The western Atlantic Plio-Pleistocene extinctions

Major environmental change and high faunal turnover have made the Neogene–recent western Atlantic a model system for studying the macroevolutionary dynamics that led to the region’s modern coastal ecosystems [1,2]. This system is characterized not only by high extinction, peaking around 2 Ma, but also with significant origination leading to similar modern levels of diversity relative to the Plio-Pleistocene, albeit with different taxonomic composition and differences in species’ abundance [2–6]. The extinction was regional but severe and basin wide, with around 70% of gastropod species [3,7], and 47–65% of bivalves [3,8] going extinct. This system is well documented, with a rich fossil record, and the extinction occurred in relatively recent geologic history, making it ideal for studying not only how extinction impacts ecologically relevant traits but also how recovery can interact with apparently selective extinction pulses. Most work studying this event has focused on the southwest Caribbean, especially Panama [2,6,9,10], though the event is known to have impacted Florida as well [3,7,11,12].

The turnover event was not simply a transient perturbation but rather represented a permanent shift in ecosystems associated with changing environmental conditions. Nutrients declined in the western Atlantic following the closure of the Central American Seaway. The formation of the Central American Isthmus was completed ~2.8 Ma [13,14], cutting off connections to eastern Pacific waters. Resulting oceanographic changes impacted local upwelling [2,12,15–17]. Subsequent global cooling during the Pleistocene modified the hydrologic cycle and may have impacted delivery of nutrients via runoff [18,19]. Hardground habitats and seagrasses also expanded, altering the habitats occupied by marine mollusks [2,5,6,20]. Some studies indicate that predation intensity may also have declined in the region but evidence for changes to predation intensity is limited [21–23]. A possible increase in parasitism in some taxa [24] may be related to declines in predation intensity [25,26], but it is difficult to determine how much of this signal may be taphonomic without further study. In contrast, Sime & Kelley [27] showed stability in predation intensity across the turnover event and highlighted the potential importance for regional variation in response.

Habitat association [2,6], feeding mode [2,6,14,28], and larval ecology [2,17,29,30] have each been observed to be associated with survivorship in this event and success across the boundary, as has metabolic rate, a trait related to body size [1].

Body-size evolution

The impact of evolutionary processes and mass extinction on macroinvertebrate body size has long been of paleontological interest (e.g., 31–35). From a theoretical perspective, selection pressures can act in the same or opposite directions on species within a clade and the constituent populations of each species [36–41]. This results in complex relationships between body-size variation across species-rich clades and among organisms within individual species. Body-size evolution is also an interesting feature to examine in the context of the Plio-Pleistocene of the western Atlantic (PPWA) because the relationship between available nutrients and the distribution of average or maximum species body sizes in a community may not be tied to the microevolutionary responses observed in the constituent species under high and low nutrient conditions. For example, the distribution of sizes in a population may be centered around a smaller mean under low-nutrient conditions. Under high-nutrient conditions, however, it is possible the community supports more species overall, including both small and large species.

Marine mollusks represent a major component of the Mesozoic to Pleistocene fossil record. Compared with other marine poikilotherms, mollusks exhibit distinctive body-size trends and responses to environmental changes [42] including the effect of body size on species duration and extinction risk (e.g., work by Payne & Heim [34], Monarrez et al. [35], Pietsch et al. [41], Jablonski [43], and Crampton et al. [44]). Some studies have recovered conflicting trends across different clades [44,45] whereas others have revealed no trend [43]. There can even be conflicting trends in body-size responses to different extinction events [46]. This suggests that understanding body-size evolution in different clades of mollusks is a complex problem affected by several factors. In some cases, the most pronounced factors are how body size influences group-level characters like species’ geographic range (see Jablonski & Hunt [47] for discussion of geographic range size as a group-level character), especially via the relationship between geographic range size and extinction risk. In other cases, body size is correlated with other organismal traits such as fecundity or other aspects of an organism’s life history or ecology (see reference 41 for further discussion of traits that have been linked to body size in a variety of taxa), and these other factors may be what is primarily determining extinction risk. Lastly, there are factors related to chance and contingency [48].

