Sex and the shifting biodiversity dynamics of marine in deep time

Andrew M. Busha,1, Gene Huntb, and Richard K. Bambachb

aDepartment of Ecology and Evolutionary Biology and Center for Integrative Geosciences, University of Connecticut, Storrs, CT 06269-3043; and bDepartment of Paleobiology, National Museum of Natural History, Smithsonian Institution, Washington, DC 20013-7012

Edited by Michael Foote, University of Chicago, Chicago, IL, and accepted by Editorial Board Member David Jablonski October 6, 2016 (received for review June 30, 2016)

The fossil record of marine animals suggests that diversity-dependent (OU-OU), in which diversity approaches and then fluctuates processes exerted strong control on biodiversification: after the around an equilibrium value]. We also tested two-phase models Ordovician Radiation, richness did not trend for hundreds of incorporating a shift between simple models. These methods can millions of years. However, diversity subsequently rose dramatically in rigorously identify shifts in the pattern of diversity history that the Cretaceous and Cenozoic (145 million years ago–present), indicat- suggest changes in underlying dynamics. ing that limits on diversification can be overcome by ecological or evolutionary change. Here, we show that the Cretaceous–Cenozoic Fertilization Mechanism radiation was driven by increased diversification in animals that trans- Fertilization mechanisms have never been considered explicitly fer sperm between adults during fertilization, whereas animals that in analyses of marine fossil biodiversity. However, they have broadcast sperm into the water column have not changed significantly been extensively discussed in the case of land plants, with ∼ pollination proposed as a key factor in the K-Cz radiation of in richness since the Late Ordovician ( 450 million years ago). We – argue that the former group radiated in part because directed sperm angiosperms (14 18). Pollination of plants by animals, as op- transfer permits smaller population sizes and additional modes of posed to wind, is postulated to permit additional modes of pre- zygotic reproductive isolation, which could enhance speciation in prezygotic isolation, as has been argued previously for the coincident conjunction with other isolation mechanisms (19). This effect radiation of angiosperms. Directed sperm transfer tends to co-occur may be strongest when pollinator and plant are closely with many ecological traits, such as a predatory lifestyle. Ecological coevolved, limiting gene flow among species, as in orchids (20, specialization likely operated synergistically with mode of fertilization 21). Pollination could also permit increased rarity because un- in driving the diversification that began during the Mesozoic marine common, widely spaced individuals may still reproduce as long as revolution. Plausibly, the ultimate driver of diversification was an in- pollinators are present (18). Interestingly, the greater diversity in crease in food availability, but its effects on the fauna were regulated terrestrial relative to aquatic habitats has also been attributed in by fundamental reproductive and ecological traits. part to differences related to fertilization. For example, motile marine animals do not commonly transmit the gametes of sessile diversity dependence | Phanerozoic | Mesozoic marine revolution | species, as pollinators do on land (22–24). Also, chemical and fertilization | copulation visual signals are transmitted more effectively in air, which could enhance sexual selection and speciation; it could also enable he fossil record provides direct, historical evidence that can animals to find mates at greater distances, which could permit Thelp constrain controls on biodiversification, such as the greater rarity (22, 23). relative roles of diversity-dependent and diversity-independent Significantly, marine species also vary widely in fertilization processes (1–3). Statistical analyses of the well-preserved fossil re- strategy. At a broad scale, these strategies are phylogenetically cord of marine animals support diversity dependence, albeit with a shifting value of equilibrium diversity (4-6, but see refs. 7 and 8 for Significance alternate opinions). Recent analyses suggest that such shifts might be quite substantial: after not trending appreciably for 300 million Fertilization mechanisms explain broad patterns in the taxonomic years (Late Ordovician–Jurassic; ∼458–145 MA), global genus distribution of diversity in marine animals, as argued previously richness approximately doubled in the Cretaceous and Cenozoic (9) for land plants. We argue that fertilization via copulation (or some (Fig. 1A). Here, we probe the fossil record for insights into this other significant interaction among adults) permits additional massive shift in global diversity and present a theory on the un- mechanisms of reproductive isolation and smaller population sizes derlying evolutionary and ecological dynamics. We show that the relative to fertilization via the broadcasting of sperm into the wa- Cretaceous–Cenozoic (K-Cz) radiation was highly selective, with ter, thus enhancing diversification potential. Maximum-likelihood diversity increase concentrated in animals that copulate or other- modeling of paleontological data indicates that the diversity of wise transfer sperm between interacting adults. In contrast, the sperm broadcasters has been limited by diversity-dependent fac- aggregate diversity of animals that broadcast sperm into the water tors for the last 450 million years. In contrast, animals that copu- column has not increased appreciably since the Ordovician. We late, etc., were also limited until the Cretaceous and then radiated argue that directed sperm transfer promoted diversification by dramatically, coincident with apparent major changes in marine permitting additional modes of prezygotic isolation, smaller pop- productivity. Fertilization may have acted synergistically with ulation sizes, and greater ecological specialization. Previous work ecological specialization in promoting diversification, particularly EVOLUTION has cited increased productivity as a driver of the K-Cz radiation in predators. (10–12), but we argue that synergistic interactions between mating strategy and ecology modulated the effects of environmental change Author contributions: A.M.B., G.H., and R.K.B. performed research, analyzed data, and on biodiversification. wrote the paper. To characterize diversity history, we generated sampling-stan- The authors declare no conflict of interest. dardized genus richness curves from the Paleobiology Database This article is a PNAS Direct Submission. M.F. is a Guest Editor invited by the Editorial (PBDB; paleobiodb.org) and used maximum likelihood methods Board. (13) to model diversity history. Models of diversity change include 1To whom correspondence should be addressed. Email: [email protected]. random walk, directional change, and two models that are consis- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. EARTH, ATMOSPHERIC,

