1 SUPPLEMENTARY MATERIAL 2 Contents 3 Taxon ranges and diversity estimates 4 Outcrop area estimates - Methods 5 Facies bias analysis 6 References 7 Table S1 – Facies analysis 8 Table S2 – European outcrop data 9 Fig. S1 – Relative abundance of cetacean occurrences per rock type 10 11 12 Taxon ranges and diversity estimates 13 The earliest cetaceans most likely evolved in the area of modern Pakistan and India about 55 14 million years ago during the early Eocene (Thewissen and Williams 2002), and started to 15 spread through Tethys and, ultimately, round the world, during the middle and late Eocene 16 (Fordyce 2002). Whilst the oldest sirenians reach a similar age (Domning 2002), 17 pinnipedimorphs are somewhat younger, with the oldest fossils dating back only to the late 18 Oligocene about 25-27 Ma (Berta 2002). Raw taxon counts for all three groups, sampled at 19 the genus level and recorded at the level of the geological stage were downloaded from the 20 Palaeobiology Database (http://paleodb.org) on the 05th October 2008, using the group taxon 1 1 names 'Cetacea', 'Pinnipedimorpha' and 'Sirenia', and the following parameters: continents or 2 paleocontinents = Europe, time fields = stage. A sampled-in-bin taxon counting protocol was 3 used, i.e. genera were not ranged through time intervals in which they had not actually been 4 sampled. 5 6 Outcrop area estimates 7 European outcrop area was estimated at the level of the geological epoch for marine 8 sedimentary rocks ranging in age from the Eocene to the Pleistocene, using the International 9 Geological Map of Europe and the Mediterranean Regions 2nd/3rd edition, scale 10 1:1.500.000, consisting of 45 separate sheets issued by the German Federal Institute for 11 Geosciences and Natural Resources and UNESCO, Hannover 1964-2000. In order to include 12 all potential European occurrences of marine mammals, outcrops ranging in age from the 13 Eocene to the present were recorded. Approximate outcrop area was obtained for the area 14 defined by 11°W (west coast of Ireland), 58°E (Ural mountains), 34°N (south coast of Crete) 15 and 72°N (North Cape/Knivskjellodden, Norway), with North Africa, Turkey, the southern 16 Caucasus and the area east of the Caspian Sea being excluded. The map was divided into 30 x 17 30 km squares aligned to 11°W. For every single square the presence or absence of marine 18 sedimentary rock of a given geological epoch, according to the Gradstein et al. (2004) 19 geological timescale, was scored. Where an outcrop spanned more than one epoch, rock was 20 scored as present for all the epochs covered, with the single exception of rock defined to be 21 ‘Tertiary and Cretaceous’ in nature, which was assumed to be mostly Palaeocene in age. 22 Molasse (post-orogenic deposits formed in depression zones either side of the Alps) was only 23 recorded where explicitly defined as marine. 24 In order to make fine-scale stage-level comparisons with the diversity estimates 25 possible, outcrop area data for a given epoch were apportioned to geological stages within 2 1 that epoch (Gradstein et al. 2004) according to their relative durations, following the 2 approach taken by Crampton et al. (2003). Whilst this procedure makes an a priori 3 assumption of constant sedimentation rates within epochs, which is unlikely to be accurate, it 4 has been speculated that this kind of error will not confound the overall pattern of outcrop 5 abundance (Crampton et al. 2003). However, because this study relies on an equal-grid 6 sampling method, the apportioned data had to be corrected for the fact that any given map 7 square could potentially contain outcrops belonging to more than one stage. To illustrate this, 8 suppose two stages of equal length (stages A and B) were represented by outcrops in ten map 9 squares each, with five squares being unique to either stage and a further five being shared. If 10 outcrop area could be scored at the level of the stage, ten map squares would thus be recorded 11 per stage. Yet, with outcrop area being recorded at the level of the epoch with no way of 12 distinguishing between stages, only 15 squares would be scored: five unique to stage A, five 13 unique to stage B and five in which rock of both ages occurred. If these data were then 14 apportioned to stages according to stage duration, as had been done in this paper, only 7.5 15 squares would be apportioned to either stage. Whilst this is not a problem within any given 16 epoch, since relative proportions should stay the same, this bias will lead to a distorting 17 boundary effect at epoch transitions, caused by the varying number of stages constituting 18 each epoch. For example, whilst the Oligocene consists of only two stages, the subsequent 19 Miocene comprises six. This means that the chance of any given map square containing rocks 20 belonging to more than one stage