Recently, Monarrez et al. [49], in an important paper, demonstrated that geographic range is likely to be a more significant influence on survivorship than body size. However, selectivity for body size was distinguishable from selectivity on geographic range in their analyses. Further, extinction models that included body size were preferred over geographic range-only models for gastropods but not bivalves [49], indicating patterns may vary across mollusks. Smaller-bodied gastropod genera were found to have higher extinction risk during both background and mass extinction events, whereas bivalves exhibited almost no selectivity on body size during mass extinctions [49].

While global-scale patterns of selectivity during mass and background extinctions observed at the generic level may yield consistent long-term patterns, the regional event in the PPWA may not conform to this pattern. Indeed, Monarrez et al. [49] noted that local extinction dynamics may deviate from the broader-scale patterns they observed and additively contribute to extinction dynamics during background intervals. The extinction event in the PPWA has been characterized previously as resulting in a smaller body sized post-extinction fauna [5].

In the PPWA, chionine bivalves (placed in Chione, Chionopsis, and Lirophora) decreased in size in the western Atlantic across the boundary [50]. Corbulid bivalves also exhibited a body-size decline for the PPWA, yet over the same interval in the eastern Pacific body size increased [51,52]. Strombina-group gastropods did not exhibit a significant long-term trend in body-size change in the Neogene western Atlantic, although they did show a significant increase in size in the eastern Pacific [29]. However, it is notable that the three extant species in the western Atlantic are larger compared to Plio-Pleistocene forms [29].

The differences between patterns in the PPWA and other regions and times may be attributable to the very specific environmental changes experienced in the PPWA, such as changes in nutrient availability. The association between the PPWA extinction event and a shift to lower-nutrient conditions could be expected to influence the relationship between body size and extinction. For instance, body size in mollusks is generally correlated with nutrient availability [53], which declined in the post-Pliocene of the western Atlantic. Therefore, all other factors being equal, one may expect the pre-extinction fauna to consist of larger body-sized mollusks than the post-extinction fauna [54–58]. In this study, we tested the hypothesis that modern members of gastropod clades in the region differ in body size from the representatives of those same clades in PPWA. We compared all constituent species belonging to three gastropod clades from before and after the extinction, as well as examined body-size evolution in surviving species lineages across the extinction event.

Taxon selection

Three phylogenetically diverse families spanning the Plio-Pleistocene and recent were chosen for analysis and represent clades with distinct ecologies: Conidae (Caenogastropoda: Neogastropoda), Tegulidae (Vetigastropoda: Trochoidea), and non-Vermicularia Turritellidae (Caenogastropoda: Cerithioidea). These taxa were selected because each clade experienced notable turnover at the extinction, and each has at least one species lineage which survived from the Plio-Pleistocene to the modern (Fig 1; species list in S1 Table).

Conidae

Cone snails are an extremely diverse family of venomous, predatory gastropods with high species diversity in both modern communities and fossil assemblages in Florida [69,74]. Multiple species frequently co-occur, most likely due to high degree of specialization on different prey species [75]. Conidae are also of evolutionary interest because of their extremely rapid rates of speciation [74,76]. Fossil Conidae of the western Atlantic Coastal Plain have also been subject to relatively recent systematic revision [69], which included the collection of measurements from several hundred fossil specimens. Twenty-four species of Conidae (19 of Conus and five of Conasprella) were present in the PPWA prior to the extinction interval. Twenty-four species also live in the region today (17 of Conus and seven of Conasprella), and eight of these species have a fossil record (three of Conasprella and five of Conus). Conus patricius was present in the PPWA, but is now restricted to the eastern Pacific. Conus amphiurgus has been reported from the Bermont Fm. [77], which we consider post-extinction, but we note that this has not been verified based on museum specimens [69].

