Species Longevity in North American Fossil Mammals 7 8 8 9 9 10 10 11 Donald R

Home , Chalicotherioidea, Helohyidae, Hyopsodontidae, Hypertragulidae, Hyracodontidae, Oromerycidae

Integrative Zoology 2014; 9: 383–393 doi: 10.1111/1749-4877.12054

1 ORIGINAL ARTICLE 1 2 2 3 3 4 4 5 5 6 6 7 Species longevity in North American fossil mammals 7 8 8 9 9 10 10 11 Donald R. PROTHERO 11 12 12 13 Department of Vertebrate Paleontology, Natural History Museum of Los Angeles County, Los Angeles, CA, USA 13 14 14 15 15 16 16 17 17 18 Abstract 18 19 Species longevity in the fossil record is related to many paleoecological variables and is important to macroevo- 19 20 lutionary studies, yet there are very few reliable data on average species durations in Cenozoic fossil mammals. 20 21 Many of the online databases (such as the Paleobiology Database) use only genera of North American Cenozoic 21 22 mammals and there are severe problems because key groups (e.g. camels, oreodonts, pronghorns and probosci- 22 23 deans) have no reliable updated taxonomy, with many invalid genera and species and/or many undescribed gen- 23 24 era and species. Most of the published datasets yield species duration estimates of approximately 2.3–4.3 Myr 24 25 for larger mammals, with small mammals tending to have shorter species durations. My own compilation of 25 26 all the valid species durations in families with updated taxonomy (39 families, containing 431 genera and 998 26 27 species, averaging 2.3 species per genus) yields a mean duration of 3.21 Myr for larger mammals. This breaks 27 28 down to 4.10–4.39 Myr for artiodactyls, 3.14–3.31 Myr for perissodactyls and 2.63–2.95 Myr for carnivorous 28 29 mammals (carnivorans plus creodonts). These averages are based on a much larger, more robust dataset than 29 30 most previous estimates, so they should be more reliable for any studies that need species longevity to be accu- 30 31 rately estimated. 31 32 32 33 Key words: Cenozoic, genus, mammals, North America, species 33 34 34 35 35 36 36 37 INTRODUCTION lutionary and extinction-related characteristics. Stanley 37 38 (1975, 1979, 1981) uses species longevity in mammals 38 39 The topic of species longevity has long been of inter- versus bivalves as one of his tests of the ‘species selec- 39 40 est to paleontologists (see summary in Hopkins 2011). tion’ model of evolution. 40 41 The duration of species through geologic time has been 41 However, the comparison of species longevity to oth- 42 related to geographic range, body size, reproductive 42 er variables is often hampered by the problem of outdat- 43 rates, extinction probabilities, niche variation, intrapop- 43 ed or flawed species-level taxonomy in many groups. In 44 ulational variation, speciation rates and many other evo- 44 some cases, there is no recent reliable species-level tax- 45 45 onomy any other paleontologist can use, so comparison 46 46 to other ecological and evolutionary variables is impos- 47 47 sible. Nevertheless, paleontologists have tried to make 48 48 Correspondence: Donald R. Prothero, Department of Vertebrate such comparisons, using genera instead of species, in 49 49 Paleontology, Natural History Museum of Los Angeles County, databases such as FAUNMAP, MIOMAP, NEOMAP, as 50 50 900 Exposition Blvd., Los Angeles, CA 90007, USA. well as the Paleobiology Database (PBDB). Using gen- 51 Email: [email protected] 51

1 era as a proxy for species is done as a matter of neces- 1998), oreodonts (see Stevens & Stevens 1996; Prothero 1 2 sity and by default, but there are many biologists and & Sanchez 2008), antilocaprids (see Janis & Manning 2 3 paleontologists who argue that the species is the only bi- 1998) and peccaries (Wright 1998; Prothero 2009; D. R. 3 4 ologically natural unit in the biota and regard genera as Prothero, unpubl. data). 4 5 relatively poor proxies for true species diversity (see re- These and many other important groups make up a 5 6 view of the topic in Cracraft 1989). significant part of many Cenozoic faunas and my own 6 7 It is well known that there are many higher-level taxa experience has taught me that their published taxono- 7 8 (e.g. the families of fossil mammals) that have not been my is so bad as to be positively misleading. I have per- 8 9 revised recently but potentially will undergo big chang- sonally revised other important Miocene groups, such 9 10 es in total species and generic diversity. Once they are as the Rhinocerotidae (Prothero 2005), the dromomery- 10 11 revised, the use of largely invalid generic data on these cine palaeomerycids (Prothero & Liter 2008), the blas- 11 12 poorly known groups is ‘garbage in, garbage out’. This tomerycine musk deer (Prothero 2008) and the proto- 12 13 is particularly a problem with North American Cenozoic ceratids (Prothero 1998b), and documented the effects 13 14 mammals, where some groups are grossly oversplit and of the outdated taxonomy on the generic databases now 14 15 have no valid taxonomy available but were used in the available. For the example, the PBDB recognizes only 7 15 16 compilation of MIOMAP and PBDB nevertheless. Un- North American fossil genera of peccaries but Wright’s 16 17 like the database of Eurasian Cenozoic mammals, which (1998) revision suggests that there are a large number of 17 18 has been continuously revised for many years by a vari- undescribed and unnamed genera, and my revision (D. 18 19 ety of paleontologists, the North American mammalian R. Prothero, unpubl. data) recognizes at least 18 genera. 