Reduced body sizes in climate-impacted Borneo moth assemblages are primarily explained by range shifts

Both community composition changes due to species redistribution and within-species size shifts may alter body-size structures under climate warming. Here we assess the relative contribution of these processes in community-level body-size changes in tropical moth assemblages that moved uphill during a period of warming. Based on resurvey data for seven assemblages of geometrid moths (>8000 individuals) on Mt. Kinabalu, Borneo, in 1965 and 2007, we show significant wing-length reduction (mean shrinkage of 1.3% per species). Range shifts explain most size restructuring, due to uphill shifts of relatively small species, especially at high elevations. Overall, mean forewing length shrank by ca. 5%, much of which is accounted for by species range boundary shifts (3.9%), followed by within-boundary distribution changes (0.5%), and within-species size shrinkage (0.6%). We conclude that the effects of range shifting predominate, but considering species physiological responses is also important for understanding community size reorganization under climate warming.


32
2. This species is the largest larentiine in Borneo and is only known from Kinabalu, falling into the highest (summit) 33 association. Its sister species occurs at lower altitudes more widely in Borneo and is distinctly smaller.

34
3. This species was also abundant in 1965, but the specimens could not be located when the forewing measuring was 35 undertaken.

36
4. This species, with titanis and sayata, is characteristic of the summit zone on Kinabalu and is only known from there.

37
It too has a smaller congener at lower altitudes that is not endemic to Borneo. 38 5. This is the largest species in a genus with 9 species in Borneo, 5 endemic, 4 only known from Kinabalu, and sayata is 39 the highest of these.

41
Supplementary  When forewing length of a species at a site is not available (e.g. damaged or missing 206 specimens; 417 out of the 1079 species-site combinations), we conducted a two-step 207 estimation for missing data based on the 662 species-site combinations we had measured. 208 First, the site-specific forewing length of a particular species in one year, if available, was 209 used to estimate the missing data of that species in the other year. Second, if not available, the 210 forewing length of that species from the nearest elevation site(s) in the same year was used to 211 estimate the missing data. An interpolation was applied if two nearest elevation sites were 212 both available (e.g. adjacent higher and lower sites). Our estimation method increased the 213 measured/collected specimens from 662 to 893 out of 1079 species-site combinations, 214 improving the ratio of measured/collected specimens (females) at each site from 39.6 -215 90.0% (raw data) to 69.8 -100% (Supplementary Table 4). 216 217 Robustness of results applying missing body size data estimation 218 To ensure that the results and conclusions drawn in this paper are robust under our missing 219 data estimation approaches, we conducted the same set of analyses on assemblage body size 220 structure (i.e. changes in average and variation in species body size, and contributions by 221 species range shift and species body size change), using 1) raw and 2) nearest-site estimation 222 only data sets (Supplementary Figure 14-17). We found qualitatively consistent patterns, 223 regardless of the type of data set used, that average body size of assemblages was reduced at 224 higher elevation, there was increasing species body size variation at most sites, and relatively 225 stronger contribution of range shifting, in comparison to size shrinkage, in determining 226 assemblage body size structure. Given that our results were robust to the different ways of 227 considering missing data, we only report results including missing data estimation in the main 228 text. 229 230 that the geometrid species characteristic of the top three sites, together with two noctuoids, 288 are the largest in their genera in Borneo (see also footnotes to Supplementary Table 3) and all  289 are only found on the mountain. A more analytical approach is being considered for the 290 fauna, particularly the Larentiinae. A more general biogeographic context for montane 291 Lepidoptera in the Indo-Australian tropics was provided by Holloway (1986) 2 , noting the 292 unique features of Borneo, though this was before a taxonomic revision 3 when these features 293 were found to be more marked. 294 295