tent with diversity dependence [stasis and Ornstein-Uhlenbeck 1073/pnas.1610726113/-/DCSupplemental. AND PLANETARY SCIENCES

www.pnas.org/cgi/doi/10.1073/pnas.1610726113 PNAS | December 6, 2016 | vol. 113 | no. 49 | 14073–14078 Downloaded by guest on September 27, 2021 conserved in most higher taxa of Metazoa that are abundant as A All Metazoa fossils, so they can be inferred reliably based on taxonomic af- filiation (SI Appendix, Reproductive Faunas). Our first group 1000 (broadcasters) includes animals that broadcast sperm into the water column. Some species broadcast eggs, whereas others re- 800 tain them, but we include both in this group because sperm broadcasting is sufficient to limit interactions during mating. As 600 in wind-pollinated plants, fertilization depends on the vicissi- tudes of transport in a fluid, and fertilization potential decreases dramatically over short distances (25, 26). Thus, individuals often 400 produce copious gametes, and fertilization is enhanced in large, dense populations. Species recognition during reproduction oc- curs only between sperm (or spermatophore) and egg (or mother) (27), and adult–adult interactions are limited to factors such as timing of gamete release. Broadcasting animals include bivalves, brachiopods, bryozoans, cnidarians, sedentary echino- 500 400 300 200 100 0 derms, and several less diverse groups (SI Appendix, Reproductive 600 Faunas). Most broadcasters are nonmotile or not regularly mo- B Nonbroadcasters tile and are suspension feeders, as a sedentary lifestyle precludes 400 many other feeding modes. Most fossil cnidarians are corals, many of which capture small living organisms from the water as food (microcarnivory) and may house photosymbionts. 200 Other animals (nonbroadcasters) transfer sperm directly be- tween adults via copulation or some other interaction (e.g., the

s male fertilizing eggs as the female releases them). Whether

es 100 fertilization is internal or external, adults must recognize each 80 other and interact directly, and reproductive isolation is driven ichn not just by gametic incompatibility, habitat separation, and the like R 60 but also by behavioral and anatomical barriers at the adult level – 40 (e.g., courtship rituals, physical inability to copulate) (28 32). Behavioral barriers to gene flow and sexual selection are generally

Genus believed to facilitate speciation (15, 33–35), and reduced Allee d

e 500 400 300 200 100 0 effects may permit species to survive, even if rare. To the extent z i that animals are adept at recognizing and mating with members of d 600 C Broadcasters their own species, they would be analogous to orchids, where 400 pollen delivery is more species-specific than with many other plants (20, 21). Even if abundance or biomass is limited by re- Standar 200 sources, a change in the predominant reproductive strategy from broadcasting to nonbroadcasting could permit greater numbers of 100 more specialized species characterized by smaller populations. 80 Adults must find, recognize, and interact with each other directly, 60 so nonbroadcasters tend to be motile, with well-developed senses. 40 Feeding modes vary, but many species are predators, omnivores, or scavengers. Major taxa include arthropods, cephalopods, ver- 20 tebrates, and some groups of gastropods (Caenogastropoda, Neritopsina, Cocculinoidea, and Heterobranchia) (36). The final group (intermediate taxa) is composed of mobile 10 animals that broadcast sperm or that belong to taxa in which broadcasting is plesiomorphic and common. Given their mobility, 500 400 300 200 100 0 many of these species aggregate before spawning to increase fer- – – Intermediate tilization success (37 39), potentially introducing adult adult inter- 100 D actions unavailable to more sedentary broadcasters. Some living 80 species exhibit interactions similar to nonbroadcasters, such as 60 mounting during spawning (40–43), or even direct sperm transfer 40 (43–47). Such variation would be impossible to identify in the fossil record, so this category is of necessity more variable than the other 20 two, although most members do broadcast gametes. The interme- diate group includes noncopulating gastropods, several minor SI Appendix, Reproductive 10 molluscan taxa, and mobile ( 8 Faunas). Feeding modes vary and include suspension feeding, 6 ODCPTrJKS Pg N Paleozoic Mesozoic Cen on 1,000 SQS replicates. SQS quorum was 0.70 for A and 0.50 for B–D. Black lines mark the trajectory of expected genus richness according to the pre- 500 400 300 200 100 0 ferred model, as determined by maximum likelihood methods (A and B: Time (MA) stasis followed by directional trend; C: OU-OU; D: random walk). These data are plotted on arithmetic axes in SI Appendix, Fig. S1, and values are listed in Fig. 1. Sampling-standardized genus richness of marine animals on a SI Appendix, Tables S1 and S2. C, Carboniferous; Cen, Cenozoic; D, Devonian; logarithmic scale. (A) All animals, (B) nonbroadcasters (copulators, etc.), J, Jurassic; K, Cretaceous; N, Neogene; O, Ordovician; P, Permian; Pg, Paleogene; S, (C) broadcasters, and (D) intermediate taxa. Error bars mark ± one SE based Silurian; Tr, Triassic.

