EVOLUTIONARY BIOSTRATIGRAPHY

Rep learned and utilized evolutionary thinking in the context of the Modern Synthesis, and he always conceptualized his data in an evolutionary framework. Although he was most concerned with the application of fossils to geological problems, he thought about fossils as the remains of members of once-extant biological populations. He adopted a Simpsonian emphasis on population-level variation as a fundamental criterion for taxonomic allocation (thus exempting himself from one of Simpson's claims that biostratigraphers tended towards typological conceptualizations; Simpson 1953, p. 340). Rep struggled to strike a balance between his perspectives on the biological and geological relevance of fossils, and to integrate those perspectives into a holistic approach to biostratigraphy. The methods he used to achieve that integration changed through time, and had a powerful influence on his view of the classification, taxonomy, paleobiogeography, and stratigraphic utility of arvicoline rodents.

Polyphyly and Paraphyly as Organizing Principles

Rep was a strong proponent of classifications and taxonomies that reflected evolutionary history. He published three classifications of arvicolines, including a global classification (Repenning et al. 1990), and classifications of the "Pitymys" group (Repenning 1983a) and of Allophaiomys and its purported descendents (Repenning 1992). In subsequent papers he updated those classifications in light of new discoveries and taxonomic arrangements (e.g., Fejfar and Repenning 1998; Repenning 2003). Although published relatively late in his career, his classifications, and his taxonomic practices, were at odds with newly emerging contemporaneous philosophies that specifically sought classifications and taxonomies based upon nested sets of monophyletic taxa (e.g., Rowe 1987, Gauthier et al. 1988; de Queiroz and Gauthier 1990, 1992; de Queiroz 1992).

The ascendancy of monophyly as an organizing principle was anathema to Rep, and he eschewed the explicit recognition of monophyly, once remarking that monophyly (at least in taxa higher than the species) "is a myth" (Repenning, letter to Christopher Bell, 6 July, 1993, p. 3). Instead, he accepted the idea shared by many paleontologists of the Modern Synthesis that macroevolutionary dynamics were best exemplified by evolutionary hypotheses that postulated what we today call paraphyly. He recognized five independent originations of what he called a 'microtine grade' of dental specialization (Repenning 1987; rendered as 'arvicolid grade' by Repenning et al. 1990), and allocated the taxa from each independent lineage to a separate subfamily (Lemminae, Prometheomyinae, Arvicolinae, Ondatrinae, Dicrostonychinae). He considered the term 'microtine' to be an adjective with no taxonomic status (Repenning 1987, p. 238) and was perpetually frustrated by the variable taxonomic treatments (reviewed by Bell et al. 2004b) that used 'microtine,' 'arvicoline,' and their cognates as formal or informal taxonomic designations (alas, as we do here!) to unite what he conceptualized as five independent lineages. That frustration was especially acute when there was an implication of monophyly. He explicitly emphasized paraphyly of many recognized 'microtine grade' genera, and was a strong proponent of a polyphyletic origin of 'microtine grade' taxa as a whole (Repenning 1987; Repenning et al. 1990), and of some of the groups that achieved that 'grade' of dental specialization (Repenning 1992).

His efforts to trace the detailed evolutionary history of the members of those five lineages were strongly influenced by two related concepts. The first was that evolution was gradational and anagenetic at both higher and lower taxonomic levels within each lineage (e.g., Repenning 1992, p. 17). The second was that the fossil record of arvicolines can and does adequately preserve population-level evolutionary signals, as long as sample sizes are sufficiently high. Discrete time-slices of populations preserved in fossil deposits could, therefore, reflect various stages in the evolution of higher taxa that could be collated into a fairly complete history. A necessary complication was that a diagnostic feature of a given taxon (at any hierarchical level) must once have been present only as a morphological variant within an ancestral population (Simpson 1953). Such a feature would, with an adequate fossil record, be traceable as it increased in abundance within and between populations. Evaluation of population-level variation thus became a central focus for Rep and a key consideration for his taxonomic treatment of fossils.

Population-level Variation, Transitional Forms, and Taxonomic Practice

As a consequence of his views on evolutionary dynamics and classification, Rep came to recognize population-level variation as an essential consideration for both the identification of fossil specimens and the placement of those fossils in a taxonomic hierarchy. His thoughts on the issues developed over time and are reflected by shifts in his perspective on the importance of sample size for reliable determination of taxonomic affinity. As late as 1988 (Repenning and Grady 1988, p. 10), he was comfortable naming the new species Phenacomys brachyodus based on a relatively limited sample of six specimens, only two of which were lower first molars -- the arvicoline tooth accepted by Rep and other paleontologists as being the most diagnostic tooth in the head of arvicolines. By the early 1990s his standards became more rigorous, and he was adamant in his view that isolated arvicoline teeth could not reliably be identified. Instead, he insisted that a minimum sample size of 25 specimens was required to establish definitively the presence of a taxon in a fossil assemblage (Repenning 1992). It is not clear why he settled on 25 as the required number, but he emphasized the importance of that sample size to his colleagues and pushed for a wider adoption among paleontologists documenting fossils of arvicolines (e.g., Repenning, letter to James Mead, 11 January, 1993, pp. 2-3; letter to Anthony Barnosky, 9 March, 1993, p. 3).