Tegulidae

Tegulids are grazers on micro- and macroalgae as well as sea grasses [78,79]. Recent species inhabit shallow water habitats with some species living within the splash zone and intertidal waters and others restricted to subtidal zones [80]. Florida fossil Tegulidae have been relatively understudied compared with the other families considered herein. Modern tegulids are represented in the region by two species of the genus Agathistoma, three of Tegula, and Cittarium pica (Linnaeus, 1758). There is some dispute as to whether Cittarium pica, which is significantly larger in body size than the other tegulid genera, is properly nested within Tegulidae or is a close relative [81], but we include it here and perform analyses both including and excluding this species. Cittarium pica appears in the late Pleistocene Miami Limestone (~130 ka based on uranium series dating [82,83]), which is post-extinction. One species of Agathistoma and three of Tegula were present in the PPWA. We treated Tegula (Monodonta) kiawahensis Tuomey and Holmes, 1856 as a likely synonym of Tegula exoleta (Conrad, 1843) following Dall [84] and Campbell [67]. Agathistoma fasciatum (Born, 1778) ranges across the extinction boundary.

Turritellidae

Turritellidae are primarily semi-infaunal suspension feeders, also known to deposit-feed [85–87]. They have a diverse and abundant fossil record [87–89] and were disproportionately impacted by the PPWA extinction [2,3,5,6]. We excluded the turritellid genus Vermicularia, which has a different ecology (generally more reef-associated) and morphology (Vermicularia are uncoiled, making axial length comparisons non-analogous) when compared with other turritelld genera [90]. A recent phylogenetic treatment of Pliocene-to-recent species in Florida and the Atlantic Coastal Plain [66] indicated that at least three evolutionary lineages of non-Vermicularia turritelline gastropods were present, with two of these recognized as the genera Caviturritella and Torcula. Twenty fossil species were present in the Plio-Pleistocene of Florida and the Atlantic Coastal Plain, but only two remain in the region: Torcula exoleta (likely the direct descendant of T. perattenuata [66]) and Turritella (sensu lato) perexilis (= Turritella acropora [66]).

Clade-level analyses

For Conidae, the average species size was 40.9 mm (typical sizes) or 59.0 mm (maximum sizes) in the Plio-Pleistocene and 35.4 mm (typical sizes) or 62.0 mm (maximum sizes) in the modern, and these differences were not statistically significant (t>0.75, p = 0.44; t>0.19, p = 0.84 for typical and maximum sizes, respectively). Extinct species also did not differ significantly in size from surviving species (i.e., without considering species which originated post-extinction) by either metric. The mean of the resampled modern body-size distributions fell within the 50% confidence window of the non-selective turnover scenario, indicating no meaningful change (Fig 2). These results were replicated when maximum sizes were used to represent species’ sizes (S2 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Clade body sizes using log transformed mean adult size to represent each turritellid and tegulid species’ size, and typical size to represent the size of each species belonging to Conidae (see Kohn [74] and supplemental text (S1 Text) for additional information on this methodology).

Data are presented for each family from left to right, as the actual distribution of species body size prior to the extinction, a distribution of means when the data are resampled up to the number of pre-extinction species 1000 times with replacement, a non-selective turnover scenario represented by the distribution of mean sizes when the pre-extinction species’ sizes are resampled 1000 times using the number of extant species, a distribution of means when the modern species’ sizes are resampled 1000 times with replacement using the modern number of species for each sample, and the actual size distribution of the modern taxa.

https://doi.org/10.1371/journal.pone.0313060.g002

S2 Fig Clade body sizes using log transformed maximum recorded size to represent each species’ size. Data are presented for each family, from left to right, as the actual distribution of species body size prior to the extinction, a distribution of means when the data are resampled up to the number of pre-extinction species 1000 times with replacement, a non-selective turnover scenario represented by the distribution of mean sizes when the pre-extinction species’ sizes are resampled 1000 times using the number of extant species, a distribution of means when the modern species’ sizes are resampled 1000 times with replacement using the modern number of species for each sample, and the actual size distribution of the modern taxa.