14074 | www.pnas.org/cgi/doi/10.1073/pnas.1610726113 Bush et al. Downloaded by guest on September 27, 2021 deposit feeding, grazing, and predation, although the number of 80 predators is small relative to the nonbroadcasters. PBDB Sepkoski Results 60 As shown previously (9), sampling-standardized genus richness of the total marine fauna fluctuated without trending from the

A ent of Genera)

Ordovician to mid-Mesozoic, then increased (Fig. 1 , shown on c log scale). Maximum likelihood modeling supports this inter- 40 pretation; the best model was stasis with a shift to a directional, s(Per increasing trend (SI Appendix,TableS3). The most likely position r for the shift in dynamics was Jurassic 6 (Tithonian; midpoint ∼149 20 Ma), although a shift in the next bin, Cretaceous 1 (Berriasian- Valanginian; ∼139 Ma), was nearly as good (SI Appendix,Fig.S2). There was also some support for a two-phase OU-OU model with oadcaste 0 an upward shift in equilibrium diversity, also in Jurassic 6, to a -br S DCPTrJ KPg N much higher equilibrium diversity (3.34 log units, or ∼2,200 genera)

Non Paleozoic Mesozoic Cen that is not yet approached by the end of the Cenozoic (SI Appendix, Fig. S3 and Table S3). 400 300 200 100 0 Time (MA) Major Faunas. The diversity of nonbroadcasters was also described B SI Fig. 2. Percentage of animal genera classified as nonbroadcasters in the best by stasis followed by a directional trend (Fig. 1 and ’ Appendix, Table S4). The timing of the shift was somewhat un- PBDB (red) and Sepkoski s (48) compendium (black). For the PBDB, broad- certain; the most likely shift point between the models was casters and intermediate taxa were combined to get a continuous time series ∼ from SQS. The spike in the Early Triassic reflects the selective extinction of Cretaceous 2 (Hauterivian and Barremian; 129 Ma), but shifts broadcasters in the end-Permian extinction, and their subsequent recovery. as early as Jurassic 2 (Pliensbachian; ∼187 Ma) or as late as Cre- ∼ Nonbroadcasters radiate strongly in the K-Cz and increase permanently as a taceous 4 (Albian; 107 Ma) were also within two log-likelihood percentage of the fauna. units of the best solution (SI Appendix,Fig.S2). The preferred model for the broadcasters was OU-OU (Fig. 1C and SI Appendix, Table S5). Diversity rose during the Or- rather than an evolutionarily meaningful unit, but together these dovician to an equilibrium value that persisted for the rest of the taxa do not increase in diversity in the Cenozoic. Paleozoic, fell during the end-Permian mass extinction, and then The asteroids and caenogastropods were examined separately rose to a similar (although slightly lower) equilibrium. The in- because they provide unusual combinations of fertilization and termediate group was not diverse or well-sampled enough to pro- feeding method. The caenogastropods are nonbroadcasters, but D duce a continuous standardized diversity curve (Fig. 1 ). From the include a wide range of feeding methods (tabulated from ref. 51). available data, their richness increased during the Triassic, but then Although the carnivorous caenogastropods radiated strongly in flattened out, never significantly exceeding levels reached in the the K-Cz, the noncarnivores did not (Fig. 3L). Asteroids, which Paleozoic (Fig. 1D). The random walk model received the most – – are broadcasting predators, declined as a proportion of the total support, although random walk stasis and stasis stasis (with the fauna during the Cenozoic in Sepkoski’s data (Fig. 3M), although shift most likely occurring in Triassic 1 in both cases) also received their fossil record is poor and should be interpreted with some some support, but were not favored by the likelihood ratio test (SI Appendix caution (52). The PBDB curve was extremely incomplete, but ,TableS6). also showed no radiation (SI Appendix, Table S8). Relative richness trends in Sepkoski’s (48) compendium are similar to those in the PBDB (Fig. 2). Nonbroadcasters com- Discussion prised about 20–40% of genera during the mid-Paleozoic to mid- Several analyses of the marine animal fossil record have detected Mesozoic, with some fluctuations. They climbed in proportional diversity dependence (4–6), but there is also strong evidence of a richness by about 20%, beginning in the Cretaceous and con- tinuing in the Cenozoic, indicating they were diversifying at a major K-Cz radiation (9, 53, 54). Ouranalysesprovidestatistical greater rate than the other groups. support for these patterns: the optimal model implies stasis in di- versity from the Ordovician to the Jurassic, consistent with diversity dependence, followed by a directional increase in the Cretaceous Individual Taxa. Most nonbroadcasting taxa have diversity histo- A ries that are consistent with the whole group: relatively stable and Cenozoic (Fig. 1 ). The OU-OU model also received some – support (SI Appendix,TableS3),whichwouldsuggesttheK-Cz richness in the late Paleozoic early Mesozoic, followed by di- cf versification in the late Mesozoic and Cenozoic (Fig. 3 A–C and radiation might have slowed in the late Cenozoic ( ref. 55). SI Appendix, Table S7). The histories of these groups in the early The richness of nonbroadcasters was also in stasis from the Paleozoic differ, however; for example, arthropods declined as Ordovician through Jurassic or Cretaceous, when it too shifted to B the trilobites dwindled. Cephalopods differ from the other a directional trend, leading to a tripling in richness (Fig. 1 ). All nonbroadcasters because they lost most of their fossil diversity in nonbroadcasters except the cephalopods (shelled forms of which the end-Cretaceous extinction (Fig. 3D). were largely eliminated by the end-Cretaceous extinction) con- The diversity curves of individual broadcasting and intermediate tributed to the K-Cz radiation, transitioning from nontrending to taxa vary widely (Fig. 3 E–K). Bivalves increased at a fairly con- increasing richness in the later Mesozoic (Fig. 3 A–D). E In contrast, broadcasters have essentially maintained a stable stant rate from the Ordovician onward (Fig. 3 ) (49), with per- EVOLUTION haps a slight uptick after the end-Cretaceous extinction (50). equilibrium since the Ordovician, despite fluctuations (Fig. 1C). Bryozoans declined in the end-Permian extinction, but recovered Their richness was severely depressed by the end-Permian ex- during the Mesozoic, although the details are poorly constrained tinction, but they returned to a remarkably similar equilibrium in the PBDB data (Fig. 3G). Brachiopods declined after the Pa- afterward and did not radiate in the K-Cz. The intermediate leozoic (Fig. 3F), and cnidarians, echinoderms, and noncopulating fauna also did not radiate during this time (Fig. 1D). Individual gastropods fluctuated through time (Fig. 3 H–J). Several groups, broadcaster and intermediate taxa show a range of diversity tra- mostly sponges and minor molluscan taxa (SI Appendix, Table S7), jectories (Fig. 3 E–K).Thefactthatsometaxaincreaseindiversity were combined because none were represented by sufficient data does not obviate the point that broadcasters, in the aggregate, be- to generate reasonably complete individual curves (Fig. 3K). This haved differently than nonbroadcasters; because there is turnover EARTH, ATMOSPHERIC,

