Rubenstein DR. 2011. Environmental Uncertainty and the Global Biogeography of Cooperative Breeding in Birds. Current Biology. 21, 72–78.