being apportioned to the Chattian (the upper stage of the 21 Oligocene) is considerably higher (a chance of 1 in 2) than such a square being apportioned 22 to the Aquitanian (the lowest stage of the Miocene, with a chance of 1 in 6), leading to an 23 overestimate of Oligocene outcrop area with respect to the Miocene, and thus distortion of 24 the overall record. In order to counteract this effect, the number of map squares apportioned 25 to each geological stage was multiplied by the total number of stages within the epoch the 3 1 stage was part of. Whilst this approach is oversimplifying the matter in assuming that map 2 squares containing rock belonging to more than one stage will always include all stages of 3 that epoch, the correction afforded by this should restore the overall pattern of changes in 4 outcrop area through time. 5 6 Facies bias analysis 7 When analysing groups of taxa as small as a mammalian order, the level of lithological 8 resolution of the diversity and outcrop data used may become important. Unlike estimates of 9 global diversity, smaller clades may exhibit associations with certain types of rock facies, 10 making their stratigraphic distribution non-random (Holland 1995; Marshall 1997). 11 Intriguingly, in term of marine animals this type of facies control has been found to apply not 12 only to relatively immobile benthic (Dodd and Stanton 1990), but also to a number of 13 planktonic and nektonic clades, such as ammonites (Bayer and McGhee 1985) or graptolites 14 (Lenz and Xu 1985). 15 In order to test whether any of the three groups of marine mammals was associated 16 with a particular rock facies, lithological classifications of the deposits yielding marine 17 mammal fossils around the world were downloaded for the Eocene, Oligocene, Miocene and 18 Pliocene from the Palaeobiology Database on the 24th September 2008, using the group 19 taxon names 'Cetacea', 'Pinnipedimorpha' and 'Sirenia', together with the following 20 parameters: continents or paleocontinents = global, lithologies = carbonate, mixed and 21 siliciclastic. These data were then adjusted for changes in rock type abundance through time 22 using the lithological abundance estimates of Ronov et al. (1980). 23 Analysis of the cetacean abundance data downloaded from the Paleobiology Database 24 showed that, in absolute terms, cetaceans occur more often in siliciclastic rocks than in 25 carbonates (Table S1). However, once the data had been adjusted for the varying proportions 4 1 of different rock facies, it became clear that, with exception of the Pliocene, carbonates are 2 relatively more abundant in cetacean occurrences (Figure S1), an observation normally 3 obscured by the overwhelming abundance of siliciclastic rocks during parts of the Caenozoic. 4 In addition, it turned out that rocks of mixed lithology provided the (relative) majority of 5 fossils during the Oligocene and the Pliocene. The extent to which carbonates are more 6 productive with respect to siliciclastics varied with epoch, with the smallest difference 7 observed during the Miocene (Fig. S1). Unfortunately there were not enough data available to 8 allow a reliable analysis of the pinnipedimorph and sirenian data. However, pinnipedimorphs 9 seems to occur mainly in siliciclastic deposits, whereas sirenians appeared to show a shift 10 from carbonates during the Eocene and Oligocene to siliciclastic deposits during the Miocene 11 and Pliocene (Table 1). 12 The admittedly rather crude facies analysis of the cetacean abundance data performed 13 here suggested a relatively greater abundance of cetaceans in carbonates than in siliciclastics. 14 However, the difference between these different types of rock was rather variable and 15 decreased to a more or less negligible extent during the Miocene. Moreover, even though 16 carbonates seemed to yield relatively more cetacean fossils, their low abundance during the 17 Caenozoic led to the overprinting of this abundance signal by the relatively less abundant, but 18 absolutely much more numerous fossils found in siliciclastic rocks. For these reasons, the 19 exclusion of neither type of rock from this analysis seems justified, as exclusion of the 20 siliciclastics would mean dumping most of the data, whilst the exclusion of carbonates would 21 target the more productive type of rock. This conclusion can be extended to the putative 22 trends observed pinnipedimorphs and sirenians. In terms of the former, occurrences seem to 23 be concentrated in the most abundant sediment type, anyway, whilst for the latter both 24 carbonates and siliciclastic deposits seem to be important. Thus, whilst the relative abundance 5 1 of cetaceans in carbonates and the putative trends in pinnipedimorphs and sirenians warrant 2 further research, it was decided to keep both types of rock in the analysis for the time being. 