polyphyletic group simply represents the remainder of the data, among taxa within the broadcasters, some individual taxa must AND PLANETARY SCIENCES

Bush et al. PNAS | December 6, 2016 | vol. 113 | no. 49 | 14075 Downloaded by guest on September 27, 2021 A Arthropoda B Copulating Gastropoda C Vertebrata D Cephalopoda 100

10

1 E F Brachiopoda G Bryozoa H Cnidaria 100

10

1 Standardized Genus Richness I Echinodermata J Other Gastropoda K Other Taxa LCaenogastropoda M Asteroidea

100 ) %

( 2

10 1 us Richness n

1 Ge 0 500 400 300 200 100 0 500 400 300 200 100 0 500 400 300 200 100 0 100 0 100 0 Time (MA)

Fig. 3. Sampling-standardized genus richness of individual higher taxa or groups of taxa. (A–D) Nonbroadcasters (red). (E–K) Broadcasters and intermediate taxa (blue). Error bars mark ± one SE based on 1,000 replicates; SQS quorum = 0.40. Black lines mark lowess regressions (smoothing parameter = 0.4). (L) Caenogastropods, which are nonbroadcasters, split into carnivores (thick solid line) and noncarnivores (thin dashed line) (SQS quorum = 0.25). Values are shown for the K-Cz because dietary assignments are less clear-cut for some earlier genera. (M) Proportional richness of asteroids, which are broadcasting predators, in Sepkoski’s data (48).

increase as others decrease. Critically, no broadcaster/intermediate diversity history of animals that are carnivorous broadcasters or taxon clearly shows a shift in diversification rate in the late Meso- noncarnivorous nonbroadcasters. Few such fossil taxa are pre- zoic similar to those observed for the nonbroadcasters. served and studied sufficiently for analysis, but three examples Notably, the broadcaster and nonbroadcaster faunas do not are highly suggestive. Caenogastropods are nonbroadcasters that correspond to Sepkoski’s evolutionary faunas (55), as members vary widely in diet; the noncarnivorous forms did not radiate in of his Paleozoic and Modern faunas belong to both categories. the K-Cz, whereas carnivorous forms diversified strongly (Fig. Sepkoski’s Paleozoic and Modern faunas largely correspond to 3L). The Asteroidea are broadcasting predators, and they did those groups that fared poorly and well in the end-Permian ex- not increase in relative richness in the K-Cz (Fig. 3M), although, tinction, whereas mass extinctions contributed to turnover again, their record should be viewed with caution. Finally, corals within, rather than between, our groups. are broadcasters that exhibit carnivory or microcarnivory, al- Nonbroadcasters comprised 20–40% of genera from the mid- though many also get nutrition from photosymbiosis, and they Paleozoic to mid-Mesozoic, then rose to 55–65% in the later likewise did not radiate (Fig. 3H). Cenozoic (Fig. 2). In the living marine fauna, nonbroadcasters Together, these results suggest that diversification was con- account for ∼73% of genera according to data from the World centrated in animals that combined direct sperm transfer with Registry of Marine Species (marinespecies.org)(SI Appendix, carnivory. Further testing would be useful; for example, arthro- Table S9). Thus, the selective nature of the K-Cz radiation was a pods and fish use a variety of feeding modes and could be ana- major determinant of the composition of the living marine biota. lyzed after additional data collection and ecological analysis.