In practice, his perspective resulted in complicated conventions for distinguishing between what he considered 'primitive,' 'typical,' and 'advanced' populations of particular genera or species; it also occasionally yielded awkward taxonomic treatments of what he considered to be transitional forms. His views on these issues were most thoroughly and explicitly discussed in his extensive analysis of the evolution of Allophaiomys and its many purported descendents (Repenning 1992). That discussion makes clear that an appreciation of his conceptual framework is essential for understanding the names he applied to fossils.

Rep utilized hyphenated taxonomies beginning in 1987, but without extensive explanation. That convention expressed his interpretation of the specimens in question as intermediate or transitional forms, unambiguously identified the taxa between which the transition was conceptualized, and simultaneously avoided cluttering the literature with additional binomial names (Repenning 1987). In his initial use, the hyphenated taxonomy was applied interspecifically within a genus (e.g., Mimomys (Ophiomys) taylori-parvus), and that convention also was followed by Repenning et al. (1990). By 1992 he applied hyphenated taxonomy intergenerically, but always with reference to species (conceptually, to populations of species) within sediments at a given locality. For example, he explicitly identified specimens from Cumberland Cave in Maryland as being intermediate between Microtus paroperarius and Lasiopodomys deceitensis (taxonomy rendered as Microtus paroperarius-Lasiopodomys deceitensis by Repenning 1992, p. 52). He continued to use hyphenated taxonomy until his death and it appeared in his last systematic work on arvicolines (Repenning 2003).

In addition to (and in conjunction with) his occasional use of hyphenated taxonomic nomenclature, Rep's writings on arvicolines are characterized by surprisingly frequent reference to the evolutionary origin of one genus from within a population of a species of another genus. Examples include his hypotheses of the evolutionary derivation of Synaptomys from within a southeastern population of an unspecified species of Mictomys (Repenning and Grady 1988), Phenacomys from within an unknown form of Cromeromys (Repenning et al. 1987), and Microtus (in part), Lasiopodomys, Terricola, Pitymys, Pedomys, Lemmiscus, and at least four other genera from within Allophaiomys pliocaenicus (Repenning 1992).

In two cases, he documented his view that origination events actually may have been captured in the fossil record; these are the possible derivation of Lemmiscus from within Allophaiomys at SAM Cave, New Mexico (Repenning 1992), and the evolution of Microtus paroperarius (in part) from within Lasiopodomys at Hamilton Cave, West Virginia (Repenning and Grady 1988). In both cases he later altered his opinion. He subsequently acknowledged that sufficient 'intermediate forms' were lacking between Lemmiscus and Allophaiomys in SAM Cave (Rogers et al. 2000). In the latter case, he reinterpreted specimens of Microtus in Hamilton Cave as representing extreme morphological variation within a new, but unnamed species of Lasiopodomys (Repenning 1992), but applied the awkward hyphenated taxonomy of 'Microtus paroperarius - Lasiopodomys deceitensis' to indicate his conception of the anagenetic derivation of Microtus paroperarius from within a population of Lasiopodomys deceitensis in Cumberland Cave (Repenning 1992, p. 53).

His taxonomic decisions were predicated upon his assessment of the variation within a given sample of fossils relative to the known variation represented in the typical populations (literally, from the type populations) of species to which those fossils might be referable. His determination of 'typical' morphological condition was specifically framed on the basis of the type population (not solely the morphological configuration of the type specimen), ideally represented by a complete sample of specimens collected from the type locality. This was true regardless of the 'stage of evolution' of the type population relative to any other reference fauna or sample. He used hyphenated taxonomy if the relative abundances of different morphotypes in a fossil sample under study were between 40% and 60% of the sample (those percentages were not explicitly justified and appear to have been arbitrarily established). The dominant morphotype was listed first in the hyphenated name. Samples (interpreted by Rep to be populations) with greater than 60% of one morphotype were assigned to the genus and species whose typical morphotype they most closely approximated. Morphotypes in the remaining 39% (or less) of the sample were considered to be individuals of that taxon, but with a morphology that overlapped the morphology of another taxon.