19 20 database is still full of invalid and outdated taxonomy In contrast, Prothero and Liter (2008) found that one- 20 21 that hampers efforts to look at species-level processes or third of the dromomerycid genera and half of the spe- 21 22 to conduct analyses on the species level. cies in the PBDB were invalid. Thus, the generic-lev- 22 23 23 This anomaly is due to several factors. First, the in- el databases can severely overestimate or underestimate 24 24 credible Frick Collection of fossil mammals in the the true number of taxa if the compilers have not kept 25 25 American Museum of Natural History remained virtu- up with these rapid taxonomic changes. Those who use 26 26 ally unstudied until the 1970s (and is only now being the PBDB for Neogene mammals may not fully appre- 27 27 slowly published piecemeal). That collection is so im- ciate the inherent flaws in calculating generic diversity 28 28 mense, with so many more complete and well-docu- based on bad taxonomy. 29 29 mented fossils, that it forced a complete revision of ev- Rather than accept generic longevity as a proxy for 30 30 ery group of Mid and Late Cenozoic mammals on the the evolutionary trends in species through time, it is 31 31 continent. In many cases, a major group remained virtu- worthwhile reviewing the published literature for areas 32 32 ally unstudied for decades until work on the Frick Col- where the taxonomy of a large group (such as Cenozoic 33 33 lection produced a monographic burst (e.g. Wang 1994; mammals) is well studied with widely accepted, up-to- 34 34 Wang et al. 1999; Tedford et al. 2009, on canids; my date species taxonomy and examining whether the true 35 35 own work on rhinocerotids in 2005). In addition, many longevity of species shows any consistent patterns with- 36 36 fossil mammalian families were burdened with the work in and between taxa and whether generic longevity data 37 37 of taxonomists who are widely considered incompetent, correlate well with species durations. 38 38 from Frick’s 1937 ruminant monograph, to the Schul- 39 39 tz and Falkenbach (1968) oreodont monograph, to Os- 40 MATERIALS AND METHODS 40 born’s brontothere monograph (1929) and proboscidean 41 41 monograph (1936). Although some of these works have Species longevities for Cenozoic fossil mammals 42 42 since been revised (e.g. Mihlbachler’s 2008 brontothere were compiled from the published literature and from 43 43 monograph replaces Osborn 1929), many others have large datasets from a number of sources. For the with- 44 44 not (e.g. there is still no updated taxonomy on probosci- in-family approach, I extracted longevities of species as 45 45 deans, oreodonts, pronghorns or many of Frick’s rumi- reported in the most recent revision of a group (cited be- 46 46 nants). In the Frick Collection alone, there is an entire low), modified as necessary when I was aware of prob- 47 47 floor of fossil camels and another floor of proboscideans lematic taxa. The time scale of the species ranges fol- 48 48 that are virtually unstudied. I have personally experi- lows the individual chapters in Woodburne (2004), as 49 49 enced this problem with the current outdated state of the modified in Janis et al. (1998, 2008). The time duration 50 50 taxonomy of camels (see Prothero 1996a; Honey et al. of each species was tabulated from their actual biostra- 51 51

1 tigraphic range using range-through methods. As much large mammal groups whose species-level taxonomy is 1 2 as possible, I used the chronostratigraphic dating of the still insufficiently studied: Camelidae, Antilocapridae, 2 3 actual documented stratigraphic first and last occur- oreodonts, primitive carnivorans (‘Miacidae’), Xenar- 3 4 rence data for each species, rather than lumping them thra and the Proboscidea. All the rest of the families list- 4 5 into coarser-resolution bins of ‘land mammal ages’. I ed below have sufficiently updated taxonomy (cited be- 5 6 did not extend ranges using inferred ghost lineages below). I entered the raw data into an Excel spreadsheet 6 7 cause the cladistic relationships in some families are too and then calculated the statistics. Histograms were plot- 7 8 poorly known to do this reliably. Generic ranges were ted using StatPlus LE. 8 9 calculated from the oldest and youngest known mem- 9 10 bers of valid species and do not reflect incomplete spec- RESULTS 10 11 imens that could not be identified to species but might 11 12 extend the generic range (in contrast to the methods Two approaches can be taken to looking at fossil spe- 12 13 used by Janis et al. 1998, 2008). I did not use any ex- cies longevity in fossil mammals: a broad faunal survey 13 14 tant species to avoid the ‘pull of the recent’ effect and approach (all the named taxa in a temporal interval that 14 15 any isolated stem-group taxa not assignable to a mono- has been well studied and has a stable species-level tax- 15 16 phyletic crown group at the family level. For reasons onomy) or a phylogenetic within-family approach (all 16 17 of space, I focused on the larger carnivorous and her- of the species within a particular family-level group that 17 18 bivorous mammalian groups (primarily Carnivora, Ar- has undergone recent systematic revision). 