Fertilization vs. Feeding. Mode of fertilization tends to covary with The Mesozoic Sexual Revolution. Our results, in concert with pre- a number of ecological traits, including feeding, motility, and vious demonstrations (4–6), suggest that diversification in marine cephalization, so it is possible that the radiation documented animals was limited by diversity-dependent processes from the here in nonbroadcasters was causally related to one of these mid-Paleozoic to mid-Mesozoic, at which point the global equi- other variables. The most obvious candidate is feeding mecha- librium diversity increased dramatically (Fig. 1A). This radiation nism: broadcasters, especially sedentary broadcasters, are pri- was driven by increased rates of diversification in nonbroadcasters marily suspension feeders, whereas the nonbroadcasters include (Figs. 1B and 3 A–C), particularly carnivorous forms (Fig. 3L). In many predators and scavengers. There is considerable evidence theory, radiations in the arthropods, copulating gastropods, and for increased diversity of predators and increased intensity of vertebrates could have been driven by independent, unrelated fac- predation in the late Mesozoic and Cenozoic (which correspond tors, such as the evolution of key innovations. However, the coin- to our nonbroadcaster radiations) (56–59), and these radiations cident timing of these shifts in diversification suggests a common have been linked to increased food supply to marine animal food environmental driver. The evolution of key innovations may be webs (10–12). related to the radiation of particular subgroups within these taxa, To distinguish whether feeding or fertilization mechanisms such as the teleost fish. However, we would argue that chang- more closely correlate to diversification, one can examine the ing environmental conditions drove these radiations, with key

14076 | www.pnas.org/cgi/doi/10.1073/pnas.1610726113 Bush et al. Downloaded by guest on September 27, 2021 innovations permitting particular subgroups to capitalize on Conclusions the situation. A novel morphology only permits radiation if After not trending for hundreds of millions of years, marine ani- environmental conditions are permissive. mal diversity doubled in the K-Cz, but this radiation was highly Food is often discussed as the limiting resource that drives selective. Animals that transfer sperm from adult to adult, whether diversity dependence (1), and indeed, numerous paleontologists internally or externally, tripled in genus richness, whereas animals have argued that increasing food and/or oxygen availability in the that broadcast sperm did not radiate significantly. By analogy with late Mesozoic permitted ecological reorganization, including similar arguments about land plants, we propose that adult–adult diversification (10–12, 59–61). As animals diversified, they also fertilization facilitated diversification by permitting smaller pop- became, on average, larger, fleshier, more motile, and more ulation sizes and/or allowing additional mechanisms of prezygotic energetic (10, 58, 62, 63). Predation increased (57–60), and isolation. In permitting diversification, fertilization mechanism several groups of large phytoplankton radiated (64). All these may have operated synergistically with covarying ecological traits; lines of evidence suggest a widening of the base of the food chain many nonbroadcasting taxa are predators, which can specialize during the Mesozoic, and the ultimate drivers of increased pri- ecologically in ways not available to suspension-feeding, broad- mary production have been discussed by numerous authors (10– casting taxa. The fossil record of marine animals indicates that 12, 62, 64–66). long-standing limits to diversification can be overcome, quite However, the effects of increased food and/or oxygen avail- possibly by changes in resource availability, and the effects of these ability on diversity dynamics were strongly modulated by re- shifts on the fauna are controlled by fundamental aspects of re- productive biology and ecology, with nonbroadcasters possessing productive biology and ecology. several traits that predisposed them to radiation when conditions permitted. Direct sperm transfer permitted additional mecha- Methods nisms for speciation to occur, or for incipient species to differ- Data were downloaded from the PBDB following previous work (9) (SI Ap- entiate. It also permitted smaller viable population sizes (i.e., pendix, Paleobiology Database). Exceptional modes of preservation that are greater rarity), which permitted the co-occurrence of a greater concentrated in particular time intervals were excluded (e.g., silicification, original aragonite; refs. 9 and 71). All marine animals were included except number of species, given the available food resources (cf. ref. 8). for annelids, which add little diversity in the fossil record, and for which The proliferation of a greater number of less abundant species functional assignments were uncertain in some cases. was also enhanced because these animals (largely predators, Sampling was standardized using shareholder quorum subsampling (SQS), scavengers, and omnivores) have many ways to specialize eco- following previous works (9, 72), except that a single collection was drawn per logically (e.g., different types of prey, feeding at different times literature reference, and the results were not smoothed using the 3T correction of day). In contrast, most broadcasters are suspension feeders, to maintain independence among data points. One thousand iterations were which take