Introduction
Changing physical environments play a critical role in ecology and evolution (Schindler Reference Schindler1990; Wootton et al. Reference Wootton, Parker and Power1996; Jablonski Reference Jablonski2003; Vrba Reference Vrba2005; Jackson and Erwin Reference Jackson and Erwin2006). However, predicting how specific environmental changes will influence biotic systems is difficult, because the changes may influence biological interactions over multiple temporal and spatial scales (Levin Reference Levin1992; Lynch and Lande Reference Lynch and Lande1993; Lavergne et al. Reference Lavergne, Mouquet, Thuiller and Ronce2010; Pereira et al. Reference Pereira, Leadley, Proença, Alkemade, Scharlemann, Fernandez-Manjarrés and Araújo2010; Dawson et al. Reference Dawson, Jackson, House, Prentice and Mace2011). Fortunately, the fossil record provides an unique opportunity to assess the effect of changing environmental conditions on biotic interactions over extended timescales (Marx and Uhen Reference Marx and Uhen2010) and to disentangle complicated biotic responses to environmental change (Terry et al. Reference Terry, Cheng and Hadly2011; Blois et al. Reference Blois, Zarnetske, Fitzpatrick and Finnegan2013, Reference Blois, Gotelli, Behrensmeyer, Faith, Lyons, Williams and Amatangelo2014).
Predation is a biotic interaction that plays an especially important role in shaping ecosystems through cascading effects (Harriston et al. Reference Harriston, Smith and Slobodkin1960; Paine Reference Paine1966; Pace et al. Reference Pace, Cole, Carpenter and Kitchell1999; Roemer et al. Reference Roemer, Gompper and Van Valkenburgh2009; Estes et al. Reference Estes, Terborgh, Brashares, Power, Berger, Bond and Carpenter2011). A number of variables influence predator–prey interactions, including productivity (Leibold Reference Leibold1989; Holt et al. Reference Holt, Grove and Tilman1994; Bohannan and Lenski Reference Bohannan and Lenski2000), temperature (Elliott and Leggett Reference Elliott and Leggett1996), environmental disturbance (Bertness Reference Bertness1981; Menge and Sutherland Reference Menge and Sutherland1987), and habitat complexity (Almany Reference Almany2004). The fossil record provides the opportunity to directly examine how environmental perturbations influence predation over long timescales, because certain predators leave distinctive traces of their predation on the hard parts of their prey (Kitchell et al. Reference Kitchell, Boggs, Kitchell and Rice1981; Kowalewski Reference Kowalewski2002).
The fossil record of drilling gastropod predators represents one of the few direct records of predation (Kowalewski Reference Kowalewski2002). Although several families of gastropods produce drill holes, most reported in the fossil record, including this study, closely resemble those made by muricid and naticid gastropods (Kelley and Hansen Reference Kelley and Hansen2003). Both naticids and muricids create bore holes using a combination of mechanical and chemical processes in which the proboscis/radula and the accessory boring organ are used alternately to rasp and soften the shell chemically (Carriker Reference Carriker1981). Naticids are infaunal (live within the sediment) and usually prefer infaunal prey, whereas muricids are primarily epifaunal (live on the surface of the seafloor) and therefore search for and drill primarily epifaunal prey (Kelley and Hansen Reference Kelley and Hansen2003). Naticid and muricid gastropods both produce regular holes that can be distinguished from the more irregular drill holes produced by other predators (principally octopods and worms). Naticid drill holes are normally beveled, while muricids more commonly produce smooth-sided, cylindrical drill holes, but shell structure and thickness of prey may influence the shape of the drill hole (Kowalewski Reference Kowalewski1993; Kelley and Hansen Reference Kelley and Hansen2003).