18 19 tiodactyla and Perissodactyla) so that the Eurasian–Af- Faunal surveys 19 20 rican NOW database (the NOW database: http://www. 20 21 helsinki.fi/science/now/) and North American databas- Although an up-to-date, modern species taxonom- 21 22 es are comparable. Small mammal groups are certainly ic database is not available for all Cenozoic mammal 22 23 worth examining, although their species-level taxonomy faunas, there are some time and geographic intervals 23 24 in North America is still in flux and adding all of them that have undergone thorough revision. For example, 24 25 was beyond the scope of the present study. Thus, among the large number of paleontologists who have stud- 25 26 North American mammals I did not count the following ied Pleistocene mammals in both the Old World and the 26 27 27 28 28 29 29 30 30 31 Table 1 Estimates of species durations in Cenozoic mammals from different regions and different data sources 31 32 Region Age Mean duration (Myr) Source 32 33 Europe Pleistocene 1.0 Kurtén 1968 33 34 Europe Neogene 2.2 (large mammals) Liow et al. 2008 34 35 Europe Neogene 2.4 (small mammals) Liow et al. 2008 35 36 Europe–Africa Neogene 6.3 (large mammals) Raia et al. 2012 36 37 Europe–Africa Neogene 4.8 (large mammals) Raia et al. 2012 37 38 Europe–Africa Neogene 3.7 (large mammals) NOW database 38 39 Africa Neogene 2.3 (large mammals) Vrba 2000 39 40 Asia Neogene 2.5 (artiodactyls) Flynn et al. 1995 40 41 North America Eocene 2.6 Stanley 1978, 1979 41 42 North America Eocene 0.8–1.0 Gingerich 1980, 1985 42 43 North America Oligocene 2.37 Prothero & Heaton 1996 43 44 North America Cenozoic 3.21 (large mammals) This study (raw data) 44 45 North America Cenozoic 3.21 (large mammals) This study (family averages) 45 46 North America Cenozoic 4.39 (artiodactyls) This study (raw data) 46 47 North America Cenozoic 4.10 (artiodactyls) This study (family averages) 47 48 North America Cenozoic 3.14 (perissodactyls) This study (raw data) 48 49 North America Cenozoic 3.31 (perissodactyls) This study (family averages) 49 North America Cenozoic 2.95 (carnivores) This study (raw data) 50 50 North America Cenozoic 2.63 (carnivores) This study (family averages) 51 51

1 New World allows such summaries. Kurtén (1968) pub- durations estimated by Raia et al. (2012) for some of the 1 2 lished one of the first such estimates (Table 1) and gave same taxa. 2 3 a mean duration for most Pleistocene mammal species Estimates of species durations in North American 3 4 of the Old World as approximately 1 Myr; this figure mammals are less common. Stanley (1978, 1979) quotes 4 5 was cited by Stanley (1975, 1979) for use in comparison a mean duration of Eocene mammal species at 2.6 Myr, 5 6 with rates of evolution in other groups, such as bivalves. but Gingerich (1980, 1985) gives durations of 0.8–1.0 6 7 Liow et al. (2008) look at more recent data for Neogene Myr for species from the Early Eocene of the Bighorn 7 8 mammals of the Old World (the NOW database: http:// Basin, Wyoming. Prothero and Heaton (1996) survey 8 9 9 www.helsinki.fi/science/now/) and find that small mam- the durations of all the known species from the up- 10 10 mal genera and species have longer median durations per Eocene–Oligocene White River Group in the High 11 11 than large mammal genera. In Table S1 of the Supple- Plains. They find a mean duration of 2.37 Myr for 177 12 12 mentary Data Series 1 of Liow et al. (2008), a mean du- species, with very long durations for insectivores (6.7 13 13 ration of all large mammal species of 2.17 Myr is calcu- Myr) and carnivorous mammals (4.6 Myr) and very 14 14 lated, with a median duration of 1.5 Myr (1106 species short durations for rodents (1.14 Myr) and lagomorphs 15 15 total). For small mammal species (n = 868), the mean (1.58 Myr), with hoofed mammals (artiodactyls = 2.91 16 was 2.4 Myr and the median value was 1.5 Myr. 16 17 Myr; perissodactyls = 2.20 Myr) coming in between. 17 18 Raia et al. (2012) also analyze the NOW database, These numbers are consistent with those of Flynn et al. 18 19 focusing on just 72 species of ungulates, carnivores and (1995), who find rodents tend to have shorter mean spe- 19 20 proboscideans with good fossil records in Europe and cies durations than those of larger mammals, such as ar- 20 21 Africa. I extracted the raw data from their Table S1 in tiodactyls. 21 22 the electronic supplementary material available online. The data discussed above are summarized in Table 1. 22 23 In that data table, they quote 2 different sets of statis- Even though the estimates of mean species duration are 23 24 tics: ‘Duration’ is ‘stratigraphic distribution in millions not identical, most of the datasets produce mean spe- 24 25 of years’; and ‘FA’ and ‘LA’ are last appearance datums. cies duration of approximately 2–3 Myr, with only a 25 26 I cannot determine why it is that when I calculated ‘FA– handful of outliers that are significantly shorter or lon- 26 27 LA’ that it does not match ‘Duration’ of the species in ger. The trend for carnivorous mammals to have the lon- 27 28 millions of years, so the results were calculated sepa- gest ranges in the Eocene–Oligocene of North Ameri- 28 29 rately. The ‘Duration’ column produces a mean value of ca (Prothero & Heaton 1996) is not upheld by the trends 29 30 6.29 Myr and a median value of 4.76 Myr, while sub- observed in the NOW database of Europe and Africa in 30 31 tracting ‘FA–LA’ produces a mean value of 4.8 Myr and the Neogene (Table 2). 