small particles of food from the water and have fewer run per analysis, and the results from each iteration were logged before analysis ways to specialize. We suggest that reproductive biology and (results were similar for raw values, but we concentrate on logged values be- ecology operated synergistically: motile predators had a greater cause the slopes of the diversity curves are equivalent to rate of diversification, ability to ecologically specialize, and direct sperm transfer per- and changes in slope indicate changes in rate). Metazoa as a whole were ana- mitted smaller population sizes that allowed specialized taxa lyzed at a sampling quorum of 0.70, and the reproductive faunas at a quorum of 0.50. These levels were chosen to balance the trade-off between sampling cov- to survive. Before the Cretaceous, there were presumably in- erage and the temporal completeness of the results. sufficient food resources to support large numbers of specialized We analyzed the temporal pattern of change in the standardized diversity predators, counteracting the inherent macroevolutionary po- curves using the maximum likelihood methods of Hunt et al. (13). Models were tential of these nonbroadcasters to diversify. fit that corresponded to stasis, directional change, unbiased random walk, an Broadcasters, in contrast, lacked the traits that facilitated explo- OU-OU process, as well as two-phase models incorporating a switch between sive diversification in the nonbroadcasters. Broadcasting fertiliza- simple models. The OU model conforms closely to the expectations of diversity tion favors large, denser populations because small populations may dependence because diversity approaches and then fluctuates around a stable not be reproductively viable. Thus, available food resources may be equilibrium value. Stasis is also consistent with diversity dependence, and it consumed by a smaller number of more abundant species. Also, as could represent the latter portion of an OU process (i.e., sampling begins after noted, suspension feeders have fewer ways to specialize in diet and the shift to the stable value). A linear directional increase in logged data could feeding than predators. The intermediate group, which is more correspond to exponential diversification, to the initial portion of an OU process (before the stable attractor is reached), or to some other pattern that is variable ecologically, also did not diversify appreciably through time C indistinguishable given the resolution of the data. Model support was evalu- (Fig. 1 ). Although they did not expand in diversity, these groups ated initially using AICc (Akaike information criterion with correction for finite persisted as successful components of Cretaceous and Cenozoic sample size). Hunt et al. (13) found that AICc tended to be overly liberal in ecosystems, maintaining the diversity they achieved in the Paleozoic. favoring complex models and used parametric bootstrapping as a more con- These mechanisms are not inconsistent with other ideas on servative test of when complex models were favored over simple ones, an ecosystem evolution during the Mesozoic and Cenozoic. For approach we follow here (SI Appendix, Maximum Likelihood Modeling). example, an increase in primary production may have facilitated We also generated standardized diversity curves for major taxonomic an increase in the average biomass of suspension feeders, which groups within the Metazoa. These groups largely correspond to phyla, al- provided greater food resources for predators (10, 11). If so, though the mollusks were split because they vary in fertilization method and however, this increased biomass did not translate into elevated the data were sufficient for separate analyses. Other phyla are more ho- diversity. Increased predation and seafloor disturbance may have mogeneous with respect to reproduction. Most of these curves had temporal helped drive down the diversity and abundance of some taxa, gaps (time bins that did not have enough data to subsample at the given – quorum), so we highlighted long-term patterns using lowess regression, such as brachiopods (67 69), although competition may also rather than formal modeling. have been important (70), consistent with diversity dependence. We focus our analyses on the PBDB because diversity curves can be stan- Biodiversity dynamics are often studied through the lens of the dardized for various biases, but we also tabulated the diversity of the fertilization EVOLUTION living biota; for example, inferred from phylogenies generated faunas in Sepkoski’s compendium (48) to show that the results are consistent. from living organisms. However, living organisms represent a However, we only examined proportional diversity for Sepkoski’s data because single time slice in the history of life, and they may have evolved of concerns about heterogeneous sampling among time bins. under conditions not typical of its entirety. In the case of marine animals, the living fauna evolved largely during conditions of ACKNOWLEDGMENTS. For helpful discussions and comments, thanks to exceptional diversification, whereas much of the history of ma- C. Marshall, D. Jablonski, A. Knoll, J. Valentine, P. Getty, and two anonymous reviewers. Thanks to J. Alroy for providing SQS computer code. Thanks to rine life was characterized by stasis and diversity limitation. In- PBDB data enterers, including W. Kiessling, A. Hendy, M. Clapham, A. Miller, ferences based on living organisms may not reflect the full range M. Foote, J. Alroy, M. Aberhan, M. Kosnik, M. Patzkowsky, and P. Wagner.

