Palaeo111. afr., 34, 127- 198 ( 1997) NEW FOSSILS OF ALCELAPHINI AND CAPRINAE (BOVIDAE: MAMMALIA) FROM AWASH, ETHIOPIA, AND PHYLOGENETIC ANALYSIS OF ALCELAPHINI by E. S. Vrba Department of Geology and Geophysics, Yale University, P.O. Box 6666, New Haven, CT 06511 , U.S.A. ABSTRACT Alcelaphine antelopes comprise one of the most species-rich groups among the mammalian assemblages from the Middle Awash, Ethiopia, and in Africa as a whole. I describe a new genus and spec iesAwashia suwai from Matabaietu 3, and other new alcelaphine species, Damaliscus ademassui from Gamedah I and Beatragus whitei from Matabaietu 3-5, all dated ca. 2.5 m.y. (millions o f years). Other new a lcelaphine fossils from Middle Awa h include an Early Pliocene species all ied to Dama/ops, Late Pliocene records of Parmularius c f.pandatus and Beat rag us amiquus, and Middle Ple istocene records of Megalotragus kattwinkeli, P. angusticornis, Damaliscus niro, Connochaetes taurinus olduvaiensis, Numidocapra crassicornis, and Alce/aphus buselaphus. My comparisons of these fossils with all other known fossil and Recent Alcelaphini inc ludes a c ladistic analysi . The results suggest that du ring or before the Miocene-Pliocene transition two alcelaphine subtribes dive rged for which I suggest the names Alcelaphina and Damaliscina. Alcelaphina consists of two ancient subclades: ( I) the s is te r-group of Damalacra neanica and Beatragus known since 5.0-4.5 m.y. ago, and (2) a large c lade first recorded 4.4 m.y. ago (genera Damalops, Numidocapra, Alce/aphus, Rabaticeras, Megalotragus, Oreonagor, and Connochaetes) that had a high diversification rate s ince 3 m.y. ago. The earliest record of Damaliscina is the form that Ge ntry ( 1980) named Damalacra acalla, which emerges as the hypothe tical direct ancestor of the Early-Midd le Plioce ne split intoParmularius and the Damaliscus group. The placement of the new genus Awashia remains problematic. A new ov ibovine genus and spec ies, Nitidarcus asfawi, and a new caprine genu and species, Bouria anngettyae, both from Bouri I, are also described. I discuss some evolutionary and biogeographic implications of the new fossils from Middle Awash. KEYWORDS: Alcelaphini , Caprinae, Awash, Ethiopia INTRODUCTION Excavations of Plio-Pleis tocene sedimentary deposits in eastern E thiopia's Middle Awash Valley have recently yie lded Large new assemblages of fossil vertebrates including Hominidae (White et al. 1993; Clark et al. 1994; White et al. 1994). This description of antelopes in Alcelaphini and Caprinae known to date from Middle Awash will need additions and amendments in future as new foss ils are discovered in the ongoing Awash project. A cladistic analysis of Alcelaphini will be presented that includes the e ntire record of thi s tribe and places the new Awash fossils in genealogical and chronological context. This analysis includes many more taxa tha n did m y previous c ladogram of Alcelaphini (Yrba 1979), and new characters and new evaluations of previously used characters, and is therefore a major revision of that analys is. A simil a r prev io us paper described the foss il H ippotragini of the Awash in the context of a cladi stic analysis of all known taxa in that tribe (Yrba and Gatesy 1994). Table 1 gives fossil sites and stratigraphic subdivisions referred to in this paper and Figure 1 the chronology of the Awash strata. The living and fossil species of Alcelaphini are introduced in Table 2. MATERIALS AND METHODS This stud y i s based o n observa tions and measurements on all living and known extinct alcelaphines with emphasis on the new Awash fossi ls. Some of these d ata were obtained for previous analyses (Vrba 1971, 1976, 1977a, 1977b, 1979, 1984, Laubscher et a/. 1972). Figure 2 gives cross-sections of the basal horncores, and Figures 3- L 7 photographs of the Awash fossils. Their measurements (Tables 3-12) are compared to those of other alcelaphines in Figures 18-20. Comparison of the 40 skull characte rs in T a ble 13 across alcelaphine taxa resulted in the codes in Table 14 that were used in the cladistic analyses. To estim ate the c lad is ti c codes (Table 14), ontogenetic growth patterns and graphs of measure­ ments were compared across taxa. My aim was to arrive at code separations that are independent of body size. Characters 1-3 , 9, 14, 15, 19-24, 28-30, and 38 (Table 13) were a na lysed quantitatively. The others were coded qua litatively because the presence or absence of the states could readily be evaluated (e.g., horncore torsion, 12, 13 in Table 13), or because it was difficult to quantify the character differences (e.g., the extent of the preorbitaJ fossa, 32 in Table 13). In order to exp lore the extent of AGE m.y. 0 1 2 3 4 5 6 ASAKOMA BIKIRMALI SEVERAL KOMA STRATA SAITUNE (See caption) DORA ..................... ······················ ...................... -----------------······ ...................... ....................... \ ·················-···· ...................... AGG, AME AMH, ············-·--·····-~ .. AMW,i- GAW, KUS, URU, WKH .ASK,BIK, .. ------·······-·····-·· STD SAGAN-ARAMIS BUNKETO TOLE ~ ······················· ...................... ·····-··············· ···-···--------------- ···········----------- -------·-····--------- ...................... ...................... ······················ BUN 1-5 ...................... ...................... ··-·················· ARA 1-14 SAG 1-7 ·······-······-······· ...................... ..................... ---·····---4·-··-······ ···-·-·-·-······-···-· ···-···---- ···-~-----· MIDDLE AWASH HADAR DATES (m.y.) MAKA/ MATA- WILTI WEE-EE GAMEDAH BOURI BODO BELOHDELIE BAIETU DORA (a)-(i)=sources BOD 1 0.64 ± 0.03 (a) . BOU 1-7 ······················ ...................... ····················· ....................... ······················ ...................... ······················· 1.0 (b) ··--------·····-······ ······-····-·········· ···------------------- ----------------------- ---------------·--···· ······-····--········ ...................... ··················--··· MAT 1-2 2.5 (b) 3-5 WIL 2-3 GAM 1 MATS 2.95 ± 0.02 (c) ····················.·· ..................... ...................... ....................... ..................... . ..................... ...................... ······· KH ········ :::: DD 1-3;:: 3.18 ± 0.01 (d) SH 1-4 3.22 ± 0.01 (e) WEE5 MAK1 MAT7 BOD2 "3.40 ± 0.03 (f) 3.75 ± 0.02 (g) WEE1-4 BEL 1-3 WIL 1 BOD3 3.91 ± 0.02 (h) ...................... ····················· ...................... ....................... ..................... ····················· ...................... ....................... 4.39 .:!" 0.03 (i) ...................... ..................... ············-········· ············-·········· ·········4·-········· ·······-·············· ...................... ...................... 5.0 (b) --···-···-············ ··-················· ...................... ....................... ..................... ...................... ...................... ...................... Figure I. Chronology of stratigraphic sections and fossil iferous units in the Middle Awash, and also Hadar, from which bovid fossils come. See Table I for names and acronyms. This scheme is subject to revision as field and laboratory analyses progress. Tuff names and sources for dates are as follows: (a) pooled estimate of weighted mean age for the ruff represented by MA90-20 at Hargufia and MA90-23 at Bodo which are homologous based on stratigraphic and geochemical criteria (Clarke/ a/. , 1994); (b) preliminary, approximate, unpublished detenninations (White, pers. comm.); (c) tuffB KT-2L, (d) tuff KHT, (e) tuffTf4 ((c)-(e) are after Walter, pers. comm., as in Kimbel, 1995); (f) based on the follow ing tuffs that are chemically homologous (review in Whiteet a/., 1993): MA90-28 at Wee-ee, MA90-39 at Bunketo South, MA90- 14 at Wilti Dora, SHT at Hadar, a tuff from the Gulf of Aden core 23 1, and Tulu Bor beta tuff in the Turkana Basin; (g) tuff YT-3 (=Wargolo tuff, White eta/., 1993); (h) tuff YT-1 (Whiteet a/., 1993); (i) GATC tuff complex (WoldeGabriel, eta/., 1994). N 00 allometric effects on the characters, for each of the 16 quantitative characters a least squares regression line was calculated for the individual values regressed against a proxy for body size. The proxy used was basal horncore size as defined in Figure 20. In the cases of the angle measurements (9, 14, 15, 19, Table 13), no significant correlation with body size was found and the code separations wereEtimated di rectly by inspection of the variation. Figure 18 illustrates this approach for characters 15 and 19. Horncore compression (J in Table 13) was taken to have state 0 below and state 1 above the line along whic h the ratio of the basa l diameters is 80% (Figure 19). (For this as for some other characters there seems to be variability with increasing body size in certain taxa. Such ontogenetic variation will be discussed in the re levant taxonomic sections.) For other quantitative characters, the regression lines with confidence intervals were used as an approxi­ mate guide to code separations, as illustrated in Figure 20 for basal horncore separation. In this case, taxa with mean values below, between, and a bove the 99 % confide nce intervals were respectively assigned codes of 0, 1, and 2. About 10% of codes are uncertain (coded ? in Table 14) mostly due to missing data on incomplete fossils. In a few cases the variation within a taxo n for a character is large. For instance, because the values for horncore compress io n in li ving Alcelaphus buselaphus vary ex te nsive ly (Figure 19), and because the original state in this taxon is unknown, a ?-code was assigned (Table 14). In each of a few additional cases a ?-code was assigned because the mean of a small sample, or the only available value, was found to be intermediate. An example is the the angle be tween the maxi mum horncore diameter and and the midfrontal suture (15 in Tables 13 and 14) in the single known specimen of Awashia suwai (AsMAT3 in Figure 18). Multistate characters were ordered on the basis of ontogenetic progression and s i milari~y between adjacent states (see Lipscomb 1992). The meanin g of ?-codes in two-s tate characters is uncertainty between states 0 and 1, but in multistate characters it is open to additional interpretation. In the final c ladistic result (Figure 2 1) all cases of ?