Observations were made of foraging frequency and food selection by species of Macrotermitinae in a savanna habitat in the distal reaches of the Okavango Delta, Botswana. A removal experiment in the field was also conducted. Observable foraging showed seasonal peaks in frequency, related to temperature in Macrotermes michaelseni and to moisture in sympatric soil nesting Termitidae. Foraging was mostly on woody litter of the dominant canopy trees and mammalian dung. There was considerable overlap in the selection of food items between species but M. michaelseni were seen more frequently on items of woody litter greater than 6 mm in diameter. In Petri-dish trials Microtermes spp. workers were aggressive towards M. michaelseni individuals and in the field they foraged more frequently and effectively when M. michaelseni numbers were experimentally reduced. Macrotermes michaelseni appeared to avoid competition with other fungus-growing species by spatial separation within the habitat and some limited separation in food selection, but intense competition is probably rare as space and food resources may rarely be limiting.

Damage to the antennae was equally common in kings and queens, and was randomly distributed. Damage to the legs was commoner in queens than in kings, and was not randomly distributed. Damage to many or all the legs of one individual may be symptomatic of degeneration due to age.

The queen's fresh weight can be deduced from her external dimensions, or less accurately from the size of the royal cell in the case of single queens. Cell shapes and methods of enlarging the cell are described.

The termite Macrotermes michaelseni (Sjöstedt) builds conical earth mounds with no external openings. The structure of the nests is described. The fungus combs and the nursery are in separate parts of a single hive chamber, and thus occupy the same air space. A closed system of inter-connected passages in the mound facilitates loss of metabolic heat to the atmosphere.

Major workers are the dominant caste in foraging parties of Macrotermes michaelseni; they also have the largest sternal gland, with the highest potential trail-laying activity. The termite utilizes pheromonal trails along open-air foraging routes, surface galleries and in the subterranean gallery network. Thus, surface galleries appear to serve primarily as protective structures to foraging traffic, while trails lead foragers to food sites. Well-established trails on paper lost activity within 2–3 hr under open-air field conditions. In a choice situation, a trail leading to food source was 30–40 times stronger than that directed to an empty food chamber, suggesting that food influences trail-laying and recruitment behaviour of the termite.

At intervals of 2 months over 1 year, 64 royal cells were removed from mounds of Macrotermes michaelseni to find out whether a new royal pair was produced. In five mounds replacement reproductives were found. Young reproductives from incipient laboratory colonies were never accepted as replacements if they were introduced after the removal of the primary reproductives. One out of 10 royal pairs was accepted if the royal cell was introduced immediately after removal of the original one, and three out of 10 survived if the royal cell was introduced 1 week after the removal of the original one.

Replacement reproductives were only produced when nymphs or alates were in the nest at the time the primary reproductives were removed. It was not possible to find out at what time after the removal a new royal pair was accepted but measurements of the replacement queens showed that they can become fully physogastric within 7 months.