of processes that have shaped the biosphere through time. This is Paleobiology Database publication 272. EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES

Bush et al. PNAS | December 6, 2016 | vol. 113 | no. 49 | 14077 Downloaded by guest on September 27, 2021 1. Rabosky DL, Hurlbert AH (2015) Species richness at continental scales is dominated by 38. Shepherd SA (1986) Studies on southern Australian abalone (genus Haliotis). Mar Biol ecological limits. Am Nat 185(5):572–583. 90(2):231–236. 2. Harmon LJ, Harrison S (2015) Species diversity is dynamic and unbounded at local and 39. Young CM, Tyler PA, Cameron JL, Rumrill SG (1992) Seasonal breeding aggregations continental scales. Am Nat 185(5):584–593. in low-density populations of the bathyal echinoid Stylocidaris lineata. Mar Biol 3. Marshall CR, Quental TB (2016) The uncertain role of diversity dependence in species 113(4):603–612. diversification and the need to incorporate time-varying carrying capacities. Philos 40. Run J-Q, Chen C-P, Chang K-H, Chia F-S (1988) Mating behavior and reproductive cycle Trans R Soc Lond B Biol Sci 371(1691):20150217. of (Echinodermata: Asteroidea). Mar Biol 99(2):247–253. 4. Alroy J (2008) Colloquium paper: Dynamics of origination and extinction in the ma- 41. Tominaga H, Nakamura S, Komatsu M (2004) Reproduction and development of the rine fossil record. Proc Natl Acad Sci USA 105(Suppl 1):11536–11542. conspicuously dimorphic brittle star Ophiodaphne formata (Ophiuroidea). Biol Bull 5. Alroy J (2010) The shifting balance of diversity among major marine animal groups. 206(1):25–34. Science 329(5996):1191–1194. 42. Picken GB, Allan D (1983) Unique spawning behaviour by the Antarctic limpet Nacella 6. Foote M (2010) The geological history of biodiversity. Evolution Since Darwin: The (Patinigera) concinna (Strebel, 1908). J Exp Mar Biol Ecol 71(3):283–287. First 150 Years, eds Bell MA, Futuyma DJ, Eanes WF, Levinton JS (Sinauer, Sunderland, 43. Young CM (1994) The biology of external fertilization in deep-sea echinoderms. Mass.), pp 479–510. Reproduction, Larval Biology, and Recruitment of the Deep-Sea Benthos, eds 7. Stanley SM (2007) An analysis of the history of marine animal diversity. Paleobiology Young CM, Eckelbarger KJ (Columbia University Press, New York), pp 179–200. Memoirs 4:1–55. 44. Warén A, Bouchet P (1993) New records, species, genera, and a new family of gas- 8. Benton MJ, Emerson BC (2007) How did life become so diverse? The dynamics of diversi- tropods from the hydrothermal vents and hydrocarbon seeps. Zool Scr 22(1):1–90. fication according to the fossil record and molecular phylogenetics. Palaeontology 50(1): 45. Quinn JF (1991) Systematic position of Basilissopsis and Guttula, and a discussion 23–40. of the phylogeny of the Sequenzioidea (Gastropoda: Prosobranchia). Bull Mar Sci 9. Bush AM, Bambach RK (2015) Sustained Mesozoic–Cenozoic diversification of marine 49(1-2):575–598. Metazoa: A consistent signal from the fossil record. Geology 43(11):979–982. 46. Golikov AN, Kussakin OG (1972) Sur la biologie de la reproduction des patelles de la 10. Bambach RK (1993) Seafood through time: Changes in biomass, energetics, and famille Tecturidae (Gastropoda: Docoglossa) et sur la position systématique des ses productivity in the marine ecosystem. Paleobiology 19(3):372–397. subdivisions. Malacologia 11:287–294. 11. Bambach RK (1999) Energetics in the global marine fauna: A connection between 47. Healy JM, Rowe FWE, Anderson DT (1988) Spermatozoa and spermiogenesis in Xy- terrestrial diversification and change in the marine biosphere. Geobios 32(2):131–144. loplax (Class Concentricycloidea): A new type of spermatozoon in the Echinodermata. 12. Allmon WD, Martin RE (2014) Seafood through time revisited: The Phanerozoic in- Zool Scr 17(3):297–310. crease in marine trophic resources and its macroevolutionary consequences. 48. Sepkoski JJ, Jr (2002) A compendium of fossil marine animal genera. Bull Am Paleontol 363: Paleobiology 40(2):256–287. 1–560. 13. Hunt G, Hopkins MJ, Lidgard S (2015) Simple versus complex models of trait evolution 49. Miller AI, Sepkoski JJ, Jr (1988) Modeling bivalve diversification: The effect of in- and stasis as a response to environmental change. Proc Natl Acad Sci USA 112(16): teraction on a macroevolutionary system. Paleobiology 14(4):364–369. 4885–4890. 50. Krug AZ, Jablonski D, Valentine JW (2009) Signature of the end-Cretaceous mass 14. Knoll AH (1986) Patterns of change in plant communities through geologic time. extinction in the modern biota. Science 323(5915):767–771. Community Ecology, eds Case TJ, Diamond J (Harper and Row, New York), pp 51. Beesley PL, Ross GJB, Wells A, eds (1998) : The Southern Synthesis, Part B 126–141. (CSIRO Publishing, Melbourne), Vol 5. 15. Coyne JA, Orr HA (2004) Speciation (Sinauer, Sunderland, Mass.), p 545. 52. Mah CL, Blake DB (2012) Global diversity and phylogeny of the Asteroidea (Echino- 16. Grant V (1949) Pollination systems as isolating mechanisms in angiosperms. Evolution dermata). PLoS One 7(4):e35644. 3(1):82–97. 53. Sepkoski JJ, Jr, Bambach RK, Raup DM, Valentine JW (1981) Phanerozoic marine di- 17. van der Niet T, Johnson SD (2012) Phylogenetic evidence for pollinator-driven di- versity: A strong signal from the fossil record. Nature 293:435–437. versification of angiosperms. Trends Ecol Evol 27(6):353–361. 54. Benton MJ (1995) Diversification and extinction in the history of life. Science 268(5207): 18. Regal PJ (1977) Ecology and evolution of flowering plant dominance. Science 52–58. 196(4290):622–629. 