-codes in multistate characters were interpreted appropriately. The codes in Table 14 were analysed using outgroup rooting. Cladistic results from mitochondrial DNA (Gatesy 1993; Gatesy eta/., 1992; Gatesy et al. in press) and nuclear DNA sequences (Gatesy pers. comm.) gave independent information on alcelaphine outgroups. While these results varied to some extent depending on the combinations of DNA sequences and computational methods used, they consistentl y supported monophyly of Alcelaphini and identified five additional monophyletic taxa as being among the outgroups or as subclades of outgroups: Caprinae (Caprini , Ovibovini , Rupicaprini ), Hippotragini , Reduncini, Aepyceros, and Antilopinae (Antilopini, 129 Neotragini). In a11 the DNA cladograms, Caprinae (1) or Hippotragini (2) are either the first two out­ groups or the sister-clade Caprinae-Hippotragini (3) is the first outgroup. The next-nearest outgroup is always a c lade Redunc ini-Antilopinae-Aepyceros (4) or a clade Redunc ini-Antilopinae (5), or Aepyceros (6). I arrived at the outgroup codes in Table 14 by estimation of the ples iomorphic states for skull characters for each of clades (1)- (6) using previous analyses of bov id skull morphology (Vrba et al. 1994, for Redunc ini; Vrba and Gatesy 1994 , for Hippotragini ; Vrba and Schaller, in prep., for Caprinae and Antilopinae; see also Gentry 1992). For six characters in Table 14, decisions on character polarity at the aJcelaphine outgroup node remain uncertain, and the codes I used reflect my current best estimates (characters 2, 6, 19, 21, 23, 30). For the remaining characters the outgroup states in Table 14 are more securely based as they appear to be the plesiomorphic states in each of clades (1) - (6). The cladistic analyses used options (outgroup rooting, mhennig*, bb*, and NELSEN for consensus trees) of the HENNIG86 program (Farris 1988). Mhennig* constructs several trees, each by a single pass thro ugh the data , and then applies branch­ swapping to each of these initial trees. The bb* option applies more extended branch-swapping to the initial trees and generates all the shortest trees it can find. Mhennig* and bb* are not certain to frnd all trees of minimal length. The full data set is too large to use the ie option of HENNIG86 which does find all trees of minimal length. This option was applied after treating as single taxa the most strongly-supported terminal clades in the consensus tree (e.g. , B eatragus, Figure 2 1). The results confirmed that the consensus tree for each of my cladistic analyses does represent the consensus of all trees of minimal length. The systematics of Mega/otragus and Awashia were subjected to additional cladistic explorations. In the case of Megalotragus, the codes in Table 14 were analysed in two cladograms that differed only with respect to taxonomic subdivision of thi s genus. In o ne analys is, M . kattwinkeli as prev iou s ly recognized (e.g., Gentry and Gentry 1978), Mega­ lotragus from Bouri in the Awash, and M. isaaci from the Koobi Fora Formation (Harris 1991) were treated as three separate taxa (Mk, MkBOU, Mki in Table 14). The second analysis treated these three forms as one variable -species (Mk* in Table 14). The results, which differ only with respect to branching within Mega/otragus, will be discussed. In the case of Awashia, many character states that turned out to be important to the basal branching sequence in Alcelaphini are difficult to interpret in the s ing le known s kull , e ithe r because it is intermediate or because its eroded condition leaves open what the correct states are. I will report on the outcomes of using different interpretations of such characters in Awashia. 130 TABLE!. Fossil sites and stratigraphic subunits referred to, with codes in capital letters that are used in tables and figures. Stratigraphic subunits are arranged roughly from latest at the top to earliest at the bottom. Table 2 cites sources for bovid records. For stratigraphy and chronology see Figure 1 for the Middle Awash, and Vrba's (199Sb) review for other sites. Loc. = Locality; F.= Formation; M . = M ember; sm. = submember. Site A in Boucherit Ain Hanech Ain Jourdel Anabo Koma A wash, Middle (Figure I) Bodo Loc. 1 Bouri Loc. I, 2 , 6 Matabaietu Loc. 1-5 Gamedah Loc. I Wilti Dora Loc. 2-3 Matabaietu Loc. 6, 7 B unketo Loc. 1-5 Maka Loc. I Wee-ee Loc. 1-5 BelohdeUe Loc. l -3 Wilti Dora Loc. I Aramis Loc. I, 4, 6, 8 Sagantole Loc. l -7 Agera Gawetu Loc. I A mba East Loc. I Amba West Loc. I Amboul Hare li Loc. I Gawto Loc. l Kuseralee Loc 1, 2 Urugus Loc. l Worku Hassan Loc. Asa Koma Loc. I Bikirmali Koma Loc. I Saitune Dora Loc. I Cornelia Elandsfontein Florisbad Hadar Formation Kada Hadar M. Denen Dora M., sm. 1-3 Sidi Hakoma M., sm. 1-4 lsi mila Kakesio Kanapoi K.romdraai A Koobi Fora Formation Chari Member Okote Member KBS Member Upper Burg i Member Tulu BorMember Lokochot Member Langebaan weg, Varswate r F. Laetoli Beds Makapansgat Limeworks M. 1-5 Nachukui Formation Upper Lomekwi Member Olduvai Gorge Beds I-IV Peninj Rabat Rusinga Island Shungura Formation, M. B-J Siwaliks formations, Pinjor Swartkrans Formation, M. 1-3 Sterkfontein Formation, M. 4 Tadzikistan, Kurufsai Temifine Upper Ndolanya Beds WadiNatrun 1 Country Algeria Algeria Algeria Djibouti Ethiopia South Africa South Africa South Africa Ethiopia Tanzania Tanzania Kenya Transvaal, S. Africa Kenya Cape, South Africa Tanzania South Africa Kenya Tanzania Tanzania Morocco Kenya Eth iopia India and Pakistan South Africa South Africa Tadzikistan Algeria Tanzania Egypt Time period Late Pliocene Early Ple istocene Early Pleistocene Early Pleistocene Plio-Ple istocene Middle Pleistocene Middle Pleistocene Late PI iocene Late Pliocene Middle-Late Pliocene Late Pliocene Middle Pliocene Middle Pliocene Middle Pliocene Middle Pliocene Middle-Late Pliocene Early Pliocene Early Pl iocene Early Pl iocene Early Pliocene Early Pliocene Early Pliocene Early Pliocene Early Pliocene Early Pliocene Early Pliocene Late Miocene Late Miocene Late Miocene M iddle Pleistocene Middle Ple istocene Late Pleistocene Middle-Late PUocene Middle Pleistocene Early Pliocene Early Pliocene Middle Pleistocene Plio-Pleistocene Early Middle Pliocene Late Pliocene Plio-Ple istocene Pleistocene Pleistocene Ple istocene Late Pleistocene PI io-Pleistocene Plio-Ple istocene Lower Ple is tocene Late Pliocene Late Pliocene Middle Ple istocene Late Pliocene Earliest Pliocene Acronym AINB AINH AINJ AK BOD! BOUI , 2, 6 MATI -5 G AMl WIL2-3 MAT6, 7 BUN I -5 MAKl WEEI-5 BELI -3 WILl ARAl , 4 , 6, 8 SAG I-7 AGG AME I AMWl AM HI GAWI KUSl , 2 URU I WKH1 ASK! BIKI STDI COR E FLOR KH DD SH lSI KK KP KA C HA OKT KBS UBU TUL LOK LAN LIT Ml-5 ULM PE R RU S-B - S-J p SKI -3 ST4 TAD T UN WADN Vrba et at . ( 1994) used significance tests to estimate code separations in quantitative characters. Most of the present samples are too small to apply such methods. The difficulties inherent in cladistic coding are especially felt in an analysis such as this one with its small samples for closely related taxa. There remain problems of homology among taxa, of which characters to exclude so as to eliminate overlap between characters, and of where the code separations should be located in the spectrum of var iation. Each statement about a character in Table 14 i s an hypothesis to be tested, and to be rejected by additional data. These hypotheses test each other in the cladistic analysis. SYSTEM A TIC RESULTS The cladistic analysis that treated Mk, Mki and MkBOU in Table 14 as three separate taxa resulted in the consensus tree in Figure 21. I will argue below that all three belong to a single species, an inclusive M. kattwinkeli. The analysis that treated them as a single species (Mk* in Table 14) resulted in a consensus tree of identical branching topology to that in Figure 21, except for the Megalotragus clade in which M. atopocranion, M. priscus and the inclusive M . kattwinkeli formed a trichotomy. The phylogenetic tree in Figure 22 is based on th is second result, with taxa plotted against time using the dates in Table 2. Eac h taxon without autapo­ morphies in the consensus tree (Figure 21) is the hypothetical potential ancestor of its s ister-group. Such taxa are shown in the appropriate ancestral positions in Figure 22. Some taxa have been cited as alcelaphine in the pas t , but are not inc lude d in my cladogram. Aep yceros ( impalas) has been regarded as alcelaphine (Gentry 1978) or as the sister-taxon of Alcelaphini (Yrba 1979), but cladistic results from DNA sequences cited earlier strongly suggest that Alcelaphini are more closely related to caprines and hippotragines than to Aepyceros. Thomas (1984) proposed that Alcelaphini consists of two basa l sister-lineages: Maremmia haupti , a small, dentally advanced antelope from the Turolian of Baccinello, Italy, and the sister-group (Aepyceros + [Alcelaphini sensu Figure 2 1]). I doubt that thi s is correct. Alcelaphini in my sense, their closest re latives based on DNA studies (living Caprini and Hippotragini), and Aepyceros, all share some kind of hollowing at the horncore bases, while Maremmia lacks thi s feature. I prefer the second alternative offered by Thomas (1984): that Maremmia may be the sister­ taxon of the Middle Mioce ne , Afro-Euras ian Caprotragoides. Klein a nd Cruz-Uribe (1991) described a fo rm f rom Elandsfontein as ?Parmularius sp. nov. It shares with Parmularius a Lateral horncore boundary that is posteriorly much lower than anteriorly; very low horncore separation and divergence, and long pedicels. It shares with P. pandatus strong backbending of the homcore, 131 and with P. braini strongly compressed horncores, but it is unlike Parmularius in the shape of the horncore-pedicel boundary, c lockwise tors ion in the right horncore and in lacking posterolateral basal horncore swellings. There are resembla nces in horncore torsion and course (e.g., in the horncores remaining very close along their entire length and being strongly backbent with rapid diminution), and in the shape of the horncore-pedicel boundary to caprines like Sinotragus wimanni (Bohlin 1935). However this latter Late Miocene form is geo­ graphically and te mporally far remove d f rom Elandsfontein. I see the Elandsfontein form as e ither a very aberrant alcelaphine allied to or basal within Parmularius, or as a caprine. The following are alcelaphines, and inc luded in Table 2 as such, but were omitted from the cladistic computations because they are based on scant material. Damaliscus "hipkini" (the name that the late L. H. Wells once intended to give this species, although he neve r described it, pe rs. comm .; Damaliscus sp. nov. of Klein and Cruz-Uribe 1991) is a small form known from the Middle Ple istocene of Elandsfontein (based on frontlets EFT 3781 and EFT 5143) and Cornelia. (I stress that specific names in quotation marks, like "hipkini" , are not formally described . I use these informal descriptive names in quotation marks, rather tha n a less inte lligible numbering system, to make more comprehensible to the reader my extensive references to several undesc ribed species.) D. " hipkini" has s trongly compressed horncores with flattened lateral surfaces, lacking basal swellings and ari sing from moderately long pedicels. These features suggest that it belongs in Damaliscus. Damaliscus or Parmularius "maka­ pani" is based on a small frontlet M8246 from Makapansgat Member 3. Features that resemble Parmularius are horncores with some basal swelling and lacking lateral flattening, long pedice ls and a definite parietal boss. The species may be a new small Parmularius species or a juvenile of a larger Parmularius species, rather than connected to the ancestry of Dama/iscus as I previously suggested (Yrba 1995b). The next taxa, although omitted from the cladogram, will be discussed in relation to similar Awash fossils. Beatrag us "e landsfonte ini" (Beatragus sp. of Gentry 1978; Damaliscus aff. lunatus of Klein and Cruz-Uribe 1991) is known from Elandsfontein and may be conspecific with a horncore from Swartk.rans Member 2 (Yrba 1976). I argue below (following Gentry 1978), that this form is closely a llied to , and perhaps descended from B. antiquus. Damaliscus gentryi (Yrba 1977a), based on a single horncore from Early Pleistocene Makapansgat Member 5 may be close to D. niro. (Damalops) sp . indet. in T able 2 is based on Parestigorgon gadj ingeri from Upper Ndolanya, Tanzania, which was described by Dietrich (1950) based on a horncore and some dentitions. These fossils are related to Damalops palaeindicus (Gentry 132 TABLE2. Living and extinct alcelaphine taxa and their first a nd last appearance data (FAD, LAD, first a nd last entries). Other sites are cited in some cases. See Table 1 for letter codes for fossil sites, Figur e 1 for Awash dates, a nd Vrba's r eview (1995b) for other dates and litera ture sources. Only subspecies that a ppear in this paper are given. Genus names in brackets r efer to past assignations that are not upheld by the cladistic r esults in Figure 21 as monophyletic and/or worthy of generic distinction. Such cases include taxa assigned to a genus during original description or later revision, as well as undescribed new species that were loosely associated with a genus in the litera ture. As the following undescribed species, will be extensively referred to, it will be done by descriptive na mes in quotation marks rather than use a less intellig ible numbering system (at least one good specimen and its source site are added in each case in brackets): (Damalops) "sidihakomai" (AL 208-7, Hadar SH), (D.) "denendorai" (AL 161-5, Hadar DD), (Rabaticeras) " lemutai" (8.208, Olduvai Bed II), Damaliscus "hadari" (AL 146-1, Hadar SH, Damaliscus " h ipkini" (EFT 3781 a nd EFT 5143, Elandsfontein), Beatragus "elandsfonteini'' (EFT 16561, Elandsfon tein), a nd Damaliscusi?Parmularius "makapani'' (M8246, Makapansgat Member 3). Genus Damalacra (Damalacra) Damalops (Damalops) (Damalops) (Parestigorgon) Beatragus Awashia gen. nov. Damaliscus Species Subspecies neanica a calla palaeindicus "sidihakomai" "denendorai'' sp. indet. hunteri antiquus whitei sp. nov. "elandsfon- teini" suwai sp. nov. dorcas lunar us l.jimela I. korrigum l. lunarus niro agel a ius gentryi ademassui sp . nov. "hadari" "hipkini" Site of FAD Site of LAD LA WAD LAN LAN TAD PI ARA KK KP LIT MAK WEE5 LOK TUL SH DD S-B KH UN C HA Wll.