Brush manufacture occurred prior to contact with the termite nest in 96 per cent of observations (52/54). In the two other observations, an adult female and subadult male used the blunt end of the stem to contact the termite exit holes, possibly to test whether the hole was open. Both of these individuals subsequently manufactured a brush and then began termite fishing. We observed five different types of modification to herb stems during brush manufacture: fray end by pulling through teeth (75% of modifications); split probe lengthwise (16%); separate fibres by biting (3%); clip probe length (2%); and remove extraneous vegetation (3%). Brush-tipped tool manufacture consisted of 3.2±1.7 modifications (range: 1–10; total: 175 modifications) and lasted 7.9±6.4
s (range: 2.0–29.6; n
After 93 per cent of brush manufacture observations (50/54, involving AF/SAF=11, AM/SAM=12, JUV=4), chimpanzees were observed to pull the brush tip through a partially closed fist to compact (or straighten) brush fibres for insertion into the nest (for more detailed information about ‘brush straightening’ within termite-fishing sequences, see Sanz & Morgan in press
). If the fishing probe was not successfully inserted, the chimpanzees often repeated the straightening action until the brush fibres were appropriately arranged for insertion into the nest. Chimpanzees also took measures to maintain the brush during termite fishing. In contrast to brush manufacture, maintenance consisted of fewer modifications (1.3±0.5; range: 1–2; total: 14) that occupied less time (2.2±1.9
s, range: 0.6–6.5; n
=11) and were aimed at refining tool dimensions or repairing the brush tip.
Chimpanzees consistently not only used the modified end of the tool for termite fishing, but we also observed four chimpanzees reverse the tool orientation and use the blunt end for a different function. In the midst of termite fishing, these individuals (AF, AM, SAM, JUV) were each observed to change the tool orientation and then use the unmodified end temporarily as a perforating tool to clear debris from a fishing hole. After the obstruction had been cleared, the chimpanzee immediately reoriented the tool to resume fishing with the brush tip.
Termite-fishing trials conducted by local guides showed that brush-tipped fishing probes were significantly more effective in gathering termites (Χ162=83.60, p<0.001) than unmodified probes. Seventy-six per cent of fishing attempts with a brush were successful in gathering termites, compared with only 18 per cent success during attempts without a brush. Humans gathered an average of 4.90±5.92 termites per attempt with a brush-tipped tool (n=420 attempts), compared with only 0.27±0.71 termites per attempt with an unmodified tool (n=420 attempts).