For Tegulidae, the average species size was 8.6 mm (mean sizes) or 11.5 mm (maximum sizes) in the Plio-Pleistocene and 19.0 mm (mean sizes) or 35.0 mm (maximum sizes) in the modern, but these differences were not statistically significant (for mean sizes, t = 1.1, p = 0.30, unequal variance t = 1.4, p = 0.23; for maximum sizes t = 1.01, p = 0.34, unequal variance t = 1.27, p = 0.26). The mean size of modern tegulids is heavily influenced by Cittarium pica, a very large species that originated in the late Pleistocene and which actually may not properly belong to Tegulidae (though it is closely related if not a member of the family). Excluding Cittarium pica, modern tegulids are 11.7 mm (the mean of each species average size) or 16.6 mm (the mean of each species maximum size), which preserves the direction of the signal (modern species are larger on average than Plio-Pleistocene species), but this difference is not statistically significant even when C. pica is included. Only one species survived the extinction, but it does not differ substantially in size from the species which became extinct (8.6 mm for surviving species versus 8.5 mm average size of extinct species). Resampled post-extinction means are substantially higher than for the non-selective turnover scenario, suggesting the modern tegulid species are larger than expected by chance (Fig 2). This result is replicated when maximum sizes are used to represent species’ body size (S2 Fig).

For Turritellidae, the average species size was 65 mm (mean sizes) or 96 mm (max sizes) for Plio-Pleistocene species and 38 mm (mean sizes) or 56 mm (max sizes) for the modern. These changes were not significant under a permutation t-test (t = 1.72, p = 0.10; t = 1.56, p = 0.14 for mean and maximum sizes, respectively), however only two turritellid species live in the region in the modern. Extinct species did not differ significantly in size from survivors (66 mm vs 62 mm—note that there was evolution, discussed below, in the Torcula perattenuata-exoleta lineage, and this comparison was made using the Plio-Pleistocene body-size distribution for this lineage). The resampled size distribution for the modern turritellids fell below the 50% confidence window of the non-selective turnover scenario, indicating that turritellids are smaller than expected by chance (Fig 2). When maximum sizes are used, this change is much less dramatic (S2 Fig). In this case, the actual mean falls within the central quartiles of the non-selective turnover scenario, although the resampled mean fell outside this window.

Conidae species-lineage analyses

Fossil and modern populations of Conasprella jaspidea (Gmelin, 1791), Conasprella stearnsii (Conrad, 1869), Conus anabathrum Crosse, 1865, Conus daucus Hwass, 1792, and Conus spurius Gmelin, 1791 were evaluated (Fig 3). One Conasprella species showed a statistically significant changes in body size between fossil and extant populations—modern C. stearnsii are smaller than their fossil counterparts (17.9 mm vs. 20.5 mm; t-test p <0.01), although this result was not significant when a Bonferroni [99] corrected α of 0.007 is applied). In contrast, one Conus species also showed a statistically significant difference, but of opposite sign: modern Conus daucus are larger than their fossil population (24.9 mm vs 34.5 mm; t-test p <0.001), a result which remains significant when using a Bonferroni corrected α of 0.007. Conasprella jaspidea did not change in size (20.8 mm for modern examples and 20.8 mm for fossil specimens; t-test p = 0.99). Excluding the measurements taken from figured specimens rather than measurements made directly from collections materials (7 specimens of C. jaspidea as detailed in S3 Table), the modern examples are smaller (18.3 mm) on average, but this difference is not statistically significant. Neither of the remaining Conus species showed statistically significant differences in specimen length in fossil versus modern populations. Extant C. anabathrum averaged 30.6 mm, while fossil C. anabathrum averaged 31.5 mm. Extant C. spurius averaged 42.1 mm in length, while fossil C. spurius averaged 45.4 mm (S3 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Shell-length distribution of species belonging to the family Conidae that survived the extinction event comparing Plio-Pleistocene to modern shells.