31 32 a median of 3.5 Myr. P. Raia (pers. comm.) provided me 32 33 with the spreadsheet of their original compilation of the 33 34 NOW dataset, with 555 species of European and Afri- 34 35 can mammals. This dataset gives a mean of 3.739 Myr 35 36 for all of these species combined. Table 2 Comparison of durations (in Myr) of species within 36 37 Flynn et al. (1995) use the large chronostratigraphi- the same order between the Neogene of the Old World (OW), 37 38 cally-controlled dataset of Miocene–Pliocene mammals from the NOW database, after Raia et al. (2012) and the Paleo- 38 39 from the Siwalik Hills of Pakistan to compile species gene (Late Eocene–Oligocene) of the North America (NAO) 39 40 durations in southern Asia. They find a mean duration of (Prothero & Heaton 1996) and the Cenozoic of North America 40 41 1.3 Myr for the rodents and 2.5 Myr for the artiodactyls, (NAC) (this study) 41 42 but do not average or plot the data from other orders Taxon Species Duration N (spp.) 42 43 (perissodactyls, proboscideans) because of the small Carnivorous mammals OW 4.6 14 43 44 sample size. NAO 4.6 32 44 45 NAC 2.9 366 45 46 Vrba (2000) and Vrba and DeGusta (2004) estimate Perissodactyls OW 7.1 18 46 47 the average species duration of larger African mam- NAO 2.2 29 47 48 mals at 2.33 Myr. Their original data are not available NAC 3.1 209 48 49 for analysis, so it was not possible to calculate the medi- Artiodactyls OW 5.6 34 49 50 an, or to break this number down into taxonomic or eco- NAO 2.9 41 50 51 logical categories, but it is considerably shorter than the NAC 4.4 150 51

1 Within-family species durations ily. The mean for all large mammals was 3.21 Myr (vir- 1 2 tually the same as obtained by averaging family means). 2 3 As described in the Materials and Methods section, I The mean for artiodactyls was 4.39 Myr (compared to 3 4 compiled all the temporal ranges for all the valid species 3.96 Myr from averaging family means). The mean for 4 5 and genera within North American mammalian fami- perissodactyls was 3.14 Myr (compared to 3.31 Myr by 5 6 lies with valid, updated taxonomy. Results are shown in averaging family means). Finally, the mean for carniv- 6 7 Table 3 and Figure 1. A total of 39 mammalian families orous mammals was 2.95 Myr (compared to 2.63 Myr 7 8 were tabulated, containing 431 genera and 998 species, obtained by averaging family means). 8 which averages approximately 2.3 species per genus. 9 These numbers are all within the range of values 9 10 Mean durations for species in these genera averaged by 10 family (Table 3) ranged from 0.75 Myr (Pantolambdidae) to quoted in earlier studies (Table 1), although they show 11 carnivores with shorter durations than ungulates, the op- 11 12 6.58 Myr (Hypertragulidae). The mean for all artiodac- 12 tyl species was 3.96 Myr; for perissodactyls, 3.31 Myr; posite of what Prothero and Heaton (1996) find for the 13 mammals of the Eocene–Oligocene White River Fauna. 13 14 and for carnivores, 2.63 Myr. The combined mean dura- 14 Figure 1 shows a least-squares regression of the re- 15 tion of all 983 species was 3.19 Myr. Slightly different 15 lationship between mean species duration and mean ge- 16 values were obtained when all the raw species duration 16 neric duration. Clearly, the 2 variables are moderately well 17 data within an order were averaged independent of fam- 17 18 18 19 c 19 20 20 21 a 21 22 22 23 23 24 24 25 25 26 26 27 27 28 28 29 29 30 30 31 31 32 32 33 d 33 34 b 34 35 35 36 36 37 37 38 38 39 39 40 40 41 41 42 42 43 43 44 44 45 45 46 Figure 1 Least-squares regression of the relationship between mean species duration and mean generic duration in each of the 85 46 47 mammalian families in this study. (a) Entire dataset. (b) Dataset excluding the 3 outliers (Tapiridae, Hypertragulidae, Leptomery- 47 48 cidae), resulting in a somewhat stronger correlation. In both plots, the line is the least-squares fit and the lower limit is the lower 48 49 threshold of 1 genus per species (all monospecific genera). (c) Regression of average number of species per genus in the Canidae. (d) 49 50 Same as c, but for the Equidae. In both plots c and d, the monospecific genera form a clear trend line along the lower threshold of 50 51 the dataset. 51