It is, of course, possible to interpret such samples in alternative ways. It can be difficult (here, a possible euphemism for 'impossible') to distinguish between samples that reflect some stage of incipient speciation (characterized by relatively high levels of variation, but not sufficiently high to warrant a hyphenated name) from those that preserve remains of closely related taxa in differential relative abundance, especially if the type locality of a species includes a wide range of morphotypes. An example is represented by the type collection of Microtus paroperarius from Cudahy, Kansas. The holotype specimen was collected from the Sunbright Ash Pit, but paratypes were established from Sunbright as well as Cudahy Ash pits (Hibbard 1944). The species was diagnosed by having lower first molars with only four closed triangles (Hibbard 1944). An extensive subsequent collection from the Cudahy Ash Pit site yielded thousands of specimens of Microtus paroperarius and was accepted by Repenning as representing the 'typical' condition for the species (Repenning 1992; Bell and Repenning 1999). That 'typical' condition included m1s with five closed triangles, as well as those with four closed triangles. In addition, upper second molars with distinctive enamel fields typically referred to Microtus californicus (in faunas from California) or Microtus pennsylvanicus (in faunas from the Great Plains and eastern United States) were present in the collection. Initial discussions between Rep and one of us (CJB) about the Cudahy fauna and its interpretation were heated, and resolved only by joint re-examination of the thousands of relevant specimens. Rep pursued the re-examination willingly, but in the loan request he filed with University of Michigan Museum of Paleontology to borrow the specimens, he referred (good-naturedly?) to his "meddlesome associate" (i.e., Bell; Repenning letter to Philip Gingerich, 22 April, 1995, p. 2). However, in a departure from his previous bounds, Rep ultimately accepted that the low relative abundance of specimens with five closed triangles (7.3% of the total relevant sample of m1s) could plausibly represent a different, but unidentified, species of Microtus (Bell and Repenning 1999).

At the time of his death, Repenning was beginning to modify his percentage-based approach to taxonomic allocation. His motivation derived in part from the unusual discoveries within Porcupine Cave in Colorado (Barnosky and Rasmussen 1988) and Cathedral Cave in Nevada (Bell 1995; Jass 2007; Jass and Bell 2011). At both sites discrete stratigraphic levels contained specimens referable to Allophaiomys pliocaenicus and several of its purported descendants, including Lemmiscus curtatus (represented by specimens with both four and five closed triangles on m1), Microtus paroperarius, Microtus meadensis, and Microtus with five closed triangles on m1. The faunas from the initial excavations in Cathedral Cave and the renewed excavations by the Denver Museum in Porcupine Cave were under study simultaneously in 1994 and 1995, and the Porcupine Cave material excavated by the Denver Museum of Natural History (now Denver Museum of Nature & Science) was studied by Rep individually (e.g., Repenning letter to Elaine Anderson, 17 June, 1994) and in collaboration with others for over a decade (see Bell et al. 2004a). Much of the correspondence that we reviewed between Rep and Bell was dedicated to the significance of the faunas preserved in those caves. The cave faunas posed a vexing problem for Rep, whose model of anagenetic evolution was inconsistent with ancestors being found in direct association with descendents (see discussions by Gould 1982 and Barnosky 1987). The discoveries stimulated renewed discussions about taxonomic philosophies, and the operational practices he employed in correlating faunas without external (to the fauna) age control.

Rep's initial responses to the discoveries in Cathedral Cave and Porcupine Cave were dismissive; he simply did not accept that the co-occurrences could be valid. His starting position was that the relevant specimens were misidentified, either through ignorance of the relevant anatomical features or because of inadequate appreciation of the range of morphological variation that could be present within a single population. Getting him past the first inclination often meant providing an opportunity for him to personally view the material and pronounce his judgment. One of us (CJB) took advantage of Rep's invitation to visit him in Colorado with specimens from the two cave sites in hand. During the first afternoon of a five-day visit in August of 1994, anatomical ignorance was ruled out by the simple expedient of placing specimens (without reference to their known stratigraphic context) under a microscope and independently writing down the name that each person thought should be applied to the specimen based only upon its morphology. Comparison of answers revealed that there was no fundamental disagreement over the identity of species, although slight differences in higher-order taxonomic preference were revealed, saving much frustration in subsequent conversations. Once taxonomic parity was achieved, the discussion resolved into an in-depth assessment of the implications of the stratigraphic associations.

At the time that visit took place, Rep had been convinced for over a year that identification of Allophaiomys from Porcupine Cave was in error. He noted "There is no Allophaiomys or Lasiopodomys in Porcupine Cave, they both became extinct 830,000 years ago. The presence of Allophaiomys morphotypes in the fauna has to be an individual variant of Microtus..." (Repenning, faxed letter to Christopher Bell, 7 July, 1993, p. 3). No resolution was achieved during that first meeting, and discussions continued for over a decade. Rep's concerns were expressed clearly at the end of the first day when the question of dinner was raised: "How can you eat at a time like this – when the whole profession is in a crisis? Boy, you are putting a hole in biochronology" (Bell, "notes from microtine investigations with Repenning, Aug. 18-22, 1994", p. 7; in correspondence file).

By 2004 no final resolution on how to interpret the Allophaiomys morphotypes existed, but in a coauthored summary of the Porcupine Cave material the co-occurrence with purported descendents was noted as "perplexing " (Bell et al. 2004a, p. 214), and the implications for taxonomy and evolutionary history were discussed briefly. This issue of taxonomic allocation of specimens with low sample sizes remained a perplexing problem for Rep, but was unresolved at the time of his death.