-2 GAM 0 11 Kit K MAT3-5 MAT3-5 ?SKI E MAT3 MAT3 SK2 SK3 Oil SHK BOU BODl FLOR Oil FLK om JK2 M5 M5 G~M GAM SH SH COR FAD m.y. Extinct= t Taxon LAD m.y. Common name acronym 5.0 t ne 4 .5 5.0 t ac 5 .0 2.6 t Dop 2.6- 2.0 4.4-4.2 t Dos 4.4 4. 1 3.76- 3.46 3.75- 3.40 3.75- 3.40 3.50- 3.36 3.36- 2.68 3.40- 3.22 3.22- 3.18 t Dod 2.9 2.9 2.9-2.6 Do?U 1.39- 0.67 Hunte r's Hartebeest Bh 2.5 t Ba 2.5 > 1.48±.05 2.6 t Bw 2.6 1.8 t Be 0 .6 2.6 t A s 2.6 1.1 Blesboketc Dd 0.7 Tsessebes etc Dl To pi Dlj Korrigum Dlk Tse sebe DIJ 1.66- 1.48 t Dn 1.0 0.6 .2- . l 1.75- 1.66 t Da J .33-0.96 1.8 - 1.6 t Dg 1.8 - 1.6 2.5 t Dad 2.5 3.40 t Dha 3.40 0.8- 0.6 t Dh 133 TABLE 2. continued Genus Species Site of FAD FAD m.y. Extinct= t Taxon Subspecies Site of LAD LAD m.y. Common name acronym E 0.6 Parmularius pandatus LIT 3.76- 3 .46 t Ppan ?WlL2 ?2.5 braini M3 2.8-2.6 t Pb M3 2 .8-2.6 eppsi S-C 2.6 t Pe OKT 1.6- 1.39 altidens AINB 2 .7- 2.5 t Pal 0 1 HWK 1.8 - 1.75 angusticornis OJ FLKN 1.80- 1.75 t Pan BOU 1.0 lSI 0.8-0.6 rugosus OII HWK 1.75 - 1.66 t Pru OIV HWK 0.78- 0.7 parvus KA I - 0.7 t Ppar OIY 0.78- 0.7 ambiguus T 0.7 t Pam T 0.7 Parmularius or Damaliscus cuiculi AINB 2.7- 2.5 t D/Pc AINB 2.7 - 2.5 "makapani" M3 2.8- 2.6 t D/Pm M3 2.8- 2.6 Alcelaphus buselaphus BOD I 0.6 Hartebeest Ab b. buselaphus t Abb b. caama Cape Hartebeest Abca b. cokii Coke's Hartebeest Abco b.jacksoni Jackson 's Hartebeest Abj b.lelwel Lelwel Hartebeest Abl b. major Western Hartebeest Abm b. swaynei Swayne's Hartebeest Abs b. lora Tora Hartebeest Abt Alce/aphus lichtensteini SEMLT 0.5- 0.3 Lichtenste in 's (Sigmoceros) Hartebeest Ali Rabaticeras arambourgi OUJ JK2 1.33 - 0.96 t Ra OIV PDK 0.8 R 0.6 E 0.6 (Rabaticeras) porrocornutus SK I 1.8 t (R)p S KI 1.8 (Rabaticeras) "lemutai" on Tuff iiA 1.66 t (R)le Oil Tuff IlA 1.66 Numidocapra crassicornis AI H 1.7 t c AK 1.6 ?Oil SHK 1.5 BOU 1.0 Megalotragus kauwinkeli MATt 2.5 t M k SH-DI 2.5 ST4 2 .6-2.4 BOU 1.0 OLI- IY 1.5- 0.6 k. isaaci UB U 2 .0- 1.88 t Mki OKT 1.6- 1.39 priscus COR 0.8- 0.6 t Mp NBC .012 -.009 Megalotragus atopocranion R U Late ... I Ma (R usi ngoryx) Pleistocene Connochaetes taurinus Oil MNK 1.66- 1.48 Blue Wildebeest Ct t. olduvai- Oil MNK 1.66- L.48 ens is PE 1.1 BOU 1.0 gnou COR 0.8- 0.6 Black Wildebeest Cgn african us on 1.75 - 1.33 ... I Ca Oil 1.75 - 1.33 gentryi ULM 2.5 t Cg Oil 1.66 - 1.48 Oreonagor tournoueri AINB 2.7- 2.5 t Ot AI J 1.8 ?Oreonagor/?Megalotragus BO U 1.0 ... I ?Osp.BOU sp. BOU 1.0 134 a Q•~·• §••••=•<» b (~ ,.,.,.._., . .....,.._. ........._. ·~ 0 @j,., d -- 8 .... ,,. --- ·8@ Figure 2. Cross-sectionsofba al alcelaphine homcores from Middle Awash, Olduvai Lower Bed II (i. = OLD/'7 1 HWK E Rootlet ClayS. 2 17), and living Lichtenstein 's Hartebeest (d.). Thesmallercross-section for D.niro hom core BOD 1/ 17 is from J 20 mm above the base. The cross-sect ions to left and rightofthecentral line areof actual left and right homcores respectively, except that of left homcore D. niro BOD I /17 (I.) which was transposed to the right. Actual horncore separation is represented by d is tance from the central line. and Gentry 1978) and may belong to (Damalops) "denedorai" (see below). Numidocapra crassicornis (Arambourg 1949), a large antelope from Ain Hanech, Algeria, and Anabo Koma, Djibouti , Ethiopia (Bonis et a/. 1988), and now also from Bouri in the Awash, was considered by Geraads ( 1981) as alcelaphine, but by Gentry (1978, 1990a) as caprine. I will discuss why I see crassicornis as specifically distinct from, but close ly related to a nd perhaps d irectly a ncestra l to R. arambourgi. When I refer in the systematic descriptions to clockwise or anticlockwise torsion of horncores, I mean in the right horncore from the base up. The following terms will also be used: the 'craniofacia l angle ' is the angle between the forehead and the dorsal braincase (character 19 in Table 13), and the 'parietal-occipital angle' is that between the straight line from bregma to occiput, and the straight line from occiput to the top of the foramen magnum. Tribe Alcelaphini (Rochebrune 1883) 1883 Alcelaphidae Rochebrune 1898 Bubalinae Trouessart 1945 Alcelaphini Simpson Alcelaphini are medium to large antelopes in which females as well as males have horns. T hey have a sinus in the frontal, in the area of the pedicel and horncore base, that is extensively developed such that the surrounding bones are thin relative to s inus vo lume and there are hardly any bony intrusions into the sinus. An associated feature is the high level of the frontals between the horn bases relative to the orbital rims. Shallow elongated post­ cornual fossa. Supraorbital foramina flush with the frontal surface, in very small pits. Preobital fossae and glands pleisiomorphicall y well-developed. The large premaxillae rise with roughly even width to have a long contact wi th the long nasa l bones. Ethmoidal fissures are absent in adults. Tooth rows are set anteriorly. The zygomatic arc h deepens anteriorly beneath the orbits. Basioccipital with a central longitudinal groove between we ll-developed longitudinal ridges behind the anterior tuberosities. Occipital surface facing partly laterally as well as backwards. Mastoids large. Temporal ridges wide apa rt o n t he posterodors al braincase. Teeth hypsodont , without goatfolds on lower molars (protostylids), without basal pillars (the ectostylids on lower and entostyles on upper molars) except in some earliest forms and juveniles, w ith small hypoconids on P4 and w ith rounded molar lobes. Repeated trends w ithin the tribe include shortening of the braincase, decrease of the craniofacial angle, lengthening of the face, reduction of preorbi tal fossae, reduction of premolar rows, and fusion of paraconid and metaconid on P 4 • In li ving species: pedal glands present on forefeet only; no inguinal glands; one pair of mammae; tail either long, crested and bordered on top with a fringe of black hair (Alce /aphu s and Damaliscus) or white hair (Beatragus), or very long horse-like tail with long ha ir (Connochaetes); whithers higher than rump. (See also Gentry and Gentry 1978; Gentry 1980; Yrba 1979). Subtribe Alcelaphina, nom. nov. A lce laphina are medium to large a lcelaphines. The basal horncore margin is lower laterally than medially. The horncore in lateral view is concave soon above the base such that the horncore recurves in an anterior direction. The ang le of the basal horncore to the midfronta l suture is pleisiomorphically large. There is no distinct parietal boss on the dorsal braincase. The braincase roof is straight to concave. A feature that evolves early within the subtribe is clockwise horncore torsion. Gen. indet. aff: Damalops Pilgrim, 1939, (here referred to as "Damalops") Several cranial pieces and dentitions from Aramis, Maka and Wee-ee (Figures 3, 4, Tables 3, 11, 12) belong to a single evolving lineage and species that is best known from the Hadar Sidi Hakoma Member (I will use the short form Hadar SH). It is a lso known from Kakesio in the Laetoli area,..Kanapoi, the Laetoli Beds, and the Lokochot and Tulu Bor Members in the Koobi Fora Formation. The foss ils from Hadar and Laetoli we re discussed in Gentry (1980, 1981) as ?Damalops. Harris (1991) discussed those from Koobi Fora unde r diffe rent names, including as Parmularius cf. P . angusticornis. This species has not yet been formally described, and I refer to it informally as (D. ) "sidihakomai". A similar species, that in my analysis emerges as the direct - althoug h substantially c hanged - descenda nt of "sidihakomai", is represented best in the Hadar Denen Dora and Kada Hadar Members (Hadar DO and KH). This form is also undescribed, and I refer to it as (D.) "denendorai" . It is also present probably in Shungura Bll (= Shungura Member B subunit 1 1) ca. 2.9 m .y. a nd in the Upper Ndolanya assemblage by horncores prev ious ly assigned to Parestigorgon gadjingeri (Dietric h 1950; Gentry 1987). Gentry (1980, 1981) regarded H ad a r "sidihakomai" and "denendorai" as successive parts of a single species, ?Damalops sp., and as a close 135 relative of D. palaeindicus from the Pinjor Formation of the Siwaliks (Pilgrim 1939) and Tadzhiki stan (Dmitrieva 1977). Pilgrim 's (1939: 67) origina l diagnosis for Damalops and for the type (and only known) species D. palaeindic us inc luded the following: "horn-cores mounted on a ridge, but not far away from the orbits, close togethe r, at firs t almost parallel and at right angles to the braincase, then c urving very g radua lly outward and backward", an almost straight profile of the dorsal braincase, anteriorward expansion of the nasals and muzzle, long narrow nasals and face, very deep and extensive preorbital fossa, and presence of P 2 where this can be seen. Gentry ( 198 1: 14) noted the following features of D. palaeindicus in comparison with the Hadar forms: similarities include " high and narrow skull; horn cores inserted closely together and not very uprightly above the backs of the orbits, nearly parallel proximally but increasingly divergent dista lly; ... braincase rather long by comparison with presumed later alcelaphines; braincase roof inclined with little sign of a parietal boss; preorbital fossa large; nasals pronouncedly narrow as a ridge between the preorbital fossae; and a deep face . [Diffe rences inc lude that] D. palaeindicus shows backward curvature of its horn cores, horn cores not tape ring rapidl y above the base , sides of braincase parallel and not widening posteriorly, and probably also longer hom cores, shorter braincase, TABLE3. Measurements of horncores and frontlets related to Damalops: (Damalops) "sidihakomai" horncores and partial frontlets: ARA-8/3 frontlet with left horncore, MAK-1/32a and b left and right horncores, WEE-5/10 left horncore, compared with AL349-3 left horncore from Hadar cf. SH, Laetoli cranium 1959.233 proba bly from the Laetoli Beds; and (Damalops) " denendorai'' right horncore from Hadar cf. DD. Length in mm; two values for horncore length are preserved/estimated complete lengths; e =estimated , ee =very rough estimate; max. = maximum. (Damalops) "sidihakomai" (D .) "denendorai ARA- MAK- WEE- AL349-3 Laetoli AL169-26 8/3 1/32 5/10 (cf. S H) 1959.233 (cf. DD) Horncore max. diamete r (HMAX) 38.4 48.2 55e 5 le 5 1.5 60.5 Horncore min. diameter (HMI ) 37.63 43.5 48e 47.4 43.5 50ee Homcore ratio (HMIN/HMAX) 0.90 0.87 0.93 0.90 0.90 83ee Horncore length 11 0/140 230/290 190/270 130/210 Horncore basal separation 22.4 30ee 27.0 Angle of basal horn divergence 30° Angle of HMAX to midfrontal suture 70° 60-70° 90"ee Width across horn pedicels 123ee 11 0.1 Craniofacia l angle 120°ee 105°e Max. separation supra- orbital foramina (SOF) 82e 67.7 Di tance orbit to horncore 38e 136 and a toothrow positioned more anteriorl y. P 2 is absent on the only two Hadar specimens (both in middle wear) on which its state is determinable, and thi s contrasts with its presence on a single D. palaeindicus." D. palaeindicus lacks the specializations in horn shape, associated with the evolution of horn torsion, that characterize its relatives from recent common ancestry, Alcelaphus, Megalotragus, Oreonagor and Connochaetes. It also lacks most of the innovations in skull and facial shape of these taxa. As a result D. palaeindicus retains the s tronges t overall resemblances to the a nces tral lineage from "sidihakomai" to "denendorai" as noted by Gentry (1980, 1981 ). In term s of my results, this resemblance rests on plesiomorphic retention of