Ramsey et al. raise the general problem of diagnosing behavioral innovation in the field through indirect approaches, emphasizing the high relevance of distribution data in populations. I argue that this constitutes de facto a historical approach at the population level; hence it should be possible to take advantage of complementary historical inferences from molecular phylogeography.

The notion of innovation has an obvious historical dimension, in that it qualifies a scenario of the appearance of a novel behavioral pattern and its further diffusion within populations as being a “cultural innovation” (according to Ramsey et al.'s terminology). A standard approach for inferring or testing historical scenarios of behavioral evolution is phylogenetic inference, for example, adaptation viewed as historical novelty, as previously discussed in this journal (see Andrews et al. Reference Andrews, Gangestad and Matthews2002 in BBS 25(4), and associated comments therein by Blackburn, Deleporte, and Roney & Maestripieri). But such studies are primarily concerned with highly genetically influenced behaviors inherited in different species or higher phylogenetic taxa. Here the focus is clearly on learned behavior and relatively shorter time periods.

Closely related to this concern, molecular phylogeography involves comparing population distributional data with genetic data in order to reconstruct the relatively recent history of spatial distributions and variations in gene flows (e.g., Templeton Reference Templeton1998). This perspective seems clearly more appropriate for Ramsey et al.'s concerns regarding, for example, the presence of a prevalent putative cultural innovation in some, but not all, of the studied populations.

However, the authors frame the question of informative populational data in a somewhat limited way. Their notion of “geographic prevalence” is simply defined in this article in terms of overall presence/absence statistics, namely, “the proportion of populations in which the behavior is recorded” (sect. 4.2.2), without specific reference to spatial patterns. Concerning genetic data, the authors refer to a general notion of occurrence “without concomitant genetic or environmental differences” between populations (sect. 4.1). They apparently mean that the behavioral differences at stake should not be attributable to direct genetic influence, so that the behavior being considered is likely learned. Hence, they implicitly refer to the genetics of behavior rather than to the molecular data as phylogeographic markers.

Compared with simply taking into account the proportion of populations showing the putative innovation (rough “geographic prevalence”), molecular phylogeographic studies may allow for finer considerations. One can map individuals showing the putative innovations on the geographic map of population spatial locations, and also on the cladogram or network of their genetic affinities. Contrasting these two maps reveals their possible consistence and discordances. A conjunction of contiguous spatial distribution of populations containing “innovating” individuals and molecular affinities indicating sustained gene flow between the same populations, possibly down to parental affinities by using appropriate markers, would be suggestive of cultural diffusion of a unique innovation through migration of individuals contributing to both cultural diffusion and gene flow between populations. Otherwise, the strict geographic pattern of diverse rates of “local prevalence” of the behavior inside a series of contiguous populations could also suggest routes of ongoing cultural diffusion, with decreasing prevalence along spatial gradients of progression.

An important and growing concern in phylogeography is that distances in a straight line between populations may not be the best parameter for contrasting spatial distribution with genetic similarity. Hence, the developing concept of “ecological distance” is preferred to rough geometric distances as the crow flies. The general idea once again is to keep the analysis realistically “close to the map” as well as to effective ecological constraints on animal dispersal abilities, notably through possible “landscape corridors” of suitable habitat, rather than reasoning according to an ideally isotropic environment. Such considerations may usefully be included in the previous analysis, with spatial proximity being understood through suitable habitat connections rather than by using a purely geometric basis.

It seems that the possibility of cultural losses is not directly addressed by Ramsey et al., as if innovations could appear and spread, but never vanish, for example, through replacement by other behaviors, or according to environmental changes. A likely explanation is that cultural loss, however plausible, may seem very difficult to assess. Lessons in this respect can also be taken from phylogenetic scenario testing, notably by addressing the question of possible convergent novelties (convergent “autapomorphies” in phylogenetic jargon) versus repeated losses (or “reversals”) of a previously widespread plesiomorphic biological character (i.e., a relatively ancient one). The problem is that resulting distribution patterns in phylogenetic clades, and here in populations, can be exactly the same in cases of multiple convergence or losses. Stated another way, the absence of a behavior can be due either to the fact that it never occurred in the population, or that it already disappeared. But the phylogenetic solution (optimizing scenarios of common descent) is hardly applicable at the population level for learned behavior because of the imperfect fit to genealogy and the importance of spatial connectivity allowing inter-population transfer of cultured individuals. Only complementary information can allow a decision for population innovations. Here, the distributional pattern of a cultural innovation in way of disappearance should generally be both spatially and genetically scattered, and hence not likely the direct result of common cultural inheritance and transfer to adjacent populations. A complementary qualitative criterion is suggested by the authors: that is, a high specificity and complexity of the behavior would suggest that convergence is not likely.

Other notions put forth in Ramsey et al.'s article could also profit from conceptual and methodological reflections in phylogeny and phylogeography, such as the importance of, and the difficulties involved in defining and delineating characters (Pogue & Mickevich Reference Pogue and Mickevich1990), or similar problems with delineating populations (Waples & Gaggiotti Reference Waples and Gaggiotti2006). Otherwise, the question of the fit between genetic relatedness and cultural transmission of innovations could in itself be the object of investigations, as is the case in humans; an example being the study of the westward spread of both agricultural techniques and human genes on the Euro-Asiatic continent (Cavalli-Sforza et al. Reference Cavalli-Sforza, Menozzi and Piazza1994). Time scale apart, such studies should be inspiring to primatologists, with a notable difference being that diagnosing cultural traits seems rather direct in humans.

It could be questioned whether the perspective of adding molecular phylogeographic analysis to investigations about innovation is feasible in practice. This is in effect a highly demanding approach, but there should be increasing opportunities for such studies, notably in primates, with the development of noninvasive population genetic studies for fundamental research and for management and conservation purposes, as well. Recent examples in two emblematic taxa include: the case for gorillas (Bergl & Vigilant Reference Bergl and Vigilant2007; Douadi et al. Reference Douadi, Gatti, Lévréro, Duhamel, Bermejo, Vallet, Ménard and Petit2007) and, as could be expected, orangutans (Goossens et al. Reference Goossens, Chikhi, Jalil, Ancrenaz, Lackman-Ancrenaz, Mohamed, Andau and Bruford2006).