Only Conasprella stearnsii and Conus daucus had statistically significant changes in body size.

https://doi.org/10.1371/journal.pone.0313060.g003

Tegulidae species-lineage analysis

The single lineage which survived from the Pliocene, Agathistoma fasciatum, had an average fossil length of 8.6 mm and an average modern length of 9.4 mm, but this difference was not statistically significant (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Shell length distribution for fossil and modern Agathistoma fasciatum comparing Plio-Pleistocene to extant shells.

Modern A. fasciatum are not statistically distinct from fossil forms.

https://doi.org/10.1371/journal.pone.0313060.g004

Turritellidae species-lineage analysis

One turritellid species-lineage, Turritella perexilis (= T. acropora) maintained a highly similar shell length across the turnover event (27.4 mm vs 28.1 mm for fossil and extant, respectively, a difference which was not statistically significant) (Fig 5). Torcula exoleta is, however, much smaller than its likely anagenetic ancestor Torcula perattenuatta [66], and this difference is statistically significant even after applying a Bonferroni correction (mean 48.3 mm vs 97.0 mm; t-test p <0.001; Fig 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Shell length distribution of species belonging to the family Turritellidae which survived the extinction event comparing Plio-Pleistocene to extant shells.

Torcula perattenuata is compared with Torcula exoleta, which is believed to be its anagenetic descendant following the phylogeny of Friend et al. [66] Turritella perexilis is the senior synonym of Turritella acropora, the name typically applied to extant forms [66].

https://doi.org/10.1371/journal.pone.0313060.g005

Body-size evolution and the meaning of trait changes in clades

Changes in body size among the species of molluscs in a region through time could be the result of: 1, natural selection on organismic properties; 2, ecological and environmental changes affecting phenotype (in the case of gastropods, temperature, availability of nutrients, and carbonate saturation may each have ecophenotypic effects); 3, macroevolutionary processes causing differential speciation and extinction rates between species with large body size versus small body size either directly or indirectly (if, for example, body size is linked to a species-level trait like geographic range size; or to another organismic trait such as larval type, that in turn is related to a species-level trait like geographic range [44,100–102], leading to substantial diversity differences [38]; 4, phylogenetic biases, if for some reason species with large-bodied organisms were more likely to give rise to species with large-bodied organisms, whereas species with small-bodied organisms were equally likely to give rise to small or large-bodied organisms [37]; 5, chance; and 6, a combination of some or all of these factors (see also Pietsch et al. [41]). In addition, once extinction eliminates a clade in a region, recovery in that group might be precluded, unless there were opportunities for biogeographic invasion, but these seem to have been limited in this case.

We may draw macroevolutionary rather than macroecological conclusions as the taxa considered herein, as well as the modern molluscan fauna in the region generally, are most likely the result of in situ speciation. The modern faunas of the US Atlantic Coastal Plain and Florida (the Caloosahatchian biogeographic province of Petuch [103]) are most similar to species that inhabited the region in the Pliocene (as the modern Caribbean molluscan fauna is most similar to the Pliocene Gatunian biogeographic province of Petuch [103]) indicating that the shift in taxonomic composition is most likely due to differences in in situ speciation rather than long-range dispersal [104,105]. Additionally, the life history traits of the taxa under study make it most likely that any Pleistocene-recent originations occurred locally. Extant tegulids have non-planktotrophic larvae [106–109] and most Conidae [74] and Turritellidae [17] in the modern western Atlantic have larvae expected to require little to no planktotrophy to metamorphose [74,110–112].

How each clade behaved across the turnover event

The return to past levels of diversity of Conidae in Florida and the US Atlantic Coastal Plain included a return to a similar size distribution relative to the pre-extinction community; the very largest species were lost and new species have not yet achieved these large sizes. The impression that taxa are smaller may be the result of the “no-Gullivers” phenomenon [113], rather than a decline in species average body size following the extinction (either within lineages, the lilliput effect sensu Urbanek [114], or within a clade; see Abbott et al. [115] for further discussion of various uses of the term “lilliput effect” in the literature). In this case, we—as paleontological observers—are taking exceptional notice of the absence of one or two very large species, although the clade has a similar size distribution in the modern as it did prior to the extinction (see Harzhauser et al. [116] for a contrasting example where loss of large species was associated with overall decline in clade body size).