3 4 References 5 Bayer, U. and McGhee, G. R. 1985 Evolution in marginal epicontinental basins: the role of 6 phylogenetic and ecologic factors (ammonite replacements in the German Lower and 7 Middle Jurassic). In Bayer, U. and Seilacher, A. (eds.) Sedimentary and Evolutionary 8 Cycles (eds U. Bayer and A. Seilacher). New York: Springer. 9 10 Berta, A. 2002 Pinniped Evolution. In Encyclopedia of Marine Mammals (eds W. F. Perrin, B. Würsig and J. G. M. Thewissen). San Diego: Academic Press. 11 Crampton, J. S., Beau, A. G., Cooper, R. A., Jones, C. A., Marshall, B. and Maxwell, P. A. 12 2003 Estimating the rock volume bias in paleobiodiversity studies. Science 301, 358- 13 360. 14 15 16 Dodd, J. R. and Stanton, R. J. 1990 Paleoecology: concepts and applications. New York: Wiley. Fordyce, R. E. 2002. Cetacean Evolution. In Encyclopedia of Marine Mammals (eds W. F. 17 Perrin, B. Würsig and J. G. M. Thewissen). San Diego: Academic Press. 18 Gradstein, F. M., Ogg, J. G. and Smith, A. G. et al. 2004 A geologic time scale 2004. 19 20 Cambridge: Cambridge University Press. Holland, S. M. 1995 The stratigraphic distribution of fossils. Paleobiology 21, 92-109. 6 1 2 3 4 Lenz, A. C. and Xu, C. 1985 Graptolite distribution and lithofacies: some case histories. J. Paleontol. 59, 636-642. Marshall, C. R. 1997 Confidence intervals on stratigraphic ranges with non-random distributions of fossil horizons. Paleobiology 23, 165-173. 5 Miller, K. G., Kominz, M. A., Browning, J. V., Wright, J. D., Mountain, G. S., Katz, M. E., 6 Sugarman, P. J., Cramer, B. S., Christie-Blick, N. and Pekar, S. F. 2005 The 7 Phanerozoic record of global sea-level change. Science 310, 1293-1298. 8 9 Ronov, A. B., Khain, V. E, Balukhovsky, A. N. and Seslavinsky, K. B. 1980 Quantitative analysis of Phanerozoic sedimentation. Sediment. Geol. 25, 311-325. 10 Thewissen, J. G. M. and Williams, E. M. 2002 The early radiations of Cetacea (Mammalia): 11 evolutionary pattern and developmental correlations. Annu. Rev. Ecol. Syst. 33, 73-90. 12 13 14 15 16 17 18 19 7 lithology carbonate siliciclastic mixed Eocene 7.3/11.53 28.6/45.18 27.4/43.29 Oligocene 4.6/7.85 50.2/85.67 3.8/6.49 Miocene 5.4/11.20 32.6/67.64 10.2/21.16 Pliocene 1.3/4.25 28.1/91.83 1.2/3.92 cetacean fossil occurrences carbonate siliciclastic mixed 24 67 27 4 24 7 63 350 33 1 94 9 pinnipedimorph fossil occurrences carbonate siliciclastic mixed n/a n/a n/a 0 3 0 5 102 1 0 40 1 sirenian fossil occurrences carbonate siliciclastic mixed 14 7 5 5 10 5 3 80 2 1 18 1 abundance of lithologies total/relative 1 2 Table S1 Total and relative abundances of different lithologies and cetacean fossil 3 occurrences. The first number in each cell in the ‘abundance of lithologies’ block shows the 4 abundance of the three rock types under study as a percentage of global rock volume (Ronov 5 1980; the missing percent are comprised of coal-bearing strata, evaporates, volcanics and 6 terrestrial lithologies), whilst the second number shows the relative abundance of these three 7 rock types only. Below this the total number of marine mammal occurrences per rock type 8 and epoch as downloaded from the Paleobiology Database are shown. 9 10 11 12 13 14 15 8 midpoint (Ma) stage duration (Ma) % epoch total rock (grid squares) Ypresian 52.2 7.2 32.9 2095.0 0 70.6 Lutetian 44.5 8.2 37.4 2381.6 0 3 58.00 Bartonian 38.8 3.2 14.6 929.7 2 3 18.0 Priabonian 35.5 3.3 15.1 961.6 0 0 4 35.9 Rupelian 31.2 5.5 50.5 1160.5 3 1 1 10.0 Chattian 25.7 5.4 49.5 1137.5 8 1 2 -2.2 Aquitanian 21.7 2.6 14.7 1388.3 14 0 3 -4.1 Burdigalian 18.2 4.5 25.4 2398.8 28 3 5 6.8 Langhian 14.8 2.3 13.0 1227.7 7 2 0 1.2 Serravallian 12.6 2 11.3 1067.2 28 10 0 5.1 Tortonian 9.4 4.4 24.9 2349.8 45 8 1 4.1 Messinian 6.2 1.9 10.7 1010.5 0 1 0 -2.6 Zanclean 4.5 1.7 48.6 2363.4 14 7 1 -6.1 Piacenzian 3.1 1 28.6 1390.8 22 4 0 -14.6 Gelasian 2.2 0.8 22.9 1113.6 18 2 1 -18.7 Pleistocene 0.9 1.8 100 833.0 11 6 0 -52.9 stage Cetacea Pinnipedimorpha Sirenia sea-level (m) 1 2 Table S2 Table showing absolute and relative stage duration (% epoch), European outcrop 3 area as observed in this study (total rock), as well as marine mammal diversities and global 4 sea-level change (Miller et al. 2005). Marine mammal diversities are given as genera counts. 5 All data are shown in their raw, non-detrended and non-transformed form. 6 7 8 9 10 11 9 1 Fig. S1 Relative abundance of cetacean occurrences per rock type (in %) after the varying 2 amounts of available rock have been accounted for. This number was calculated in two steps, 3 first by dividing the total number of occurrences by the relative abundance of each lithology, 4 and then calculating the relative proportion of the resulting number as a percentage per 5 epoch. Because there were not enough data available, no attempt was made to calculate 6 relative abundances of pinnipedimorph and sirenian occurrences. 7 10