1 Table 3 Summary of data of valid species and genera in each family of North American mammals 1 2 Family N spp. N genera Avg.sp.dur. Avg.gen.dur Source 2 3 Artiodactyla 3 4 Anthracotheriidae 15 6 2.32 5.4 Kron & Manning 1998 4 5 Helohyidae 10 5 3.33 4.4 Stucky 1998 5 6 Entelodontidae 15 6 5.13 8.83 Effinger 1998 6 Tayassuidae 26 17 3.32 4.14 Wright 1998; Prothero 2009; D. R. Prothero, 7 7 unpubl. data 8 Protoceratidae 30 13 4.23 6.16 Prothero 1998b 8 9 Oromerycidae 9 6 4.55 6 Prothero 1998a 9 10 Leptomerycidae 11 4 4.72 13 Webb 1998 10 11 Hypertragulidae 12 5 6.58 14.2 Webb 1998 11 12 Moschidae 9 6 4.71 8.16 Prothero 2008 12 13 Palaeomerycidae 16 12 4.47 4.71 Prothero & Liter 2008 13 Leptochoeridae 10 5 3.09 4.14 Stucky 1998 14 14 Antiacodontidae 9 5 2.37 2.54 Stucky 1998 15 Homacodontidae 16 11 2.64 3.3 Stucky 1998 15 16 Artiodactyla mean 14.46 7.77 3.96 6.53 16 17 Perissodactyla 17 18 Chalicotheriidae 8 4 3.73 5.00 Coombs 1998 18 19 Amynodontidae 8 5 3.75 6.20 Wall 1998 19 Hyracodontidae 13 5 3.50 6.80 Prothero 1996b, 1998c 20 20 Rhinocerotidae 37 16 3.77 6.44 Prothero 1998d, 2005 21 Tapiridae 14 6 2.60 11.22 Colbert & Schoch 1998 21 22 Isectolophidae 8 3 2.93 5.43 Colbert & Schoch 1998 22 23 Brontotheriidae 27 21 3.29 3.28 Mihlbachler 2008 23 24 Equidae 121 37 2.92 4.65 Froehlich 2002; Hulbert 1993; MacFadden 24 25 1998; Prothero and Shubin 1989 25 26 Perissodactyla mean 29.50 12.13 3.31 6.13 26 Carnivorous mammals 27 Canidae 135 39 3.39 7.27 Wang 1994; Wang et al. 1999; Tedford et al. 2009 27 28 Felidae 24 11 2.29 3.91 Martin 1998 28 29 Nimravidae 15 7 2.95 5.86 Bryant 1996 29 30 Ursidae 23 17 1.77 2.29 Hunt 1998b 30 31 Amphicyonidae 34 12 3.19 5.72 Hunt 1998a 31 32 Procyonidae 19 11 2.06 3.75 Baskin 1998b 32 Mustelidae 55 38 3.61 4.19 Baskin 1998a 33 33 Oxyaenidae 27 10 1.91 3.95 Gunnell 1998 34 Hyaendontidae 35 15 2.47 3.88 Gunnell 1998 34 35 Carnivore mean 40.78 17.78 2.63 4.54 35 36 Miscellaneous families 36 37 Conoryctidae 7 5 1.27 1.44 Lucas et al. 1998 37 38 Stylinodontidae 7 4 5.8 7.87 Lucas et al. 1998 38 Esthonychidae 8 5 4.2 4.34 Lucas & Schoch 1998b 39 39 Pantolambdidae 6 2 0.75 1.15 Lucas 1998 40 Barylambdidae 7 4 1.68 1.72 Lucas 1998 40 41 Uintatheriidae 6 4 5.06 6.45 Lucas & Schoch 1998a 41 42 Arctocyonidae 50 15 1.59 3.87 Archibald 1998 42 43 Hyopsodontidae 35 8 2.37 4.99 Archibald 1998 43 44 Mioclaenidae 26 9 1.69 3.28 Archibald 1998 44 45 Periptychiidae 35 13 1.19 1.79 Archibald 1998 45 Phenacodontidae 20 4 3.26 8.89 Archibald 1998 46 46 Miscellaneous mean 18.82 6.64 2.62 4.16 47 Combined means 24.61 10.65 3.19 5.40 47 48 N spp., number of species in family; N genera, number of genera in family; Avg.sp.dur, average species duration (in Myr); 48 49 Avg.gen.dur, average genus duration (in Myr). 49 50 50 51 51

1 correlated by the standards of paleoecological datasets ly similar results, with most datasets giving mean spe- 1 2 such as these data. The r2 = 0.54 for the entire dataset is cies durations of approximately 2–4 Myr. Thus, the 2 3 considerably better (r2 = 0.61) if the 3 outliers (Hyper- 1-Myr average duration for European Pleistocene mam- 3 4 tragulidae, Leptomerycidae and Tapiridae) are exclud- mals reported by Kurtén (1968) is an underestimate de- 4 5 ed (Fig. 1b). (For some reason, those 3 families have a termined before the rigorous work in dating and taxono- 5 6 lot of anomalously long-lived species.) Note in the plots my of the past 40 years (see more recent estimates in the 6 7 in Figure 1a and b that although there are several outli- NOW database). 7 8 ers above the least-squares lines (i.e. genera with unusu- Some datasets (e.g. Flynn et al. 1985; Prothero & 8 9 ally long average durations relative to their average spe- Heaton 1996) suggest that small mammals have shorter 9 10 cies longevity), there cannot be similar outliers below species durations than do large mammals, although oth- 10 11 the line. That is because there is a lower threshold in the ers (e.g. Raia et al. 2012) indicate the opposite. With the 11 12 field below the least-squares line, because it is impossi- existence of the large dataset summarized in this study, 12 13 ble for a species to have a duration longer than the ge- it should be possible to see whether or not this trend 13 14 nus to which it belongs. Thus, the lower limit for the holds true when the small mammal species are compiled 14 15 data would be a line where the number of species and and analyzed. 15 16 16 genera are exactly equal (i.e. each genus is monospecif- Most comparisons of species durations are done with 17 17 ic and monotypic). histograms, such as those seen in analyses by Liow et 18 18 One reviewer suggested that the correlation might al. (2008) and Flynn et al. (1995). Figure 2a shows his- 19 19 improve if the durations of genera and average species tograms of the 3 major ordinal-level groups of large 20 20 duration in a given family were plotted against each oth- mammals in this study: artiodactyls, perissodactyls and 21 21 er. To test this hypothesis, I chose 2 of the largest fam- carnivorous mammals (Carnivora plus Creodonta). All 3 22 22 ilies (to improve sample size and, thus, the likelihood plots are slightly skewed toward a greater representation 23 2 23 of a higher r value), the Canidae and Equidae. In fact, of the shorter-lived taxa, but not strongly so. By con- 24 24 the opposite is the case: the correlation was much poor- trast, in their study of Neogene Siwalik mammals, Fly- 25 25 er when examined at the genus versus species level (Fig. nn et al. (1995; see Fig. 2b) show a distribution of ar- 26 2 26 1c,d). Least-squares lines fitted to these plots give r tiodactyls that is strongly skewed toward shorter-lived 27 27 values of 0.2–0.4. The reason for this much poorer cor- species and also contains some very long-lived spe- 28 28 relation is apparent when the plot is examined in detail. cies. Some are longer in duration than most of the spe- 29 29 At the within-family level, there