Here, we use the fossil record of gastropod predation on bivalves to test the effect of changes in the physical environment on trophic interactions in the Caribbean during the Neogene. The final closure of the Central America Seaway (CAS) by the emergence of the Isthmus of Panama approximately 3.5 Ma (Coates et al. Reference Coates, Jackson, Collins, Cronin, Dowsett, Bybell, Jung and Obando1992, Reference Coates, Collins, Aubry and Berggren2004; Coates and Stallard Reference Coates and Stallard2013; Jackson and O’Dea Reference Jackson and O’Dea2013) caused a decrease in upwelling and planktonic productivity that provides a natural experiment for testing how changes in the physical environment may have influenced the structure and function of Caribbean nearshore marine communities (Woodring Reference Woodring1966; Vermeij and Petuch Reference Vermeij and Petuch1986; Johnson et al. Reference Johnson, Budd and Stemann1995, Reference Johnson, Todd and Jackson2007, Reference Johnson, Jackson and Budd2008; Jackson et al. Reference Jackson, Todd, Fortunato and Jung1999; Todd et al. Reference Todd, Jackson, Johnson, Fortunato, Heitz, Alvarez and Jung2002; O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007; O’Dea and Jackson Reference O’Dea and Jackson2009; Smith and Jackson Reference Smith and Jackson2009; Leonard-Pingel et al. Reference Leonard-Pingel, Jackson and O’Dea2012). The fossil record is rich and well dated, and changes in physical environments are well documented based on independent proxy data (O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007). We use this excellent framework to quantify changes in drilling predation through time and assess potential causes of change in predation, specifically addressing two alternative hypotheses:
H1: Decline in regional planktonic productivity directly resulted in decreased intensity of predation as expected from ecological theory and from laboratory experiments. Theoretically, as a prey population grows, predators respond behaviorally (functional response) by increasing their rate of prey consumption (Holling Reference Holling1959). This has been observed in both field and laboratory experiments (Ricker Reference Ricker1941; Holling Reference Holling1959; Kauzinger and Morin Reference Kaunzinger and Morin1998) and is well established in theoretical ecology (e.g., Case Reference Case2000). Conversely, then, if the prey population decreases in size due a decrease in productivity, one might expect to see a decrease in the intensity of predation.
H2: Changes in predation intensity were the result of changes in habitats and the species that inhabited them rather than a direct effect of changes in productivity on predation intensity.
Materials and Methods
Sample Collection and Predation Measures
We collected 189 fossil samples from 28 faunules across Panama and Costa Rica that span 11 Myr (Table 1, Fig. 1). Here, we define a “faunule” as a pooled collection of bulk samples from a locality that we believe reasonably represents a fossil community (e.g., bulk samples that are pooled into a faunule are from the same environment, time, and geographic location) (see also Jackson et al. Reference Jackson, Todd, Fortunato and Jung1999; Johnson et al. Reference Johnson, Todd and Jackson2007; O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007; Smith and Jackson Reference Smith and Jackson2009; Leonard-Pingel et al. Reference Leonard-Pingel, Jackson and O’Dea2012). We collected between 1 and 21 approximately 10 kg bulk sediment samples (collection bag of loosely consolidated sediment, rock, and fossils) from each locality, carefully controlling for stratigraphy and geographic extent (e.g., closely spaced sampling from the same geologic horizon and paleoenvironment). The use of several bulk samples for each locality allowed us to attain an adequate sample size. In addition, individual bulk samples are combined into faunules to avoid pseudoreplication—each individual bulk sample is not an independent replicate (Hurlburt Reference Hurlburt1984); however, each faunule does represent an independent sample. Faunules were assigned to one of four habitat types: reef, seagrass, mixed reef and seagrass, or soft sediment based on lithologic descriptions, total faunal assemblages, and the percent mud and carbonate in sediments (Coates Reference Coates1999; Jackson et al. Reference Jackson, Todd, Fortunato and Jung1999; O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007; see Table 2). Although identified by a suite of descriptive characters, reef habitats were identified either by the presence of large–reef building corals, which remained uncollected but were noted in stratigraphic descriptions, or by a high proportion of coral debris found in the fossil assemblages. Seagrass environments are also identified primarily by faunal composition, either by bivalves commonly associated with seagrass beds, or the presence of small, solitary corals typical of seagrass beds (see Leonard-Pingel et al. Reference Leonard-Pingel, Jackson and O’Dea2012). Soft-sediment habitats were recognized by faunal assemblages typical of sandy or muddy bottoms, particularly a very low proportion of coral debris in the fossil assemblages (mean=1.26% by weight, SD=1.71).
Figure 1 Map of Panama and eastern Costa Rica, with insets showing the four basins from which collections were taken; Limon Basin, Costa Rica; Bocas del Toro Basin, Panama; Panama Canal Zone, Panama; and Darien Basin, Panama. Numbers correspond to faunules listed in Table 1.
