Gut content analysis and a new feeding group classification of termites

Summary 1. Gut content analysis of termites was undertaken using microscopical techniques. The 46 study species covered the entire range of taxonomic and feeding forms within the Order. 2. Inter‐specific gut contents data were analysed using principal components analysis, placing species along a clear humification gradient based on variations in the amount of silica and plant tissue fragments in the gut. 3. Redundancy analysis was used to find morphological correlates of the observed variation in gut contents. A total of 22 morphological characters (out of 45 candidate characters) were correlated significantly with the gut contents. 4. Three of the 22 significantly correlated characters unambiguously defined feeding groups, which were designated groups I to IV in increasing order of humification of the feeding substrate. Group I contains lower termite dead wood and grass‐feeders; group II contains Termitidae with a range of feeding habits including dead wood, grass, leaf litter, and micro‐epiphytes; group III contains Termitidae feeding in the organic rich upper layers of the soil; group IV contains the true soil‐feeders (again all Termitidae), ingesting apparently mineral soil. These groupings were generally supported statistically in a canonical covariance analysis, although group II apparently represents termite species with a rather wide range of feeding habits. 5. Using existing hypotheses of termite phylogenetic relationships, it seems probable that group I feeders are phylogenetically basal, and that the other groupings have arisen independently on a number of occasions. Soil‐feeding (i.e. group III and group IV feeding) may have evolved due to the co‐option of faecal material as a fungal substrate by Macrotermitinae‐like ancestral forms. As a consequence, these forms would have been constrained to build nest structures from soil and would therefore have passed at least some soil through their guts.