Establishing the correct provenience of fossil specimens is necessary to understand the biogeography and evolution of individual species, as well as higher-level taxonomic groups. For catarrhine primates, the Early to Middle Miocene of East Africa was a period of prolific radiation of several clades (e.g., Harrison, 2010; McNulty et al., 2015; Almécija et al., 2021), represented at numerous sites centered on the volcanic terrains of eastern Uganda and from Lake Victoria to Lake Turkana in the rift valley of Kenya. Given the numbers of specimens, species, and sites from these regions, establishing the correct provenience of all specimens is essential to accurately document and interpret the biogeography and paleoecology of these radiations.

Suspicions have circulated for decades regarding the provenience of certain catarrhine specimens collected from Early and Middle Miocene East African sites during the first half of the 20th century (e.g., Andrews and Molleson, 1979; Pickford, 1985; Kelley, 1986; Kelley et al., 2002; Hammond et al., 2019). During this early period of site discovery and initial collecting, multiple sites were often visited by the same individuals or teams during the same season (see Table 1). Field practices and protocols also varied and were sometimes lax. For example, specimens were not always given field numbers, or, if they were, the field numbers were not always written on the specimens. These lapses in documentation were exacerbated by the long delay between these early collecting efforts and the first attempts to catalog and accession specimens into permanent collections. The first museum accession catalog was for the Coryndon Museum, Nairobi (CMH), compiled years after the collection period for many specimens (Pickford, 1985). This was also the case for the current catalog of the National Museums of Kenya (NMK; formerly the Kenya National Museums, KNM), in which specimens are accessioned as: KNM-[locality acronym][specimen number] (e.g., KNM-RU 1676). Coryndon Museum specimens were eventually reaccessioned into the KNM and a complete list matching CMH specimen numbers with the later KNM numbers is given in Andrews (1978, Appendix III). However, although Andrews (1978) identifies CMH numbers as field numbers, they are in fact the accession numbers for the Coryndon Museum. Examination of Le Gros Clark and Leakey (1951) reveals that some specimens first accessioned at the CMH have actual field numbers, while other specimens are identified only by their CMH numbers. The field numbers are generally a number followed by the year of collection (e.g., 683 1947), although other forms were used by the various collectors. Thus, even the current NMK museum labels cannot be considered entirely reliable for earlier-collected specimens without field numbers, or that show field numbers not written directly on the specimens. Questions concerning provenience are known to extend also to Plio-Pleistocene sites in Kenya before the advent of GPS technology and strict field-collection protocols (e.g., Hammond et al., 2021). As taxonomy and spatiotemporal distribution are the most basic data of paleontology, concerns about specimen proveniences impact all subsequent specimen-based research, including paleoecology and phylogenetics.

These issues of provenience are set against a historical background of differing and frequently changing perceptions of species representation among various sites (e.g., Le Gros Clark and Leakey, 1951; Leakey, 1967; Andrews, 1978; Bosler, 1981; Kelley, 1986; Harrison, 2002, 2010, 2013; Pickford et al., 2009, 2010; Grossman et al., 2014; Cote et al., 2016), involving not only disagreements about the taxonomic identification of individual specimens but also the recognition of new taxa within what had formerly been viewed as morphologically homogeneous samples (e.g., Harrison, 1986; Ward et al., 1999; Pickford et al., 2010; McNulty et al., 2015). Irrespective of these disagreements, as decades of collecting and increased sampling have led to more refined taxonomies, it has gradually become apparent that many catarrhine species from this period are—to a greater or lesser extent—restricted in their distributions, some to single sites while others are somewhat more widespread. These restrictions reflect both the geographic and temporal distribution of sites, as well as habitat variation. Although there continue to be disagreements in the acceptance of taxa as well as species’ site representation (e.g., Pickford et al., 2009, 2010; cf. Harrison, 1986, 2010; McNulty et al., 2015), most taxonomies reveal some degree of association of species among sites, particularly related to the temporal position. Species abundances can also vary substantially, with apparently rare or uncommon representation of taxa at particular sites being especially problematic. Questions remain as to whether these represent genuine cases of rarity, taxonomic misassignment of specimens, or errors in provenience.

Andrews and Molleson (1979) were the first to test the provenience of one of the questionable specimens alluded to above, the type specimen of Equatorius ‘Sivapithecus’ africanus (NHM M16649), described as being from Rusinga by Le Gros Clark and Leakey (1951) but with dental morphology more similar to specimens from Maboko. Using x-ray fluorescence spectroscopy (XRF) to examine the elemental signatures of specimens, Andrews and Molleson demonstrated within a reasonable probability that M16649 was in fact from Maboko. Further suspected cases of incorrect provenience also concern Rusinga vs. Maboko, both sites being islands on the margins of Lake Victoria in the Kisingiri region. Others concern Fort Ternan vs. Songhor (Fig. 1), proximate localities in the Tinderet area (Pickford, 1985; Kelley, 1986).

Until now, no other questionable specimens from these areas have been examined to determine if their elemental signatures are compatible with their reported provenience. While investigating the provenience of catarrhine fossils purportedly from Losodok using portable XRF spectroscopy (pXRF; Hammond et al., 2019), we collected data from some of the specimens questioned by Pickford (1985) and Kelley (1986), as well as several other specimens detailed below. This study was not intended to be a systematic examination of all specimens that might be regarded as having questionable provenience. Rather, the goal was to test the efficacy of pXRF for determining the likely provenience of a few select specimens that have implications for the distributions of certain species and the taxonomic composition of particular sites, or that might add to morphological information of certain species through more reliable association of specimens with taxa known to occur at particular sites.