ancestral characters. That is, a genus Damalops including the species palaeindicus, "sidihakomai", and "denendorai" i s paraphyletic , and lacks autapomorphies and definition in the sen se of De Queiroz and Gauthier (1994). If Damalops is left to include only pa/aeindicus, then a separate new genus inc luding "s idihakomai" and " denendorai" would also represent a paraphyletic grade. Nevertheless, I suggest that we need a single genus name for such an unbranching lineage of s uccessive distinct morphologies in spite of its ancestral status and, if my hypothesis of ancestry (Figure 22) is upheld, then a new generic name is needed for this long-lasting , basal lineage from c. d. Scm "sidihakomai" and "denendorai". Until these forms are described, we need to refer to them using some provisional nomenclature that maintains links with previous usage to facilitate communication. To this end I suggest the temporary names (Damalops) "sidihakomai" and (D.) "denendorai" . Damalops sidihakomai sp. nov. from Aramis 1, 4, 8 and 9, and also Maka 1, and Wee-ee (Figures 3, 4 , Tables 3, 11 , 12): The Aramis localities at which this taxon is found are dated 4.4 m.y. and Maka 1 and Wee-ee 5 ca. 3.4 m.y. (Figure 1). The earlier form from Aramis was about the size of a large blesbok. The later ones from Maka and Wee-ee are of the larger size of a living topi , with the Wee-ee material being the largest. ARA-8/3 is a broken frontlet with most of the left and part of the right horncore (Figure 3). (Catalogue numbers for Middle Awash vertebrate fossils contain the letters VP, ARA-VP-8/3 in this case, but I use the shorter form. Also, I abbreviate the names of Middle Awash localities, e.g. , Aramis 8 for Aramis Locality 8.) The horncore is little compressed (Figure 19), with a distinctly flattened posterior surface and approach to a posterolateral keel, inserted moderately closely to its neighbour (Figure 20) on pedicels that are not short but ill­ defined with sloping sides, and quite uprightly with a gentle anterior concavity. The pedicel-horncore transition is dis tinct , and late rally lower than b. Figure 3. (Damalops) "sidihakomai" sp. nov. from A ram is: a. and b. ARA-8/3 frontlet with left horncore in left lateral and anterior views; c. ARA-8/9 left maxilla with P3-M2 ; d. ARA-4/ l left mandibular teeth M 1 fragment and M 2 • 137 b. c. Fig ure 4. (Damalops) "sidihakomai" homcores (anterior always to the right; b. in right lateral view, others in medial view): a. left WEE-5/ I 0; b. and c. right and le ft MAK-1/32; d. left AL 349-3 from Hadar Sidi Hakoma Member. medially. The horncores diverge moderately, with divergence increasing slightly distally but not by a sudden or marked change in course. The maximum diameter (which is barely the maximum one and only due to the posterolateral keel) is situated at a posterior angle of ca. 70° to the midfrontal suture. T he re is no horncore tors ion. C losely-spaced transverse ridges can be seen on the anterior surface. The extensive smooth-walled sinus in the pedicel extends well into the horncore base. The coronal suture is hardly indented. The preserved pieces suggest that the craniofacial angle was large for A lcelaphini . Left horncore WEE-5/10 c losely resembles ARA-8/3 in nearly all these features, but it is larger and longer with Jess of a protruding posterolateral keel although it also has an approach to a sharp edge lateral to a flattened posterior basal surface (Figure 2a); its anterior surface is less markedly concave; the ang le of horncore to braincase may be higher; the basal sinus is a bit more extensive (ca. lOmrn into the horncore base) w ith thinner surrounding bone. Right and left horncores MAK-1 /32 (Figure 4b, c) compare closely with the previous specimens. In the large extent of the basal sinus and the thinness of the surrounding bone MAK-1/32 resembles WEE-5/10 more; in basal horncore cross-section (Figure 2b) it is closer to ARA-8/3 by its lesser anteroposterior depth than in WEE-5/10 and in the more pronounced posterolateral approach to a keel. Figure 3 shows the best of a few dentitions from Aramis, and Figure 17a and b two mandibles from Maka. The single right P 4 from Aramis (Table 11) is from a dentition in advanced wear. Yet its paraconid and metaconid had fused only recently, forming a simpler central enamel cavity (interior to the fusion) than in later alcelaphines. (When I mention 'simpler' central enamel cavities, in distinction to ' more complex' ones, I mean less centrally constricted, or 'dumbbell-shaped ' .) It is rather square in shape with the rear part less reduced than in later alcelaphines, with a thin transversely directed valley between endostylid and entoconid. The left maxilla with P3-M 2 , ARA-8/9, and the lower molars ARA-4/ 1 (Figure 3) are less advanced than those of later alcelaphines in having J ess ' pinching' of the lateral lobes of lower molars and medial lobes of upper molars (that is, anterior and posterior constrictions of the rounded parts of the lobes), and s impler central enamel cav ities. Similar observations apply to the cf. (Damalops) dentitions from Maka and Wee-ee which I cannot meaningfully distinguish from the small Aramis sample . Of four avai lable P 4 from Maka and Wee-ee, only one has fusion of paraconid with metaconid, while the other three are only close to fusion. 138 Comparisons: This species is present at Kakesio dated ca. 4.4 m.y. (M. D. Leakey pers. comm.), as is Aramis. Frontlet KK/82 270 has basal horncore dimensions (49.5/44.0mm) slightly larger than ARA- 8/3, and comparable to the Laetoli LIT233.'59 and Hadar SH horncores (Figure 19). KK/82 270 has a basal horncore cross-section like that of ARA-8/3 except that the posterolateral keel-like ridge is less sharp and more posteriorly directed; a wide braincase (ca. 90mm) with the supraoccipital far forward (occiput is missing, the suture is present); horncores laterally lower than medially with a divergence of 30-40°, and an angle to the braincase of ca. 90°; a craniofacial angle of ca. 1 05°; an extensive sinus in the pedicel and basal horncore; a hardly indented coronal suture; and no parietal boss. All comparable features attest to a very close relationship between the Aramis and Kakesio forms (an impression that is supported also by a comparison of the scant dental samples), and further suggest conspecificity with the Maka and Wee-ee forms. The Lokochot (3.5-3 .36 m.y. ) and Tulu Bor (3.36-2.68 m.y.) specimens of this lineage include a part of the sample called Parmularius cf. angusti­ cornis by Harris (I 991: e.g., crania KNM-ER 921, locality unknown, and KNM-ER 4680 labelled as from Tulu Bor on pp. 208-209, but listed as from Lokochot on p. 299). Previous to my seeing this material, A. W. Ge ntry (pers. comm.) pointed out that it is related to Damalops. (The characters that suggest (Damalops) rather than P. angusticornis are partly reflected by the code differences between these taxa, Dos and Pan, in Table 14.) In some respects these crania are intermediate between the SH and later DD/KH fossils from Hadar. Their high horncore-to-braincase and parietal-occipital angles resemble some DD spec imens. The Koobi Fora horncores resemble ARA-8/3, MAK-l/32 and WEE­ SilO in compression, and in size KNM-ER 4680 is close to MAK-1/32, and KNM-ER 921 to WEE-5/ 10. Many other features of these specimens compare closely with those of other fossil s of (D.) "sidihakomai", including the lack of horncore torsion, the basal horncore cross-section, and cranial features. This lineage is also present at Kanapoi dated 4.1 m.y. (Leakey et. a!. 1995). For example, frontlet KP-71 is similar to ARA-8/3 and other (D .) "sidihakomai" in degree of horncore divergence, a basal horncore that is flattened posteriorly and with a lower lateral than medial margin, basal horncore separation, huge basal sinuses, a slight anterior concavity of the homcore in lateral view, absence of tors ion, and horn core length. It differs from Aramis in a higher angle of horncores to braincase, in which respect it also resembles Hadar DD more closely than Hadar SH. Its relatively high brain width is a resemblance to Kakes io that cannot be assessed on the Awash fossils. In the Hadar sequence, there is subs tantial evolutionary change from the earlier SH to the later DD and KH Members (see also Gentry 1980, 1981). The later (D.) " denendorai" has horncores that are more uprightly inserted, basally larger and on average more compressed (Figure 19), and relatively shorter; a lower craniofacial angle (Figure 18) and higher parietal-occipital angle; an occipital surface facing more directly backwards instead of laterally; and wider, shorter braincase and basioccipital. The horncores also change from a straighter and distally less divergent course without torsion to a more curved and di sta ll y more dive rgent one w ith clockwise torsion on the right at least in many of the later fossils. (However, detailed study is needed to determine whether the absence or presence of torsion may be variable in each of "sidihakomai" and "denendorai", with the latter only having a higher incidence of torsion. This would necessitate changes of codes 0 for "sidihakomai" and 1 for "denendorai" in Table 14, which might have the cladistic result that tors ion does not need to be secondarily lost in D. palaeindicus.) Gentry (1981: 12) also noted that the junction between the base of the nuchal crest and the back of the zygomatic arch is more anterior in the earlier forms, and that horncore divergence in the early SH skull AL 208-7 "does not increase continuously but changes fairly sharply at a point about one third of the distance above the base. Other horn cores from SH are curved, however, and it can only be noted that no re latively straight horn cores occur above SH." Many Hadar SH horncores are almost entirely straight except for a gentle anterior concavity, while the DD horncores are less straight. The fairly sharp change "at a point about one third of the distance above the base" that Gentry noted in AL 208-7 is absent in the Aramis and W ee-ee horncores, while the Maka horncores are broken too near the base to be certain. However, there are SH (or cf. SH) horncores that resemble the Aramis and Wee-ee ones very closely in being almost straight except for a gently concave shape anteriorly, such as AL 349-3 (Figure 4). In general, the Aramis and Wee-ee horncores resemble most the relatively straight SH versions, and not the DD horncores that s how torsion or/and more curvature and tend to be more compressed basally as well as much larger. The horncores from Upper Ndolanya (Parestigorgon gadjingeri of Dietrich 1950) and Shungura B 11 differ from the Awash fossils in having greater compression and a hint of clockwise torsion like (D. ) "denendorai" to which they may belong. The horncores of D. palaeindicus are much longer, more compressed and backbent than all those from the Awash, with smaller bases than those from Maka and Wee-ee (Figure 19). Cranium LIT 233.'59 either from the Laetoli Beds or a later stratum was discussed by Gentry (1980) who suggested that it may be conspecific with the 139 TABLE4. Measurements ofA lcelaphus buselaphus crania BOD-1/20, KL270-1 and KL284-1 with associated pala te KL8-1a, both cf. Bodo 1; Numidocapra crassicornis cra nia BOU-1/21, BOU-1/31, and subadult horncores BOU-6/1; ovibovine Nitidarcus asfawi gen. et sp . nov. cranium BOU-119. Length in mm; two values for horncore length are preser ved/ estima ted complete lengths; e =estima ted, ee =very rough estimate; max.= maximum; min.= minimum; a nt.= anterior; post. = posterior. Alcelaphus buselaphus Numidocapra crassicornis Nitidarcus BOD KL284- 1_ KL BOU- BOU- BOU- asfawi 1/20 /KL8-la 270- 1 1/2 1 1/3 1 6/1 BOU- 1/9 Skull length: premaxilla to post. occ ipital condyle 373e Horncore max. diameter ( I) 66.7 78e 60ee 58.2 61.7 40.9 6 1.6 Horncore min. diameter (2) 50.4 72.5e 50ee 47e 46.2 34.7 45.7 Horncore ratio (2)/( I) 0.76 0.93e 0.81 0.75 0.85 0.74 Horncore length 3 10/390 440/480 180(? 