Wood-boring beetle larvae act as ecosystem engineers by creating stem cavities that are used secondarily as nests by many arboreal ant species. Understanding the heterogeneity and distribution of available cavities and their use by ants is therefore key to understanding arboreal ant community assembly and diversity. Our goals were to quantify the abundance and diversity of beetle-produced cavity resources in a tropical canopy, reveal how ants use these resources, and determine which characteristics of the cavity resource contribute to ant use. We dissected branches from six common tree species in the Brazilian Cerrado savanna, measuring cavity characteristics and identifying the occupants. We sampled 2,310 individual cavities, 576 of which were used as nests by 25 arboreal ant species. We found significant differences among tree species in the proportion of stem length bored by beetles, the number of cavities per stem length, average entrance hole size, and the distribution of cavity volumes. The likelihood that a cavity was occupied was greater for cavities with larger entrance hole sizes and larger volumes. In particular, there was a strong positive correlation between mean head diameters of ant species and the mean entrance hole diameter of the cavities occupied by those ant species. Wood-boring beetles contribute to the structuring of the Cerrado ant community by differentially attacking the available tree species. In so doing, the beetles provide a wide range of entrance hole sizes which ant species partition based on their body size, and large volume cavities that ants appear to prefer.

This study focused on six Cerrado tree species which are characteristic of Cerrado vegetation in general, and are abundant in our study site: Caryocar brasiliense (Caryocaraceae), Stryphnodendron polyphyllum (Fabaceae), Qualea grandiflora (Vochysiaceae), Kielmeyera coriacea (Calophyllaceae), Machaerium opacum (Fabaceae), and Tachigali aurea (formerly Sclerolobium aureum, Fabaceae). In June-August 2012 (“small stem” dataset) and June-August 2013 (“large stem” dataset), we sampled branches from 90 individual trees occurring in mid to low-canopy density savanna vegetation (cerrado sensu stricto). All trees were growing at least 15 m from conspecifics and except for two trees of the large data set, were 40 m from any trail. All were sufficiently tall, i.e., 6-10 m in height, to comprise part of the tree canopy. Together, trees sampled for the small and large data sets were distributed across approximately 1 km² and 2.1 km², respectively.

For the large stem dataset, we removed a single 10-cm diameter branch with all attached biomass from 12 individuals of each tree species. To minimize impact on trees, only large trees with three or more 10-cm diameter branches were sampled. Because of the growth form and availability of Stryphnodendron polyphyllum, branches sampled from this species ranged from 7.5-10 cm at the base (samples from only two individuals were under 9.0 cm). From an additional six individuals for each tree species (the small stem dataset), we sampled between six and fifteen 3-cm diameter branches, with more branches sampled from larger trees. Bole diameters were measured 10 cm above the soil level using digital calipers. Trees often branch below breast height in Cerrado, preventing the more common measurement of diameter at breast height. The 10-cm diameter branches and all attached stems greater than 3 cm diameter were cut using a chainsaw at 10-cm intervals and split lengthwise to expose any cavities. We measured both the xylem diameter at the basal end of each 10-cm length and the total length of cavities encountered. When present, we measured entrance-hole diameter, taking multiple measurements for non-circular shapes and averaging for analysis. Cavity diameter was measured at 10-cm intervals along the length of the cavity and then averaged for analyses. Cavity inhabitants were identified in the field, taking voucher specimens to the lab for confirmation when necessary. Because of the scarcity of dead stems of larger diameter, all stems from this dataset were living wood.

For the small stem dataset, stems (3 cm diameter and under) were measured for total length and basal xylem diameter, and then split lengthwise to expose cavities. Measurements were taken on all cavities encountered, including occupant species, cavity length, minimum and maximum diameter, and diameter of any entrance holes. We also recorded whether the cavities in these small stems were located in live or dead wood.

Voucher specimens of ants found in stems were collected for identification. Ant specimens were identified to species or morphospecies using our existing expertise in the system and vouchers deposited at the Zoological Collection at the Universidade Federal de Uberlândia, in Brazil. Vouchers were deposited in this same collection. Measurements of ant head width were made with a stereoscopic microscope with a zoom lens and a built-in micrometer, measured to the nearest tenth of a millimeter. Sample size ranged from one individual per species for rare species up to 85 individuals from as many as 32 different trees, with never more than five individuals per tree.

There are 4 data sets.

1. Ant body size: average body size of ant species encountered in cavities.

2. Characteristics and inhabitants of cavities found in small stems (< 3 cm diameter) for the 6 tree species. Each data line is a different cavity.

3. Characteristics and inhabitants of cavities found in small stems (> 3 cm diameter) for the 6 tree species. Each data line is a different cavity.

4. Summary of measures by tree for both large and small stem data sets.

All measurements are in cm except for ant body size measurements, which are in mm.

When data are "missing" for ants and other invertebrates, no organisms were found in those cavities.

The diameter measure for a cavity is the diameter of the cavity measured from one inside surface to the opposite inside surface. A wood diameter measure for a cavity is the diameter of the cavity plus the wood on the outside, i.e., the full diameter of the stem at the point of measurement.