All 12 crows had a significant preference for either stick (n=7) or pandanus (n=5) tools (); eight of the crows had an exclusive preference. In spite of the strong preferences, many of the 12 crows also sometimes used the non-preferred tool type when they found these tools on tables. The strong individual specialization that we found was unlikely to be related to factors like site or seasonal effects. This is because we collected data from individual crows at different feeding tables and in different months of the year, over a period of usually several years.
The individual specialization in the use of foraging tools by hunter–gatherers is associated with different foraging niches. We have insufficient direct observations to know if the 12 crows in this study specialized in feeding locations when foraging away from feeding tables. However, circumstantial evidence suggests, it is likely that the individuals which specialized in wide pandanus tools foraged more often in Pandanus trees than did the stick-tool specialists. At our study site, crows generally seem to use each tool type in different vegetation within the forest. Although we sometimes see crows using stick tools in Pandanus trees, they mainly use them in non-Pandanus trees. For example, we commonly find stick tools, and only very rarely wide pandanus tools (J. C. Holzhaider 2006, personal observation), at sites where crows have been extracting cerambycid larvae from dead wood. On the other hand, we frequently observe crows at our study site using wide pandanus tools in Pandanus trees to extract small invertebrates but not outside these trees. That we often find wide pandanus tools discarded in Pandanus trees suggests that crows rarely use them outside these trees.
In spite of sexual size dimorphism in crows and the circumstantial evidence that individual specialization was associated with different foraging niches, there was no significant relationship between the sex of crows and the type of tool that they preferred to use (2×2 contingency table with Fisher's exact probability test: 2-tailed p=0.22, n=12; ). This shows that parallel tool industries in New Caledonian crows, and associated foraging niche partitioning, are not based on a rigid division of labour between the sexes. It is possible that sexual size dimorphism may have a weaker effect on the tool type that a crow prefers to use which is below the limit of our sample size to detect.
Humans have an evolved disposition for basic tool skills (
Lockman 2000), but horizontal (peer-to-peer) social learning clearly plays a major role in determining which tool a hunter–gatherer uses. Social conformity is probably important in bringing about the strict sexual division of labour in subsistence seen in almost all hunter–gatherer societies (
Bird 1999). In contrast, the parallel tool industries we have documented in crows could either be a consequence of genetic differences or different, vertically inherited (parent-to-offspring) social traditions.
Kenward et al. (2006) have demonstrated that social learning can influence the development of stick-tool skills in young crows. An evolved disposition to develop basic stick tool use without social learning has also been demonstrated in juvenile crows, but there is little evidence that such a disposition exists for basic pandanus tool manufacture (
Kenward et al. 2005;
Hunt et al. in press). The social organization and development of crows is likely to promote vertical social learning and minimize opportunities for the horizontal transmission of tool manufacturing techniques. The crows at our study site on Maré spend most of their time in family units consisting of parents and dependent juveniles (J. C. Holzhaider
et al. unpublished data). The juveniles develop tool skills during their first year in close association with parents. Early sensitive periods for acquisition of foraging skills have been documented in other birds (
Tebbich et al. 2001) and would act to further minimize the possibility of horizontal transmission.
This article has been 
cm deep and 2.6