55. Sepkoski JJ, Jr (1984) A kinetic model of Phanerozoic taxonomic diversity. III. Post- 19. Kay KM, Sargent RD (2009) The role of animal pollination in plant speciation: In- Paleozoic families and mass extinctions. Paleobiology 10(2):246–267. tegrating ecology, geography, and genetics. Annu Rev Ecol Evol Syst 40:637–656. 56. Vermeij GJ (1977) The Mesozoic marine revolution: Evidence from snails, predators, 20. Armbruster WS, Muchhala N (2009) Associations between floral specialization and and grazers. Paleobiology 3(3):245–258. species diversity: Cause, effect, or correlation? Evol Ecol 23:159–179. 57. Huntley JW, Kowalewski M (2007) Strong coupling of predation intensity and di- 21. Xu S, et al. (2011) Floral isolation is the main reproductive barrier among closely re- versity in the Phanerozoic fossil record. Proc Natl Acad Sci USA 104(38):15006–15010. lated sexually deceptive orchids. Evolution 65(9):2606–2620. 58. Bambach RK, Knoll AH, Sepkoski JJ, Jr (2002) Anatomical and ecological constraints 22. Vermeij GJ, Grosberg RK (2010) The great divergence: When did diversity on land on Phanerozoic animal diversity in the marine realm. Proc Natl Acad Sci USA 99(10): exceed that in the sea? Integr Comp Biol 50(4):675–682. 6854–6859. 23. Grosberg RK, Vermeij GJ, Wainwright PC (2012) Biodiversity in water and on land. 59. Bush AM, Bambach RK (2011) Paleoecologic megatrends in marine metazoa. Annu Curr Biol 22(21):R900–R903. Rev Earth Planet Sci 39:241–269. 24. Strathmann RR (1990) Why life histories evolve differently in the sea. Am Zool 30: 60. Vermeij GJ (1987) Evolution and Escalation: An Ecological History of Life (Princeton 197–207. University Press, Princeton, NJ). 25. Pennington JT (1985) The ecology of fertilization of echinoid eggs: The consequences 61. Martin R, Quigg A (2012) Evolving phytoplankton stoichiometry fueled diversification of sperm dilution, adult aggregation, and synchronous spawning. Biol Bull 169(2): of the marine biosphere. Geosciences (Basel) 2(2):130–146. 417–430. 62. Finnegan S, McClain CR, Kosnik MA, Payne JL (2011) Escargots through time: An 26. Babcock R, Keesing J (1999) Fertilization biology of the abalone Haliotis laevigata: energetic comparison of marine gastropod assemblages before and after the Meso- Laboratory and field studies. Can J Fish Aquat Sci 56(9):1668–1678. zoic marine revolution. Paleobiology 37(2):252–269. 27. Palumbi SR (1994) Genetic divergence, reproductive isolation, and marine speciation. 63. Heim NA, Knope ML, Schaal EK, Wang SC, Payne JL (2015) Animal evolution. Cope’s Annu Rev Ecol Syst 25:547–572. rule in the evolution of marine animals. Science 347(6224):867–870. 28. Malay MCD, Paulay G (2010) Peripatric speciation drives diversification and distribu- 64. Falkowski PG, et al. (2004) The evolution of modern eukaryotic phytoplankton. tional pattern of reef hermit crabs (Decapoda: Diogenidae: Calcinus). Evolution 64(3): Science 305(5682):354–360. 634–662. 65. Falkowski P, Knoll AH, eds (2007) Evolution of Primary Producers in the Sea (Academic 29. Solignac M (1981) Isolating mechanisms and modalities of speciation in the Jaera Press, Amsterdam). albifrons species complex (Crustacea, Isopoda). Syst Zool 30(4):387–405. 66. Meyer KM, Ridgwell A, Payne JL (2016) The influence of the biological pump on ocean 30. Cruz R, Carballo M, Conde-Padín P, Rolán-Alvarez E (2004) Testing alternative models chemistry: Implications for long-term trends in marine redox chemistry, the global for sexual isolation in natural populations of Littorina saxatilis: Indirect support for carbon cycle, and marine animal ecosystems. Geobiology 14(3):207–219. by-product ecological speciation? J Evol Biol 17(2):288–293. 67. Thayer CW (1983) Sediment-mediated biological disturbance and the evolution of the 31. Bauer RT, Martin JW, eds (1991) Crustacean Sexual Biology (Columbia University Press, marine benthos. Biotic Interactions in Recent and Fossil Benthic Communities, eds New York). Tevesz MJS, McCall PL (Plenum, New York), pp 479–625. 32. Duffy JE, Thiel M (2007) Evolutionary Ecology of Social and Sexual Systems: Crustaceans As 68. Tomašových A (2008) Substrate exploitation and resistance to biotic disturbance in Model Organisms (Oxford University Press, New York). the brachiopod Terebratalia transversa and the bivalve Pododesmus macrochisma. 33. Boughman JW (2001) Divergent sexual selection enhances reproductive isolation in Mar Ecol Prog Ser 363:157–170. sticklebacks. Nature 411(6840):944–948. 69. Stanley SM (1974) What has happened to the articulate brachiopods? GSA Abstracts 34. Ellis EA, Oakley TH (2016) High rates of species accumulation in animals with bio- with Programs 6:966–967. luminescent courtship displays. Curr Biol 26(14):1916–1921. 70. Liow LH, Reitan T, Harnik PG (2015) Ecological interactions on macroevolutionary 35. West-Eberhard MJ (1983) Sexual selection, social competition, and speciation. Q Rev time scales: Clams and brachiopods are more than ships that pass in the night. Ecol Biol 58(2):155–183. Lett 18(10):1030–1039. 36. Ponder WF, Lindberg DR (1997) Towards a phylogeny of gastropod molluscs: An 71. Alroy J, et al. (2008) Phanerozoic trends in the global diversity of marine inverte- analysis using morphological characters. Zool J Linn Soc 119(2):83–265. brates. Science 321(5885):97–100. 37. Dix TG (1969) Aggregating in the echinoid Evechinus chloroticus. Pac Sci 23(1): 72. Alroy J (2014) Accurate and precise estimates of origination and extinction rates. 123–124. Paleobiology 40(3):374–397.

14078 | www.pnas.org/cgi/doi/10.1073/pnas.1610726113 Bush et al. Downloaded by guest on September 27, 2021