The evolution of larger species in Tegulidae may reflect a response to availability of the specific resources used by these species (seagrasses and algae growing on hard substrates [2,6,117]), despite overall decline in nutrient levels. Low connectivity [107] related to short duration in the plankton and high microhabitat partitioning [118] may both have contributed to high speciation and extinction (volatility) in this clade (although the relationship between larval mode and speciation rate in gastropods is not straightforward [102]).

Florida Tegulidae and Conidae originations appear to primarily occur post-extinction, with many recent species having no fossil record (Fig 1). Some extant species of Conidae may have Gelasian or Calabrian origins, but the rate of origination does not appear elevated (using the stratigraphic range data of Hendricks [69]). These clades that rediversified appear to have reoccupied much of the ecological space that may have been associated with their prior body-size distribution, especially in Florida where temperatures are similar to pre-extinction conditions. These new taxa may represent diversification within Florida and the US Atlantic Coastal Plain from the surviving species, immigration from neighboring regions, or in-situ radiation(s) following immigration events. Additional molecular genetic and phylogenetic work incorporating both fossil species and biogeographic information will be needed to determine precisely which species are of local origin. This post-extinction diversification appears to have different timing than originations in the southwest Caribbean where increased origination preceded extinctions [6,9].

Non-Vermicularia Turritellidae likely declined in size as a clade because the extinction was not followed by any cladogenesis in the surviving regularly-coiled taxa (Caviturritella, Torcula or Turritella s.l.), and because the genus Caviturritella, which was generally large bodied [119], was extirpated [66]. The extant regularly coiled western Atlantic Turritellidae may therefore be “dead clades walking” [17,120–122]. The decline of filter-feeding taxa associated with soft substrates is a pervasive characteristic of the turnover event [14], noted not only in turritellids but across bivalve taxa with similar ecology, including declines in the abundance of suspension-feeding bivalves compared to other feeding modes [6]. In contrast, cladogenesis continued to occur in Vermicularia, which differ from other turritellids in that many species are adapted to live on hardground environments, including reefs [90].

Gastropod taxa had higher speciation and extinction rates than bivalves across the turnover event [6] and, based on the patterns observed in our data, post-extinction speciation during recovery was an important part of body-size evolution and how this differed among gastropod families. The relatively rapid origination of many taxa which re-occupied a similar suite of body sizes to the extinct taxa also suggests that body size may be a particularly labile trait in gastropods. Bivalves have been documented to have decreased growth rates yet obtained similar sizes under the new environmental conditions by gaining longer lifespans [28,123] (see Palmer et al. [124] for an example from this study system). It is unknown whether gastropods exhibit similar patterns, but there is some evidence to the contrary in Turritellidae, which do not seem to have a higher frequency of long-lifespan species at high latitudes [58].

Comparison to basal metabolic rates in PPWA molluscs

Recently Strotz et al. [1] found that bivalves and gastropod species containing organisms that had higher basal metabolic rates (BMRs)–which are associated with larger body sizes–had higher rates of extinction within the same study region and interval. The pattern was most pronounced in species with narrow geographic distributions and was more significant in bivalve species than gastropods. Because Strotz et al. [1] considered taxa from a broader range of taxonomic groups, used somewhat different taxonomic concepts and body-size data, and did not consider body size directly, but rather BMR, it is not possible to directly compare their results with ours. However, the difference in pattern may be attributable to several factors, including differences in number of species, differences in number of higher taxa, or relative proportions of taxa with narrow versus broad geographic range. Evaluating a broader suite of taxa and including additional data on functional traits, including larval mode, metabolic rate, lifespan, relationship to substrate, as well as information on species’ geographic range may help determine what selectivity occurred during the extinction, and better characterize how post-recovery communities differ from the Pliocene community.