325/330 275/275 290/300 Horncore basal separation 26e 25e 20e 25e 45.0 Angle of basal horncore divergence 100°e ll 0°e 20°e 50° Angle horncore to braincase 90°e 140° 145° Angle of ( I) to midfrontal suture 90°e 90°e 20°e 40° Width across horn pedicels 104.0 106e 112.4 113.5 Craniofacial angle 780 goo 85o 95°e Parietal-occipital angle 150° Max. separat ion supra- orbita l foram ina (SOF) 60.0 60ee 66e 69.7 Distance of SOF to anterior horncore midpoint 80e 50.0 Distance across superior orbita l margins 150e 152ee 148ee 163.0 Distance orbit to horncore 60e 44.5 Bra inca e width at parie tal - squamosal suture 82.8 96.0 107.0 98.0 Braincase length: coronal suture to occiput 33.5 52e 68.7 75.8 63.2 Distance across mastoid exposures 108e 130ee 126ee Occipital he ight: top foramen magnum to occiput 44.5 46.5 58.2 Min. separation temporal lines on dorsal braincase 34.0 37e 34e 61.0 Basioccipital width across anterior tuberosities 26.0 3 1.8 Basioccipita l width across posterior tuberosities 34.2 38e 36e 47.6 Basioccipital length: ant. to post. tuberosities 30ee 3l e 35e Distance post. M3 - occiput 190e Length of nasal bones 186e Width across nasal bones 29.8 23.5e Width across ant. premaxilla 53.0 Maxi llary breadth at M3 78.9 Length P2"" 33.6 Length P3•4 29.0 Length/breadth P2 9.7/9.6 Length/breadth P3 11.4/10.6 14.4/12.2 Length/breadth£>-' 12.0/11.5 13.3/13.5 Length M 1•3 55.8 62.5e 74.5 Length/breadth M1 16.5/14.5 22.2/15.5 Length/breadth M2 19e/15e 23e/19.5e 25.8/16. 1 Length/breadth M3 20.7/15.7 22.3/14.4 26. 1/ 14.6 Length P2-M3 87.0 Length P3-M3 102.65 140 Hadar lineage and closely related to D. palaeindicus. LIT 233. ' 59 is c learl y c loser to Hadar (D .) "sidihakomai" than to (D .) "denendorai" in the fo llowing: the absence of horncore tors ion , a narrower braincase relative to length, a longer thinner basioccipital with anterior tuberosities facing more ne arly anteroposteriorl y and not splayed out posteriorly as in the Hadar DD form, absence of localized frontal raising between horncore bases in the form of a ridge, and a substantially lower parietal­ occipital angle. LIT 233.'59 does show some advance towards (D.) "denendorai" in the more uprig ht insertion of the horncores relative to the braincase than in Hadar SH fossils, and in a slightly shorter braincase. It is possible that there was advance within (D. ) "sidihakomai" from more nearly straight horncores before 3.4 m.y. (as at Aramis, Kakesio and Kanapoi) to the beginning within the SH Member of increased variation to include more distal horncore divergence that starts fairly abruptly some distance above the base, and more rapid tapering towards the tip. LIT 233.'59 does have these tendencies and thus resembles some SH horncores while the H adar DD form is even more advanced in these respects. The Aramis teeth are much more advanced than the 5-4 m. y .-old Lange baanweg dentitions of Damalacra (Gentry 1980) in more rounded medial lobes of upper and lateral lobes of lower molars, thinner more complex central enamel cavities and more outbowed walls between stylids on lower molars. The gap in tooth advancement between Langebaanweg and Aramis is very large compared to that between Aramis and living alcelaphines given the short time between the two fossil localities. The lack of comple ted fusion of paraconid wi th metaconid in three of four P 4 from Maka and Wee­ ee compares as follows with the Hadar sample: of two from Hadar SH one was fused in middle wear and one only by late wear, while the DD sample of six appears more advanced in that five were fused by middle wear. The Hadar teeth from DD are more advanced than those from SH in other respects as well , such as in greater reduction and earlier fusion of the posterior parts of P 4 , more enlarged posterior lobes (entostylids) on M3, and in generally more complex molar morpho logy. The Maka dentitions are closer to those from Hadar SH (e.g. , compare mandibles MAK-1 /76 with AL40 1-7 from SH and AL358-8 from DD). The incidence of P 2 can be assessed on two Hadar DD spec imens in middle wear (Gentry 1981) and on two Awash specimens. P 2 was absent in both mandibles from Hadar DD. It was present in one mandible in middle wear, and absent in a subadult one, from Awash. Figure 5 Partial c rania from Bouri 1 in anterior view: a. Numidocapra crassicornis BOU- 1/3 1, and b. holotype of Nitidarcus asfawi BOU-1/9. 141 l~ a. Figure 6. Par1ial crania from Bouri I in right lateral view: a.Numidocapra crassicornis BOU-I /3 1, and b. holotype of Nitidarcus asfawi BOU- 1/9. Genus Numidocapra Arambourg 1949 Type species Numidocapra crassico"rnis Arambourg 1949 Generic diagnosis: As for the single species. Numidocapra crassicornis Arambourg 1949 Diagnosis: Medium to large alcelaphines with skulls nearer to high and narrow than to low and wide; f rontals between horncore bases strongly raised ; horncores have an oval cross-section and are more massive, longer, and often basally more strongly compressed and less basally and distally divergent, with weaker clockwise torsion from the base up on the right side, than in Rabaticeras. The horncores pass forwards and upwards from the base such that their basal to middle parts appear gently concave­ forward in lateral view; basal horncores with low divergence and inserted closely, with small posterior angles of their max imum diameters to the midfrontal suture, and very uprightly with a large angle to the dorsal braincase; the frontal surface anterior to the horncores is not convex, but slightly concave with a shallow valley towards the midfrontal s uture; supraorbital foramina widely separated; keels and basal sw e llings a re a bsent on horncores and transverse ridges a bsent to weak; pa rie tofrontal suture straight to gently indented anteriorly; dorsal orbital rims project less strongly than in Rabaticeras; postcornual fossa small to absent; braincase strongly angled relative to the face with s ides parallel to slightly narrower towards the rear; no parietal boss; moderately large, somewhat shallow preorbital fossa that lacks definite rims; the nasal bones are narrow; occ ipital s urface mainly facing backward s, narrowing a bit dorsally, with a slight vertical ridge. The teeth are large relative to skull size among alcelaphines. The occlusal morphology is quite advanced with large fairly complex central cavities and pinched buccal and lingual Jobes. The premolar row is reduced. (Based on Arambourg 1949, 1979, with additions from the new Bouri fossils.) Numidocapra crassicornis from Bouri 1 and 6 (Figures 5, 6, 17e; Tables 4, 11 , 12): There are three crania with horncores from Bouri Localities 1 and 6 dated ca. 1 m.y. : BOU-1/21 includes two pieces that join securely: a posterior cranium with both basal horncores, and an anterior piece with full maxillary dentition. BOU-1/31 is a dorsal cranium with complete right and most of the left horncore, and the subadult BOU-6/1 includes two complete horncores and fragments of the frontlet and braincase. These specimens, where the features are preserved, essentially support the diagnosis. The re is variation: The unattached horncores of BOU-6/1, probably juvenile, show slight indications of diagonal ridges on the convex surfaces and 142 stronger ones towards the concave surfaces. BOU- 1/21 has the basioccipital encrusted in matrix , but one can see that it is somewhat rectangular (perhaps more so than on one R. arambourgi cranium from Elandsfontein) and has a longitudinal groove with flanking ridges as usual in Alcelaphini. Its occipital surface has slightly laterally-facing components and is dorsally less evenly rounded and a bit naiTower than in some other alcelaphines. Its auditory bulla is somewhat thin and little inflated. The anterior piece of BOU-1/21 shows naiTow posterior nasals; and a medium-sized preorbital fossa, rather shallow and lacking definite flanking rims, that is deepest over the M2/M3 junction. The dentition (Figure 17e) is the first known to be definitely associated with Numidocapra. A few other dentitions from Bouri have been assigned to this taxon (Tables 11 , 12). The teeth are close in s ize to the smalles t Connochaetes taurinus and the largest Alcelaphus buselaphus that I measured. They are large relative to skull s ize among alcelaphines. The occlusal morphology is quite advanced with large fairly complex central cavities and buccal and lingual lobes that are nanowed reminiscent of ovibovines. P2 was absent in life. Comparisons: N. crassicornis was described from Ain Hanech , Algeria, (Arambourg 1949), and is also known from Anabo Koma, Djibouti, Ethiopia (Bonis et al. 1988). Geraads (1981) considered it alcelaphine and Gentry (1978, 1990a) as a caprine. The larger Bouri specimens BOU-21 and BOU-3 1 bear a close resemblance in morphology to N. crassicornis from Ain Hanech (compare Figures 5 and 6 with Arambourg's, 1979, Plate 38: 4) and Anabo Koma . Similarities include compressed hornc ore bases of oval cross-section with a maximum diameter near-parallel to the midfrontal suture; very strong craniofacial angle and very high angle of basal horncore to braincase; comparable horncore length in relation to girth; and very low basal horncore separation and divergence. In fact, every feature cited in Arambourg (1979) and Bonis et a/. (1988) seems to be present also at Bouri, but there appears to be a trend of decreasing size in this lineage from Ain Hanech (which by biochronology may be ca. 1.8-1.7 m.y. old) to Anabo Koma (ca. 1.6 m.y. old, Bonis et al., 1988) to Bouri at ca. 1 m.y. (Figure 19). The most closely related species, Rabaticeras arambourgi, is smaller than the Bouri form and occurs in roughly coeval to later strata, thus continuing the size trend if it is indeed descended from Numidocapra . The Bouri form is larger than R. arambourgi from Rabat and Olduvai Beds III-IV and overlaps only ~ith the largest Elandsfontein fossils. Gentry and Gentry (1978) proposed that Rabaticeras gave rise- to Alcelaphus, which would further continue the same lineage, but this was questioned by Bonis et a/. (1988). I return to this debate in the next section in which I describe the earliest known Alcelaphus from Bodo, ca. 0.6 m.y. in age. In addition to smaller size, the main differences of R. arambourgi from the Bouri Numidocapra are horncores that are averagely less strongly compressed, more widely separated and more divergent, with stronger clockwise torsion from the base up on the right side; a hardly marked postcornual fossa; parietals shorter especially in relation to braincase width; probably closer supraorbital foramina; and more strongly projecting dorsal orbital rims. Gentry and Gentry (1978: 417) considered four specimens from Olduvai Bed II, SHK II 1953.280, SHK II surface 1957.92, SHK IT 1953.234, and BK II East 1953.067/5460, as similar to R. arambourgi yet abeiTant in comparison with it: " One [alternative], which we favour, is to regard them as a variant within the R. arambourgi lineage, and another is to regard them as possibly linked with ... Numidocapra crassicornis Arambourg (1949: 290) from Ain Hanech ... If the Olduvai horn cores are related to Numidocapra, it would be good to know whether they should be classified as Caprinae similar to Procamptoceras brevicornis ... or as Alcelaphini." I suggest assignation of these Olduvai specimens to N. crassicornis. BOU- 1/31 and BOU-1/21 have strong resemblances to them: a size as large as the largest Olduvai specimen and high compression in BOU- 1/3 1; weak torsion and forward curvature, although the Olduvai horncores may be even less curved. The Olduvai Bed IT levels SHK and BK are dated ca. 1.7- 1.4 m.y., comparable in age to Ain Hanech and Anabo Koma. On Gentry and Gentry 's (1978) suggestion that Numidocapra m ay be caprine, the dentition of BOU-1/21 (Figure 17) does not support membership of Caprini. The central cavities are too complicated, probably too wide open, and the dentition is large relative to other skull measures whereas Caprini have relatively small teeth. The teeth resemble those of ovibovines in upper molars that are long relative to width and have pointed medial lobes; but the premolar row is too short, and the central