Large scale disturbances are known to significantly alter aspects of both species diversity and ecosystem function. In the Caribbean, hurricane events are a significant form of disturbance, the effects of which have been shown to alter food web function, especially in the terrestrial environment. Although hurricanes have been studied from a variety of their effects on ecosystems, there is little research on how these storms affect species along elevational gradients. Within terrestrial habitats, ants form the basis of many food webs, being both numerically dominant and functioning in a variety of roles within the food web. On September 20^th, 2017 Hurricane Maria, a category 4 storm, crossed over the island of Puerto Rico, causing significant damage to both human and natural systems. We collected data on ant abundance and composition from 150 samples of leaf litter along a 700 m elevational gradient during June the year of and after the storm event. Ant abundance increased by 400% after the storm with many common ant species seeming to benefit, especially at lower elevations. There were subtle changes in ant richness, with declines generally after the storm, but yet again this response was dependent on elevation. This is one of the first studies to consider how terrestrial insect communities are affected by large hurricane events across elevations, and our results are in contrast to past work showing declines in ant abundance after such storms.

Samples were conducted in the Luquillo Experimental Forest (LEF), a 11,330 ha site located on the eastern side of the main island of Puerto Rico, U.S.A. Average annual rainfall is 3,460 mm (McDowell & Estrada-Pinto, 1988) and monthly temperatures range from 21 to 25°C (Brown et al., 1983). In both 2017 and 2018 sampling took place between June 14 and July 24, which are within the wet season for the island. The elevation transect we used occurred in a mixed forest, typical of Tabonuco forest (Uriarte et al. 2019), composed of both large hardwoods (e.g., Dacryodes excelsa) and gap specialists like Prestoea montana and Cecropia peltata, across fifteen different elevations. Sample sites were from 300 m to 1000 m elevation at 50 m elevation increments (n = 15). At each elevation, 10 plots (total 10 m x 10 m) were laid out in a 2 x 5 grid, with no space between adjacent plots. Within each plot, we sampled litter using a 0.5 m² PVC frame in either the northwest corner in the upper five plots or northeast corner in the lower five plots. As the goal of the sampling was to collect litter invertebrates, the frame was placed to avoid any large obstructions (e.g., boulders). Thus, we produced 150 samples (15 elevations with 10 plots at each). Individual plots were processed using Winkler sacs, which are a passive means to extract invertebrates by using their own behavior (Sabu & Shiju, 2010; Besuchet et al., 1987). Bulk samples of leaf litter from each plot were first vigorously agitated for 1 minute over a 1.25 cm mesh screen to separate invertebrates from large debris. This sifted material was then placed into Winkler sacs that were then hung in place for 48 hrs. At the bottom of the Winkler sac collection bag, we placed a container of 95% ethanol to catch and preserve all invertebrates. Following sorting of invertebrates from debris, ants were then separated into 0.25-dram shell vials for each plot within each elevation. This original sampling design ended in late July 2017, but on September 20^th, 2017 Hurricane Maria passed over the island and did considerable damage to the forest (e.g., removal of the of canopy with stem breakage and uprooting of stems, Uriarte et al. 2019). We decided to resample the plots at the same time during the next year (June 2018) to maintain continuity in seasonality. All methods were the same except samples on plots were collected in the opposite corner to minimize any differences due to our removal of litter in 2017. Given the severity of the storm, no sites in the LEF escaped without damage, and thus we did not have access to any control sites to compare to pre-storm samples. The LEF has also seen significant human disturbance in the form of agriculture (i.e., coffee) and logging, however these practices took place before the purchase of the property in the 1930s (Thompson et al. 2002) and no records of these disturbance exist above ~400 m; it is not clear that any of these disturbances took place along the studied transect.