are a number of genera cies in my dataset. Liow et al. (2008) analyze the Neo- 30 30 that are monospecific (genus and species range are iden- gene mammals of Eurasia and Africa and produce an 31 31 tical), producing an obvious trend of points lined along even more strongly skewed histogram for both large and 32 32 the bottom of the scatter. The least-squares line is trying small-bodied mammals, with a very high percentage of 33 33 to fit the overall scatter of points when there are actual- the species in the very short-ranging (less than 1 Myr) 34 34 ly 2 trends: the 1:1 line of monospecific genera and the categories. Their histogram (see Fig. 2c) also includes 35 35 cluster of the rest of the taxa plotted above that thresh- numerous species with durations longer than 15 Myr, 36 36 old. However, the plots of durations at the family lev- something that is completely lacking in my dataset. 37 el (Fig. 1a,b) contain no monotypic groups, because av- 37 38 The lack of updated species-level taxonomy in North 38 eraging genera and species durations prevents this from American fossil mammals has been a hindrance to ef- 39 occurring (all families had at least 2 genera and more 39 40 forts to use existing databases (e.g. PBDB, NEOMAP, 40 than 2 species). Consequently, the trend line fits a lot MIOMAP and FAUNMAP) that only contain genera. 41 better. 41 42 Without species-level data, it is difficult to analyze these 42 43 DISCUSSION data to solve other interesting paleontological problems. 43 To get around this problem, some (e.g. Alroy 2002) have 44 The compilation of raw species longevity data, based 44 45 proposed statistical models to estimate the number of 45 on a new taxonomy of North American large mammals, valid species knowing the number of genera. The pres- 46 provides a number of results consistent with previous 46 47 ent study provides empirical evidence of how many spe- 47 studies and others that differ remarkably from the pub- cies and genera are actually considered valid as of the 48 lished literature. 48 49 latest taxonomic revisions. Thus, if we know that there 49 Species durations of fossil mammals as estimated by 50 are x number of genera in the dataset, our empirical data 50 a number of different methods (Table 1) provide rough- 51 suggests that there should be 2.3x valid species (because, 51

1 a as demonstrated above, each genus has, on average, 2.3 1 2 species in the dataset). 2 3 3 4 CONCLUSIONS 4 5 5 6 Many different estimates of the average duration of 6 7 fossil mammal species have been provided over the past 7 8 50 years. Most give mean durations in the order of 2–4 8 9 Myr, although there are some time intervals (e.g. the 9 10 Early Eocene) that are alleged to have shorter durations 10 11 b of species. Comparison of the existing datasets for larg- 11 12 er mammals of the Old World (both carnivorous mam- 12 13 mals and prey, such as artiodactyls and perissodactyls) 13 14 b and North America (this study) reveals broadly similar 14 15 results, although the North American mammals do not 15 16 have such a high proportion of short-lived species, nor 16 17 are there as many species with durations as long as 15 17 18 Myr or longer. 18 19 19 20 ACKNOWLEDGMENTS 20 21 21 22 I thank Maria Rita Palombo for inviting this contri- 22 23 bution and Spencer Lucas for critiquing it. 23 24 24 25 REFERENCES 25 26 26 c 27 Alroy J (2002). How many named species are valid? 27 28 PNAS 99, 3706–711. 28 29 Archibald JD (1998). Archaic ungulates (‘Condylar- 29 30 thra’). In: Janis C, Scott KM, Jacobs LL, eds. Evo- 30 31 lution of Tertiary Mammals of North America. Cam- 31 32 bridge University Press, Cambridge, pp. 292–331. 32 33 Baskin JA (1998a). Mustelidae. In: Janis C, Scott KM, 33 34 Jacobs LL, eds. Evolution of Tertiary Mammals of 34 35 North America. Cambridge University Press, Cam- 35 36 bridge, pp. 152–73. 36 37 37 Figure 2 Distribution of total number of species of a given lon- Baskin JA (1998b). Procyonidae. In: Janis C, Scott KM, 38 38 gevity. Each bin represents the proportion of the total species Jacobs LL, eds. Evolution of Tertiary Mammals of 39 39 lasting between 0 and 1 Myr, 1 and 2 Myr, and so on. (a) Data North America. Cambridge University Press, Cam- 40 from the North American Cenozoic. The left-hand dark bar bridge, pp. 144–51. 40 41 is the carnivorous mammals, the middle gray bar is artiodac- Bryant HD (1996). Nimravidae. In: Prothero DR, Emry 41 42 tyls, and the right-hand dark bar is perissodactyls. (b) Artiodac- RJ, eds. The Terrestrial Eocene–Oligocene Transition 42 43 tyls from the Neogene of the Siwalik Hills, Pakistan (replot- in North America. Cambridge University Press, Cam- 43 44 ted from Flynn et al. 1995, fig. 3). (c) Data from the Neogene bridge, pp. 453–75. 44 45 of the Old World database (NOW) (replotted from Liow et al. 45 Colbert MW, Schoch RM (1998). Tapiroidea and oth- 46 2008, fig. 1). The dark bar is large mammals, the gray bar is 46 er moropomorphs. In: Janis C, Scott KM, Jacobs LL, 47 small mammals. Note that the North American data set is much 47 eds. Evolution of Tertiary Mammals of North Amer- 48 more evenly distributed, with not nearly as high a proportion of 48 ica. Cambridge University Press, Cambridge, pp. 49 short-ranging species (except in the carnivores), and does not 49 569–82. 50 range to species with durations greater than 9 Myr. 50 51 51