cavities are not narrow enough for Ovibovini, quite apart from the numerous differences in horncore and skull morphology. The caprine Procamptoceras brevicornis from Seneze (Schaub 1923) differs decis ively from the Bouri form: apart from being very much smaller (with basal horncore size that compares with the smallest Alcelaphini in Figure 19), it has no horncore torsion and much less distal divergence, hollows more than half-way up the horncore of which there is no sign at Bouri; and a much lower angle of the basal horncore to the braincase. (R.) porrocornutus from Swartkrans resembles the Bouri fossils in strong basa l horncore compression (shared with BOU-1/3 1) and widely separated supraorbital foramina. It differs in higher separation between horncore bases and between these and the supraorbital foramina, more divergent 143 a . [~ b. [~ F igure 7. Alce/aphus buse/aphus cranium BOD- 1/20 from Bodo I: anterior (a.) and palatal (b.) v iews. horncores with strong and closely spaced transverse ridges , more prominent orbits, and in a higher craniofacial angle and a lower angle of the horncore to the cranium. If Numidocapra and Rabaticeras are ancestor and descendant in their own monophyletic group separate from Alcelaphus, then the correct name for thi s group is Numidocapra Arambourg 1949 . I suggest that nomenclatural changes at the generic level will be better made with additional information. It is possible that additional support emerges for a single ancestral-descendant lineage from Numido­ capra to Rabaticeras to Alcelaphus. In that case the name of the group would be Alcelaphus with the other two generic nam es as synonym s. Also, a generic rev is ion would be be tte r based on a cladogram that includes N. crassicornis and with a branc hing seque nce be tte r r esolved tha n the polytomy at node 18 in Figure 21. It is interesting that the largest form in this proposed lineage was widespread at some time in t he interval 1.8- 1.4 m .y ., imply ing continuity between the North African and Somalian-eastern Ethiopian biotas and the ir extension southwards to include at least the Olduvai area. In contrast, by the time of Bouri at 1 m .y. and some time within the interval of Olduvai Beds ill-IV (ca. 1.3-0.7 m.y.) , the Olduvai re presentative was a smalle r a nd c hanged descend ant w hile contemporane ous Numidocapra still persisted in eastern Ethiopia. T his may indicate the effects of global cooling during the earlier time, and of later warming that affected Olduvai while cooler conditions persisted at higher latitudes and altitudes in the Awash area. Genus Alcelaphus De Blainville, 18 16 1775 Bubalis Frisch 1820 Bubalis Goldfuss 1827 Damalis H. Smith 1827 Acronotus H. Smith 1 9 12 Sigmoceros Heller 1979 Sigmoceros Yrba T ype species Alcelaphus buselaphus (Pallas 1766) Generic diagn osis: Medium to la rge s ized a lcelaphines w ith e longated c ra nia that a re moderately to very high and of low to medium width ; horncores inserted far behind the orbits on a moderate ly to ver y long boss that results from backward protrusion of the frontal bone and possibly also from pedicel fusion ; horncores with posterior fl attening and with a large angle to the midfrontal suture ; moderately long ho rncores that vary in compression and have strong clockwise torsion from the base up on the right; horncore tips recurving to point downwards in mature indiv iduals; di stance from pos te rior horncores to occiput s hort to extremely short; small craniofacial angle and large parietal-occipital angle; preorbital fossae of moderate s ize. 144 Alcelaphus buselaphus (Pallas 1766) (Includes many subspecies; see Table 2 for those to which I refer; Ansell (1971) for the full list.) Diagnosis: Differences from the other living con­ generic spec ies A. lichtensteini are: The basa l horncores are less separated. Their divergence may be as high to lower and is therefore more variable. They vary a lso in compress ion from form s resembling A. Lichtensteini to forms with nearly round cross-sections. The horncores are m ounted on a more elongated and narrower frontal boss and are more di stant from the supraorbital foramina . The skull is narrower especially in the orbital region. A rai sed boss on the · midfrontal s uture of the forehead, as present in A. lichtensteini, is absent. ALceLaphus buseLaphus (Pallas 1766) from Bodo 1 (Figure 7, Table 4): BOD-1/20 is an almost complete cranium from Bodo 1, ca. 0.6 m .y. old. Crania KL284- l /KL8- 1a and KL270-1 are of unknown provenance but probably also from Bodo 1 based on stratigraphic criteria. BOD-1/20 is a high narrow skull; face long, narrow with a long frontal boss and lacking the central boss on the forehead of A. lichtensteini; part of a shallow postcornual fossa can be seen on the right; toothrow substantially anterior to the orbit; a long diastema; an extensive nasa l- maxillary contact; ante rior zygomatic arch with a strong deep ridge; elongated thin nasals with a central indentation of the fronto­ nasal suture; horncores with bases flatte ned posteriorly (Figure 2c) , marked clockwise tors ion on the right, strong transverse ridges, strong basal divergence with distal parts reapproaching and with a downward component; homcores surprisingly long relative to basal size; oval and moderately s ized foramen ovale; the spaces for the missing auditory bullae look fairly large; the basioccipital has a longitudinal groove with flanking ridges; occipital surfaces face partly laterally on ei ther side of a moderately marked median vertical ridge; mastoids are large; a strongly developed preorbital fossa with superior and ventral rims; palate strongly domed upwards in the centre, with m edian pa lata l indentation far forward of lateral ones and narrowed to a point; the uppe r teeth not noticeably less advanced than in living AlceLaphus with comparably long premolar rows with P2 present. KL284- l is a frontlet, with an adjoining ventral cranial piece, of a large hartebeest. It was encrusted with matrix at the time of study. It has fairly long, nearly complete horncores with little basal but stronger distal compression, strong clockwise torsion and basal divergence, curved back down in lateral view and slightly outwards in anterior view towards the tips, transverse ridges that are very faint near the base but marked over the curved anterior surfaces, and large smooth-walled basal hollows that extend some 20-30 mm into the horncore. The basioccipital looks typically alcelaphine. The mastoid is entirely on the occipital surface. The piece KL8- I a appears to belong to the same skull as KL284-l. It includes the left anterior orbit, part of the nose, and a partial palate and dentition (Table 4) that resemble those of BOD-1 /20. The preorbital fossa inc ludes a fairly deep anterior part. Another fossil of this species is KL270-l, a well-preserved uperior cranium with a basal left horncore and orbital margin, a substantial f ronta l boss, a huge parietal-occipital angle, a nd mastoids all on the occipital surface. It strongly resembles BOD-1 /20 , as does ano the r cranium lack ing horncores and w ith partial dentition, KL237 -1, which was too encrusted with matrix for measure me nt. These Bodo foss il s are the most complete and earliest of ALcelaphus. I will continue their description by a detailed comparison with living A. buseLaphus and A. lichtensteini. Figure 19 shows the samples of the two liv ing species and subspecies that I meas ured . I wi 11 r efer to the va ri e ti es of A. buselaphus only by the ir subspecific names tora, swaynei, cokii, jacksoni, Leiwe!, major, and caama. Ruxton and Schwarz (1929) gave data on another subspecies that recently became extinct, the North African A. b. buseLaphus. They regarded caama as a separate species, and the tora-swaynei-cokii group as more plesiomorphic than other A. buselaphus in their weaker, less twi s ted horns on a s horte r , narrower f rontal boss, but as more advanced in high horncore divergence. KL284-1 is much larger than BOD- 1/20 and KL270- l , although the first two both have adult dentitions. This suggests that KL284-1 is male and the other two female, and I will refer to the fossil s as such. The basal skull length of the female BOD-1/20 is closest to female means of tora, swaynei, cokii, Leiwe!, caama, A. Lichtensteini, shorter than female means for jacksoni and major, and longer than values for one female and one male of the extinct buseLaphus. Space constraints in Figure 19 did not allow me to show all individual points for hartebeests and information on sex. In a plot of all the data, BOD- l/20 and KL270-l fall into a cluster of mostly female living hartebeests, and lie very close to swaynei, tora and A. lichtensteini females. KL284-l is comparable in size to the larges t horncores in A. buselaphus, namely of caama, jacksoni, and major males. (My comparative sample included no males of tora and swaynei which are today geographically closest to the Awash, the Tora Hartebeest in the Blue Nile area and Sway ne's Harte beest in easte rn Ethiopia and Somalia.) Horncores in caama, jacksoni, and perhaps major are very little compressed and show a growth gradient towards lower compression with increasing size (Figure 19). Horncores in swaynei, cokii, and especially tora and A. Lichtensteini, are more compressed. The sample of five A. lichtensteini s uggests a growth gradie nt towa rds grea ter compression with increasing size. The Bodo females are very close to females of tara, swaynei and A. /ichtensteini, w hile the Bodo m a le i s less compressed and close to males of cokii and major. That is, the growth gradient of the Bodo form is towards less compression at larger size as in A. buselaphus. I compared the ratio of basal horncore size to tooth s ize in the Bodo sample with those in A. /ichtensteini, the subspecies jacksoni, -and other living and extinct Alcelaphini. The ratio is excep­ tionally large in the Bodo form and closest to that of A lichtensteini: (maximum horncore di a mete r)/ (length M 1-3) for BOD-1/20 is 1.20, for KL284- l it is 1.25, for an A. lichtensteini skull it is 1.31 , while for jacksoni it i s near 1.0, and for all othe r alcelaphines even lower. Basal horncore separation and divergence is higher in A. lichtensteini than in any A. buselaphus, with tara, swaynei, and cokii the next highest. The Bodo fossi ls group with subspecies of A. buselaphus and fa ll decisively below A. lichtensteini (Figure 20). Surprisingly, for divergence the Bodo crania group closely with A. lichtensteini and with the highest cokii, be ing a bit above the single tara and others. This. early hartebeest from Bodo already had a strongly bent braincase relative to the face (as extreme as in livi ng A. buselaphus and A. lichtensteini), and in its large basal horncore­ midfrontal angle it resembles A. buselaphus closely and may be less advanced than A. lichtensteini which has an even larger one (Figures 2, 18). In hartebees ts, the angle between horncores and cranium increases with increasing horn size so that females tend to have less upright horncores than males. This angle in the female BOD-1/20 is closest to those of tara, cokii, and small A. lichtensteini, and substantially lower than those of other subspecies of A. buselaphus. In its moderately widely separation of the supraorbital foramina, BOD-1/20 lies above the caama mean, and close to other A. buselaphus and to the lowest A. lichtensteini. In the lengthening of the frontal boss, as measured by the distance from supraorbital foramina to basal horncores, the Bodo female is higher than swaynei and cokii, as advanced as tara and other A. buselaphus and above even the largest values in A. lichtensteini. In the width of its frontal boss, or fused pedicels, the Bodo female is close to tara and to the jacksoni and caama means, above females of cokii, swaynei , jacksoni and caama, and substantially below female A. lichtensteini, while the Bodo male has a boss as wide as the largest A. lichtensteini. Breadth across the mastoids and occipital height in the Bodo fossils are lowe r than in A. lichtensteini and