Gut content analysis and a new feeding group classification of termites

<jats:title>Summary</jats:title><jats:p>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.</jats:p><jats:p>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.</jats:p><jats:p>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.</jats:p><jats:p>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 <jats:italic>true</jats:italic> 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.</jats:p><jats:p>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.</jats:p>