1 Coombs MC (1998). Chalicotherioidea. In: Janis C, Scott Hunt RM Jr (1998b). Ursidae. In: Janis C, Scott KM, 1 2 KM, Jacobs LL, eds. Evolution of Tertiary Mammals Jacobs LL, eds. Evolution of Tertiary Mammals of 2 3 of North America. Cambridge University Press, Cam- North America. Cambridge University Press, Cam- 3 4 bridge, pp. 560–8. bridge, pp. 174–95. 4 5 Cracraft J (1989). Species as entities in biological the- Janis CM, Manning E (1998). Antilocapridae. In: Ja- 5 6 ory. In: Ruse M, ed. What the Philosophy of Biology nis C, Scott KM, Jacobs LL, eds. Evolution of Tertia- 6 7 Is. Kluwer, Dordrecht, pp. 31–52. ry Mammals of North America. Cambridge Universi- 7 8 8 Effinger JA (1998). Entelodontidae. In: Janis C, Scott ty Press, Cambridge, pp. 491–507. 9 9 KM, Jacobs LL, eds. Evolution of Tertiary Mammals Janis C, Scott KM, Jacobs LL, eds (1998). Evolution 10 10 of North America. Cambridge University Press, Cam- of Tertiary Mammals of North America, Vol. 1: Ter- 11 11 bridge, pp. 375–80. restrial Carnivores, Ungulates and Ungulate–Like 12 12 13 Flynn LJ, Barry JC, Morgan ME, Pilbeam D, Jacobs Mammals. Cambridge University Press, Cambridge. 13 14 LL, Lindsay EH (1995). Neogene Siwalik mammali- Janis C, Gunnell GM, Uhen MD, eds (2008). Evolution 14 15 an lineages: species longevities, rates of change and of Tertiary Mammals of North America, Vol. 2: Small 15 16 modes of speciation. Palaeogeography, Palaeoclima- Mammals, Xenarthrans and Marine Mammals. Cam- 16 17 tology, Palaeoecology 115, 249–64. bridge University Press, Cambridge. 17 18 Frick C (1937). Horned ruminants of North America. Kurtén B (1968). Pleistocene Mammals of Europe. Al- 18 19 Bulletin of the American Museum of Natural History dine Publishing Company, Chicago. 19 69, 1–669. 20 Kron DG, Manning EM (1998). Anthracotheriidae. In: 20 21 Froehlich DJ (2002). Quo vadis Eohippus? The sys- Janis C, Scott KM, Jacobs LL, eds. Evolution of Ter- 21 22 tematics and taxonomy of Early Eocene equids tiary Mammals of North America. Cambridge Uni- 22 23 (Perissodactyla). Zoological Journal of the Linnean versity Press, Cambridge, pp. 381–8. 23 24 Society 134, 141–256. 24 Liow LH, Fortelius M, Bingham E et al. (2008). High- 25 25 Gingerich PD (1980). Evolutionary patterns in Ear- er origination and extinction rates in larger mammals. 26 26 ly Cenozoic mammals. Annual Reviews of Earth and PNAS 105, 6097–102. 27 Planetary Sciences 8, 407–24. 27 28 Lucas SM (1998) Pantodonta. In: Janis C, Scott KM, 28 Gingerich PD (1985). Species in the fossil record: con- Jacobs LL, eds. Evolution of Tertiary Mammals of 29 cepts, trends, transitions. Paleobiology 11, 27–41. 29 30 North America. Cambridge University Press, Cam- 30 Gunnell GF (1998). Creodonta. In: Janis C, Scott KM, 31 bridge, pp. 274–83. 31 Jacobs LL, eds. Evolution of Tertiary Mammals of 32 Lucas SM, Schoch RM (1998a). Dinocerata. In: Janis 32 North America. Cambridge University Press, Cam- 33 C, Scott KM, Jacobs LL, eds. Evolution of Tertiary 33 bridge, pp. 91–109. 34 Mammals of North America. Cambridge University 34 35 Honey J, Harrison JA, Prothero DR, Stevens MS (1998). Press, Cambridge, pp. 284–91. 35 Camelidae. In: Janis C, Scott KM, Jacobs LL, eds. 36 Lucas SM, Schoch RM (1998b). Tillodontia. In: Janis 36 Evolution of Tertiary Mammals of North America. 37 C, Scott KM, Jacobs LL, eds. Evolution of Tertiary 37 Cambridge University Press, Cambridge, pp. 439–62. 38 Mammals of North America. Cambridge University 38 39 Hopkins MJ (2011). How species longevity, intraspecif- Press, Cambridge, pp. 268–73. 39 40 ic morphological variation and geographic range size 40 Lucas SM, Schoch RM, Williamson TE (1998). Tae- 41 are related: a comparison using Late Cambrian trilo- 41 niodonta. In: Janis C, Scott KM, Jacobs LL, eds. Evo- 42 bites. Evolution 65, 3253–73. 42 lution of Tertiary Mammals of North America. Cam- 43 43 Hulbert RC Jr (1993). Taxonomic evolution in North bridge University Press, Cambridge, pp. 260–7. 44 American Neogene horses (subfamily Equinae): the 44 45 rise and fall of an adaptive radiation. Paleobiology MacFadden BJ (1998). Equidae. In: Janis C, Scott KM, 45 46 19, 216–34. Jacobs LL, eds. Evolution of Tertiary Mammals of 46 North America. Cambridge University Press, Cam- 47 Hunt RM Jr (1998a). Amphicyonidae. In: Janis C, Scott 47 bridge, pp. 537–59. 48 KM, Jacobs LL, eds. Evolution of Tertiary Mammals 48 49 of North America. Cambridge University Press, Cam- Martin LD (1998). Felidae. In: Janis C, Scott KM, Ja- 49 50 bridge, pp. 196–227. cobs LL, eds. Evolution of Tertiary Mammals of 50 51 51