more comparable to those in A. buselaphus. In sum, in all respects the Bodo form resembles one or more of the tora-swaynei-cokii group, the most ples iomorphic group in A. buselaphus. In some of these features it is clearly less advanced than the specialized j acksoni-major-caama group. It has parti c ularly c lose resemblances to the Tora 145 Hartebeest that today li ves not far from the Awash area, and some of these are also resemblances to A. lichtensteini: shorter basal skull le ngth, horncores that are more compressed and possibly also longer, higher basal diverge nce and separation, and the lower angle of homcore to cran ium. The Bodo form clearly differs from A. lichtensteini, and resembles A. buselaphus as a whole in its more slender and longer frontal boss, in lacking both a central raising on the forehead and strong basal expansion of homcores, and in its narrower braincase and mastoid region. Gentry (1990b) described the earliest secure A. lichtensteini specimen, a horncore from the Lower Terrace Complex of Semliki , Zaire, dated ca. 0 .5-0.3 m.y. He noted that it has a curvature in the middle part tha t is not o nl y upward as is typical of A/celaphus, but m ore forward than in li v ing A. buselaphus and resembling A. lichtensteini, and that the succeeding curve backwards in the more distal part of the fossil homcore is less abrupt than in either living species. The BOD-1/20 homcores in their middle parts show the less forward-curving course of living A. buselaphus, while their tips are less abruptly back-curved than in either living species, supporting G entry's (1990b) notion that the latter is plesiomorphic. The character codes for hartebeests in Table 14 were based on li ving A. buselaphus and A. lichtensteini, independe ntly of the conclusion that the Bodo form belongs to A . buselaphus. It is a reasonable inference that A. buselaphus at its origin did not yet have the high level of polymorphism that it has today and, second, that A. buselaphus from Bodo should most c losely resemble the common ancestor shared w ith A. lichtensteini and, third , that the features which Tora and Lichtenstein 's Harte bees ts share w ith tha t from Bo do, are pl es iomorphic retentions. Thi s hypothes is is supported for most of the characters by the present cladistic result and by comparison with the outgroups of the A. buselaphus-A. lichtensteini sister-group . These outgroups are the potential direct ancestor Rabaticeras arambourgi, (R .) " lemutai", (R. ) porro­ cornutus, and Damalops pa/aeindicus (Figure 22). The observed strong horncore compression in all these outgroups, even in some populations of R. arambourgi although thi s taxon is variable (Fig ure 19), s upports the notion that higher compression is ples iomorphic fo r the li v ing A. buselaphus and A. tichtensteini. The hypothesis that horncores of intermed iate length re lative to basal size are plesiomorphic for the entire group from node 14 in Figure 21 , with D . palaeindicus evolving increased and the jacksoni group decreased length , is consisten t with the evidence. Homcore le ngth is not known for (R .) " lemutai" and (R. ) porrocornutus , but for some R. arambourgi it is intermediate, comparable to Bodo, shorter than in D. pa/a eindicus and lo nger than in advanced A. buselaphus. Basal ho rncore divergence i s 146 intermediate in D. palaeindicus, (R .) " lemutai", (R.) porrocornutus and varies in R. arambourgi with some E landsfontein horncores being higher but sti ll lower than A. lichtensteini and the Bodo A. buselaphus. T herefore the p resent hypotheses, that the hig h di vergence of the last two forms is plesiomorphic and that R. arambourgi is ancestra l to them, require that higher divergence was attained by at least one late population of R . arambourgi as indicated by the ambiguous change at node 27 in F igure 2 1 (characte r 9: 1->2). If true, the n this population is likely to have been deri ved from a form such as the Elandsfontein R . arambourgi. Basal ho rn core separatio n is intermedia te in D. palaeindicus, (R .) porrocornutus and R . arambourgi (Figure 20). This is consistent with the hypothesis tha t ho rncore separation was pleisio mo rphica ll y inte rmedi ate a nd then na rrowed towards the advanced jacksoni group and widened towards A. lichtensteini. (In that case (R .) "lemutai" would have had to evolve independently a separation as high as in A. lichtensteini.) The angle of the basal horncore to the cranium was scored conservatively such that the score 1 was given only to taxa with the highest values (Table 14). D . palaeindicus, (R.) "lemutai" and (R .) porrocornutus are lower than R. arambourgi and the average A. buselaphus, which supports the notion that the primitive state before origin of R. arambourgi was indeed a lower ang le. A. lichtensteini has a ?-code as it varies from moderately high in the smallest adults to very high in the largest, and this allowed the cladistic outcome of a high angle as a synapomorphy of R . arambourgi and living hartebeests. A less conservative coding would have to acknowledge that R . arambourgi has a higher angle than even the highest A. buselaphus, namely the jacksoni group (and an even hig her angle tha n the o ther supposed descendants : the Bodo, Tora and Lichtenstein 's Hartebeests). T hus the hypothesis that a lower angle was plesiomorphic for living hartebeests does not fit the data without admitting homoplasy. Recently there has been increasing interest in our f ield in heterochrony- partic ularly in the kind of paedomorphos is known as neoteny and in its inverse, the form of peramorphosis called acceleration - in relation to phylogeny and climate. (Reviews and examples can be found in Vrba et a/. 1994; Vrba 1994, in press). He teroch ro ny inc ludes a ll evolutionary changes in the timing of appearance of characters during ontogeny, and in the rates of s hape a nd s ize developme nt. I n neo te ny the descendant adult, in spite of being of the same size or larger, retains phenotypic characters that appeared in a ncestral juvenile stages . In accele ratio n the descendant adult, in spite of retaining the ancestral size or being smaller, has characters that appear more advanced than the ancestral adult - or ' hyper­ adult '. O n ave rage co lder c l imates are often associated w ith larger bodies (Bergmann 's Rule) tha t retain neotenic compone nts; w hil e warme r environments correlate with smaller bodies and with features that evolved by acceleration. Thus, evolution in warmer climate by acceleration is expected to result in relative increases in features such as hom curvature, torsion and girth, angulation of the basal horncore to the midline, length of frontals and frontal bosses , and premolar length (all of which were re lative ly larger in t he ancestra l adu lt s t ha n juveniles), and decreases in other characters such as separation of horncores and supraorbital foramina (which generally become relative ly reduced with growth to maturity). T hese are precisely most of the changes that did evolve towards Alcelaphus (given descent from Rabaticeras). T hus acceleration in respo nse to occupation of increasingly warmer habitats provides a viable hypothesis for evolution of a lineage from Numidocapra to Rabaticeras to Alcelaphus. Size red uction also occurred within Numidocapra and between it and Rabaticeras and early Alcelaphus. However, one cannot explain by acceleration the evolution of less steep horncore insertions (with lower and higher angles of homcores to braincase and face respectively), that we observe in Alcelaphus if it evolved from Rabaticeras. As noted above, in hartebeests and many other bov ids steepness of horncore insertions tends to increase with maturation and growth to larger sizes. For this character in Alcelaphus one would need to invoke the simpler kind of heterochrony called hypo­ morphosis (the smaller descendant adult has the subadult shape of the ancestor at that size), evolving in mosaic fashion alongside the overall imprint of acceleration. Bonis et a/. ( 1988) noted the problem that the carriage of the head in Numidocapra and Rabaticeras would have differed from that in living Alcelaphus because the horns of the first two are more extremely angled to the braincase with lower basal angles to the face. I agree with that. Based on it, they suggested that R. arambourgi is not a suitable ancestor of living Alcelaphus. Nevertheless, I suggest th at G e ntry a nd Gentry's ( 1978) not io n t ha t Alcelaphus evolved from Rabaticeras remains viable. T h e re a re in fact models that p redict mosaic evolution in the same li neage, of different kinds of heterochrony for different characters depending on their ancestral growth profiles, as required in this case (e.g., Vrba 1994, in press). Genus 1925 1932 1953 1965 1984 Megalotragus Van Hoepen, 1932 Rhynotragus Reck: 451 Pelorocerus van Hoepen: 65 Lunatoceras Hoffman: 48 Xenocephalus Leakey: 62 Rusingoryx Pickford and Thomas: 445 Type species Megalotragus priscus (Broom 1909) Generic diagnosis : Gentry and Gentry ( L 978:356) gave the diagnosis : "Very large extinct alcelaphines, including the largest known, with narrow skull s and 147 TABLES. Measurements of Megalotragus kattwinkeli crania BOU-2/21, BOU-1/99, frontlet BOU-2/20, and horncores BOU-116 and MAT-1/13; and cranium of ?Oreonagori?Megalotragus sp. BOU-1/97. Length in mm; two values for horncore length are preserved and estimated complete lengths ; e =estimate, ee = very rough estimate; max. = maximum; min.= minimum. Horncore maximum diameter (I ) Horncore minimum diameter (2) Horncore length Horncore basal separation A ngle of horncore divergence Angle horncore to bra incase A ng le of ( I ) to midfrontal suture W idth across horn pedicels Craniofacial angle Parietal-occipital angle Maximum separation supra­ orbital foramina (SOF) D istance SOF to horncore D istance orbit to horncore Braincase wid th at parietal­ squamosal suture Braincase length: coronal suture to occiput M in. separation temporal lines Basioccipital w idth across anterior tuberosities Basioccipital w idth across posteri or tuberosities Basioccipital length : anterior to posterior tuberosities Length M 1•2 Length/breadth M 1 Length/breadth M2 Megalotragus kattwinkeli BOU- BOU- 2/21 1/99 68.2 76.6 53.0 61.'4 300/420 33.0 40° 60•e 70•e 11 5.5 II OOe 140- 160• IJ Oee 11 8e 83e 15e 58.5 27.5/ 19.7 3 le/? 