1 North America. Cambridge University Press, Cam- Prothero DR, Liter M (2008). Systematics of the dro- 1 2 bridge, pp. 236–43. momerycines and aletomerycines (Artiodactyla: Pa- 2 3 Mihlbachler MC (2008). Species taxonomy, phylogeny laeomerycidae) from the Miocene and Pliocene of 3 4 and biogeography of the Brontotheriidae (Perissodac- North America. New Mexico Museum of Natural His- 4 5 tyla, Mammalia). Bulletin of the American Museum tory and Science Bulletin 44, 273–98. 5 6 of Natural History 311, 1–473. Prothero DR, Sanchez F (2008). Systematics of the lep- 6 7 7 Osborn HF (1929). The titanotheres of ancient Wyo- tauchenine oreodonts (Mammalia: Artiodactyla) from 8 8 ming, Nebraska and South Dakota. US Geological the Oligocene and earliest Miocene of North Ameri- 9 9 Survey Monograph 55, 1–701. ca. New Mexico Museum of Natural History and Sci- 10 ence Bulletin 44, 335–56. 10 11 Osborn HF (1936). Proboscidea. American Museum 11 Prothero DR, Shubin N (1989). The evolution of Oligo- 12 Press, New York. 12 cene horses. In: Prothero DR, Schoch RM, eds. The 13 Prothero DR (1996a). Camelidae. In: Prothero DR, 13 Evolution of Perissodactyls. Oxford University Press, 14 Emry RJ, eds. The Terrestrial Eocene–Oligocene 14 New York, pp. 142–75. 15 Transition in North America. Cambridge University 15 16 Press, Cambridge, pp. 591–633. Raia P, Passaro F, Fulgione D, Carotenuto F (2012). 16 17 Prothero DR (1996b). Hyracodontidae. In: Prothero DR, Habitat tracking, stasis and survival in Neogene large 17 18 Emry RJ, eds. The Terrestrial Eocene–Oligocene mammals. Biology Letters 8, 64–6. 18 19 Transition in North America. Cambridge University Schultz CB, Falkenbach CH (1968). The phylogeny of 19 20 Press, Cambridge, pp. 634–45. the oreodonts. Bulletin of the American Museum of 20 21 Prothero DR (1998a). Oromerycidae. In: Janis C, Scott Natural History 139, 1–498. 21 22 KM, Jacobs LL, eds. Evolution of Tertiary Mammals Stanley SM (1975). A theory of evolution above the spe- 22 23 of North America. Cambridge University Press, Cam- cies level. PNAS 72, 646–50. 23 24 bridge, pp. 426–30. Stanley SM (1978). Chronospecies’ longevities, the or- 24 25 Prothero DR (1998b). Protoceratidae. In: Janis C, Scott igin of genera and the punctuational model of evolu- 25 26 KM, Jacobs LL, eds. Evolution of Tertiary Mammals tion. Paleobiology 4, 26–40. 26 27 of North America. Cambridge University Press, Cam- Stanley SM (1979). Macroevolution: Pattern and Pro- 27 28 bridge, pp. 431–8. cess. W. H. Freeman, San Francisco. 28 29 29 Prothero DR (1998c). Hyracodontidae. In: Janis C, Scott Stanley SM (1981). The New Evolutionary Timetable. 30 30 KM, Jacobs LL, eds. Evolution of Tertiary Mammals Basic Books, New York. 31 31 of North America. Cambridge University Press, Cam- 32 Stevens MS, Stevens J (1996). Merycoidontinae. In: 32 bridge, pp. 589–94. 33 Prothero DR, Emry RJ, eds. The Terrestrial Eocene– 33 34 Prothero DR (1998d). Rhinocerotidae. In: Janis C, Scott Oligocene Transition in North America. Cambridge 34 35 KM, Jacobs LL, eds. Evolution of Tertiary Mammals University Press, Cambridge, pp. 481–555. 35 36 of North America. Cambridge University Press, Cam- Stucky RK (1998). Eocene bunodont and bunoseleno- 36 37 bridge, pp. 595–605. dont Artiodactyla. In: Janis C, Scott KM, Jacobs LL, 37 38 Prothero DR (2005). The Evolution of North American eds. Evolution of Tertiary Mammals of North Amer- 38 39 Rhinoceroses. Cambridge University Press, Cam- ica. Cambridge University Press, Cambridge, pp. 39 40 bridge. 358–74. 40 41 Prothero DR (2008). Systematics of the musk deer (Ar- Tedford R, Wang X, Taylor BE (2009). Phylogenet- 41 42 tiodactyla: Moschidae: Blastomerycinae) from the ic systematics of the North American fossil Caninae 42 43 Miocene of North America. New Mexico Museum of (Carnivora: Canidae). Bulletin of the American Muse- 43 44 Natural History and Science Bulletin 44, 207–25. um of Natural History 325, 1–218. 44 45 Prothero DR (2009). The early evolution of North Vrba E (2000). Major features of Neogene mammalian 45 46 American peccaries (Tayassuidae). Museum of North- evolution in Africa. In: Partridge TC, Maud R, eds. 46 47 ern Arizona Bulletin 65, 509–42. Cenozoic Geology of Southern Africa. Oxford Uni- 47 48 48 Prothero DR, Heaton TH (1996). Faunal stability during versity Press, Oxford, pp. 277–304. 49 49 the Early Oligocene climatic crash. Palaeogeogra- Vrba E, DeGusta D (2004). Do species populations real- 50 50 phy, Palaeoclimatology, Palaeoecology 127, 239–56. ly start small? New perspectives from the Late Neo- 51 51

1 gene fossil record of African mammals. Philosoph- nidae). Bulletin of the American Museum of Natural 1 2 ical Transactions of the Royal Society of London B History 243, 1–391. 2 3 359, 285–93. Webb SD (1998). Hornless ruminants. In: Janis C, Scott 3 4 4 Wall WP (1998). Amynodontidae. In: Janis C, Scott KM, Jacobs LL, eds. Evolution of Tertiary Mammals 5 5 KM, Jacobs LL, eds. Evolution of Tertiary Mammals of North America. Cambridge University Press, Cam- 6 bridge, pp. 463–76. 6 7 of North America. Cambridge University Press, Cam- 7 Woodburne MO, ed (2004). Late Cretaceous and Ceno- 8 bridge, pp. 583–8. 8 zoic Mammals of North America. University of Cali- 9 Wang X (1994). Phylogenetic systematics of the Hes- 9 fornia Press, Berkeley. 10 perocyoninae (Carnivora: Canidae). Bulletin of the 10 Wright DB (1998). Tayassuidae. In: Janis C, Scott KM, 11 American Museum of Natural History 221, 1–207. 11 Jacobs LL, eds. Evolution of Tertiary Mammals of 12 Wang X, Tedford RH, Taylor BE (1999). Phylogenet- 12 13 North America. Cambridge University Press, Cam- 13 ic systematics of the Borophaginae (Carnivora: Ca- 14 bridge, pp. 389–401. 14 15 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23 24 24 25 25 26 26 27 27 28 28 29 29 30 30 31 31 32 32 33 33 34 34 35 35 36 36 37 37 38 38 39 39 40 40 41 41 42 42 43 43 44 44 45 45 46 46 47 47 48 48 49 49 50 50 51 51