40ee 106.5 86.7 horncores inserted oblique ly in side v iew , behind the level of the orbits and close together, with a torsion that is clockwise from the base upwards on the right side; molar teeth tending to have a simple occlusal patte rn ; very short premolar rows; long legs." The present analysis adds the following. The preorbital fossae are minimally developed to absent. T he distance between the coronal suture and the occ iput on the dorsal braincase is markedly reduced, while the occipital surface is relatively high and wide across the mastoids. The tendency to at least incipient development of horn pedicel fusion into a joint boss is present within each species. Although the nasal area of the face is unknown in M. priscus, an upward -doming of the posterior nasals and adjacent bones to form a crest probably characterizes the entire genus, as also concluded by Gentry et al. (1 995). Megalotragus katMinkeli (Schwarz 1932) 1925 Rhy no trag us semttt cus Reck: 451 , unnumbered figure 1932 A lcelaphus kattwinkeli Schwarz: 4, no f igure 1937 Gorgon taurinus semiticus Schwarz: 60, 85, in part I 9 65 Alcelaphus howardi Leakey: 60, Plate 79 1965 Xenocephalus robustus Leakey: 62, Plates 8 1 and 82 1965 Incertae sedis Leakey: 69 (d) in part 1976 Megalotragus ?kattwinkeli Gentry : 285 BOU- 2/20 90e 77e BOU- 1/6 98.3 80.0 MAT- 1/ 13 70.0 59.3 ?0./?M. sp. BOU- l /97 68.8 55.8 390/460 65e 5SOe 80•e 50•e 135.0 LI O• 130° 104.0 35e 75.2 37.5e 4 1e 45ee 1976 Megalotragus cf. M. kattwinkeli Harris: 298 1978 Megalotragus kattwinkeli Gentry and Gentry: 356, Plate 12:2, Plate 13 1985 Megalotragus sp. nov. Harris: 156 1991 Megalotragus isaaci Harris: 187, Figures 5.46 to 5.48 Diagnosis: An a lcelaphine that varies in size from large-medium to very large. Horncores short to mode rately long, w ith transverse ridges that are better developed a bove the base, insertions that vary from far behind the orbits to very far such that the ir bases overhang the occipital surface, with associated variation in the distance from horncore bases to occiput; horncore bases strongl y angled to the midfronta l suture, with compress ion that is mostly low at the base and increases towards the mid-horncore. Divergence above the horncore bases varies from low to moderate, with tips that have a greater tendency to reapproach and to curve inwards distally in the smaller individuals whereas in the larger individuals they curve less inwards and rather more upwards. In smaller individuals most of the basa l part of the horncore is anteriorly concave. Towards larger skull s izes there is an increasing tendency to strong backward c urvature of the horncore above a more upright basa l stem. The horncores have anterior basal swellings that are especially marked in larger horncores with well­ deve loped basal backward c urvature. There is a 148 ' a. Figure 8. Partial crania from Bouri L in anterior view: a. BOU-2/ 2 1 of Megalotragus kattwinkeli and b. BOU- L/97 of ?Oreonagor/?Megalotragus sp. variable te ndency, increasing in larger specimens, for the frontals between the horncores to be raised such that the pedicels appear fused into an incipient joint boss. The orbits project prominently relative to the narrow distance across the horn pedicels. A prominent nasal crest, formed by upward-doming of the posterior nasals and adjacent bones, is present. The cranium has an occipital surface that faces mainly backwards and only a little laterally and has a median vertical ridge, a wide basioccipital w ith large anterior tuberosities, and small , little inflated auditory bullae. Comments on synonymy and diagnoses: The nature and seque nce of di scoveries in thi s group has resulted in a complicated systematic history (Gentry et a/. 1995). The holotypes of Rhynotragus semiticus and Megalotragus kattwinkeli were collected during a 1913 expedition to O lduvai Gorge. Because it was tho ught that they had been destroyed during the second world war, Gentry and Gentry (1978) designated as neotype of M. kattwinkeli the skull BM (NH ) M21447. They w e re uns ure of the affinities of R. semiticus just as Reck (1925, 1935) had been. Pickford and Thomas (1984) described material from Late Quaternary strata on Rusinga Island in Ke nya as a new genus and species of Alcelaphini , Rusingoryx atopocranion . They did not mention any possible affin ity of this form with Megalotragus, just as by that date no one had yet suspected any affinity between Rhynotragus and Megalotragus. It was not until 1991 that the close re la tions hip between these three taxa was recognized. Harris (1991) described new fossil s from Koobi Fora as a new species M. isaaci. He argued that Rusingoryx is a junior synonym of Megalotragus, and that M. atopocranion is a species separate from other members of the genus. My own analyses strongly support both of these conclusions (see also Gentry 1990a; and Gentry et al. 1995). Pickford and Thomas ( 1984) thought that some features of atopocranion are unique among bovids, notably the very large craniofacial angle. I agree with Harris (1991) that this angle only appears so large because the nasals and anterior frontals are strong ly domed upwards in a nasal crest; and that this nasal crest in atopocranion is homologous with the inflated nasal region that he described in M. isaaci and that Reck (1935) noted in R . semiticus from Olduvai. Recentl y Gentry et a/. ( 1995) re po rte d the discovery, during 1992 in the U niversitats-Institut fur Palaonto log ie und His torische Geo logie in Munich, of many of the bovid fossils from the 1913 Reck expedition to Olduvai Gorge, including the holotypes of R. semiticus and M. kattwinkeli. They reported that a horncore listed by Schwarz (1937) as Gorgon taurinus fits exactly onto the R. semiticus holotype, and confirmed that R. semiticus is conspecific with M. kattwinkeli. With this astonishing discovery, the name Rhynotragus became the senior generic synonym for Megalotragus, and R. semiticus the senior specific synonym for M. kattwinkeli. A pe tition to the Inte rnatio na l Commiss ion on Zoological Nomenclature from A. W. and A. Gentry to conserve the use of the names Megalotragus and M . kattwinkeli, which have been muc h used in recent years, is pending. The new foss il s of Megalotrag us from Bouri confirm many essential conclusions of Harris (1991 ) and Gentry et al. (1995). Yet they do add yet a new perspective. Harris (199 1) argued that the Koobi Fora form is a new species, M. isaaci, separate from M. kattwinkeli but conspeciftc with R. semiticus. In partial contrast, Gentry et al. (1995) regarded R. semiticus as a synonym of M. kattwinkeli as known from O lduvai Beds II to IV and elsewhe re, and accepted M. isaaci as a separate species. I shall arg ue be low that the combina tion of features preserved in the new Bouri fossil s, in the context of all the other evidence, suggests that all three belong to a single variable species, namely that both R. semiticus and M. isaaci are synony ms of M . kattwinke li . Thus my diag nosis refers both to 149 10 em Figure 9. Partial crania from Bouri I: a. BOU-2/21 of Megalorragus katMinkeli in left lateral view, and BOU-1 /97 of ?Oreonagor/?Megalotragus sp. in right lateral view. characters cited in Gentry and Gentry 's ( 1978:356) diagnosis of M. kattwinkeli, and to charac ters in Harris's (199 1:187) diagnosis forM. isaaci, and differs from both in being the diagnosis of a more variable species. Megalotragus katn-vinkeli from Bouri l -2 and Mataba ietu 1 (Figures 8, 9, Table 5): Bouri 1 and 2, ca. 1 m. y. old , have yielded four cranial fossi ls of Megalotragus: c rania BOU-2/2 1 and BOU- 1/99, frontlet BOU-2/20, and horncore BOU-1/6. BOU-2/2 1 (Figures 8, 9) includes a large part of the face up to parts of both orbital rims and much of the nasal and max illa bones, most of the left and the basal part of the ri ght horncore, the dorsa l braincase, and some associated teeth. T he ho rncore bases a re compressed , w ith some posteromedial flattening, and not particularly swollen ante rio rl y . The hornc ores are quite long, near 420 mm in life, and have definite clockwise torsion on the right side. They are inserted very far behind the orbits and close to the occiput with their major axes at a fairly strong angle to the midfrontal suture (the posterior angle be ing ca. 70°), and at a moderate angle to the bra incase. They pass g raduall y backwards above the insertion such that the bases are moderately convex anteriorly, then straighten out in their mid-secti ons, and recurve gently upwards towards the tips. They are inserted closely together, with their latera l basal margins much lower than their medial ones, and with divergence that is basally low, increases higher up, and lessens slightl y towards the ti ps. Fairly well-marked transverse ridges are evident anteriorly above the basal third of the horncore. The frontals between the horncores are raised, and from the back this raising can be seen to be loca lized underneath the horncores such that the pedicels appear fused into an incipient joint boss. The orbits project outwards quite prominently relative to the narrow diameter across the horn pedicels. There is a small supraorbita l foramen on the right set almost flush with the slightly convex frontal, with a long groove anterior to it, and notably distant from the midline and the horncores. A long, shallow, thin postcornual fossa can be seen on the left. Near the anterior orbit, the posterior nasals and adjacent bones start to dome upwards strongly to form a prominent, rounded nasal crest that extends forward on the face. E nough of the lateral face is preserved to de te rmine that there was e ither no preorbital fossa or only a very small one. The dorsal brai ncase behind the horncores is extremely abbreviated with a supraoccipital exposure that is anteroposteriorly so thin that it is confined to the nuchal crest. The fac ial updoming a nd extreme shortness of the exposed braincase make it difficult to judge the c ra niofac ial a ng le, bu t my es timate makes it surpri s ing ly hig h , ca. 110°, while the parietal­ occipital angle was very high, perhaps as high as 160°. The occip ital surface faces mainly backwards. It has a rounded dor al marg in that shows the beginnings of onl y gentle lateral concavities towards its missing basal parts. The very hypsodont teeth show that this was a young adult. Cranium BOU- 1/99 includes abo ut 200 mm of the right basal horncore and the dorsal braincase. It I SO belongs to a larger individual than BOU-2/21. Its horncore is less compressed at the base, shorter, with hig he r rates of diminuti on and increasing compress ion from the base upward , and with more anterior basal swelling and rugosity. There i again some posterior flattening, and defi nite c lockwise torsion on the right. The orbital area, although the rim is missing, sugges ts that the horncores are inserted closer behind the orbits than in BOU-2/21, a nd w ith s imil a r low basal divergence. The orientation of the basal horncore cross-section also differs in that the angle to the midline is larger: While BOU-2/2 1 has an orientation of its maximum diameter resembling Figure 2e, that of BOU- l/99 is more like Figure 2h. The horncore passes backwards above the insertion with more of an upright stem a bove which the base is more trongly convex a nteriorly than in BOU-2/2 1, a nd c hanges more abruptly in curvature towards a straighter course some 80mm above the base. There is no sign of upward recurvature on the pre erved di stal part . The late ral basal horncore margin is again much lower than the medial one. Anterior transverse ridges, above the basal rugose part, and large basal sinuses are ev ident. The vestige of a small postcornua l fossa can be seen on the right. There is a rugose ridge on the frontal anterior of the horncore passing from anterolateral to posteromedial. The face as far as pre erved (up to the middle of the orbit and above the missing supraorbital foramina) is quite flat, showing no s ign of the facial up-doming in BOU-2/21. This is nevetheless cons istent with the notion that these specimens are conspecific, as the rounded nasal crest in BOU-2/2 1 is present only from the anterior orbit forward. The cra niofac ial angle is ca. 110°. The coronal suture is not centrally indented. The surface of the po terior braincase roof and occipital surface is damaged such that the position of the occiput is uncertain, but one can see that the supra-occipita l expos ure was aga in anteroposteriorJy very thin, while the distance from the posterior horncores to the occiput was longer than in the previous specimen. There is no sign of a parietal boss. The occipital surface faces mainly backwards and is basally very wide. The poorly marked temporal lines on the braincase are very far separated at their closest approach. Frontlet BO U-2/20, with muc h of the ri ght horncore, is from an even larger individual than the previ o us one. The horncore base i s ha rdly compressed and quite swollen anter iorl y. The complete horncores were of comparable length as in BOU-2/21 , but shorte r relative to their much greater girth. They are inserted more steeply relative to the anterior part of the dorsal braincase than in the prev ious spec ime ns. They pass abruptly backwards (some 60 mm above the base), unlike the gentler curvature in BOU-2/21 and with more of a basal upright stem as in BOU- 1/99, although the backward curvature commences sooner above the base than in BOU-1/99. The horncore then has only a short straighter part, recurving upwards soone r and more strongly than in BOU-2/21. Although the mid-frontal a rea is miss ing, it seems that the horncores were inserted c losely together. Again , their lateral basal margins are much lower than their medial ones, and divergence is