Table 1 Age, geologic formation, and locality information for each faunule. Numbers correspond to locations mapped in Figure 1.
Table 2 Age, habitat designation, environmental data, and abundance data for each faunule (O'Dea et al. 2007; Leonard-Pingel et al. Reference Leonard-Pingel, Jackson and O’Dea2012). Mixed habitat refers to a combination of seagrass and coral dominated environments.
We washed bulk samples on a 2 mm sieve and sorted them to gross taxonomic groups. We then sorted bivalves to genus and gastropods to family following the nomenclature of Todd (Reference Todd2001). Identification of bivalves to the generic level is based on dentition (number and arrangement of teeth on the hinge plate), shell shape, and ornament; all characters that are readily preserved in the fossil record. Morphological characters that distinguish gastropod families are also well preserved. Individuals that we could not identify to at least family with a high degree of certainty, particularly due to poor preservation, represent <1% of the total number of bivalve valves (~109,000) identified and counted. Unidentified individuals were not included in this analysis. For bivalves, numbers of individuals were estimated from the number of valves with a hinge and umbo and divided in half to account for disarticulation; gastropods with an apex were counted as individuals (Gilinsky and Bennington Reference Gilinsky and Bennington1994). We examined bivalves for the presence of distinctive traces left by naticids or muricids (Kitchell et al. Reference Kitchell, Boggs, Kitchell and Rice1981; Vermeij Reference Vermeij1987; Kelley et al. Reference Kelley, Hansen, Graham and Huntoon2001; Leighton Reference Leighton2002; Walker Reference Walker2007). However, we did not distinguish between muricid and naticid drill traces in the final analyses because of ambiguity introduced by differential shell preservation, variability in drill shape among individuals (Kowalewski Reference Kowalewski2004), and variability in drill shape among prey with different microstructures (Hoffman et al. Reference Hoffman, Pisera and Ryszkiewicz1974). We then tallied the number of valves displaying at least one successful drilling trace (Kowalewski Reference Kowalewski2002). We did not observe unambiguous edge drilling of bivalves (Dietl and Herbert Reference Dietl and Herbert2005; Chattopadhyay et al. Reference Chattopadhyay, Zuschin and Tomašových2014). We calculated the drilling frequency (number of bivalve valves drilled/half of the total number of bivalve valves in the sample) at the assemblage level for each faunule as a proxy for predation intensity sensu Kowalewski (Reference Kowalewski2002) (Table 2). We also calculated the drilling frequency for each bivalve genus with at least 25 valves pooled over all samples (see Table 3).
Table 3 Bivalve genera pooled across all samples with drilling frequency and preferred habitat. Preferred habitat is inferred from a reivew of the literature with key references cited.
Statistical Analyses
Bivalves
We performed a Spearman rank correlation on the assemblage-level drilling frequency in relation to faunule age to test for changes in predatory drilling through time. To test for the effect of habitat on drilling we used Wilcoxon rank-sum tests to compare the drilling frequency of faunule assemblages categorized as either soft-sediment or biogenic habitats (reefs, seagrasses, or a combination of both). We also compared drilling frequency for genera characteristic of biogenic versus soft-sediment environments using data pooled across all the faunules. Assignment of genera to different habitat types was based on a literature survey (Table 3). Additionally, we used drilling frequency data on individual genera to determine whether the same genus living in different habitats had similar drilling frequencies. We chose four common genera typical of soft-sediment habitats and four common genera typical of biogenic habitats, calculated the drilling frequency for each genus in each faunule where it occurred in a reasonable abundance (>5 valves), and then compared the drilling frequencies in soft-sediment and biogenic habitats for each genus using Wilcoxon rank-sum tests.
We also looked for significant differences in drilling frequency between functional groups that are strongly related to habitat (i.e., infaunal vs. epifaunal bivalves and chemosymbiotic vs. siphonate bivalves) using G-tests of independence. For comparison of predation frequency upon infaunal versus epifaunal bivalves, we excluded groups whose life habits do not fit well within either of these broad functional types. The most abundant of these are small corbulid bivalves that live byssally attached to sediment grains upon or immediately below the sediment surface in gregarious clusters (Mikkelsen and Bieler Reference Mikkelsen and Bieler2001). We also excluded Pectinidae (scallops) because of their ability to move freely or swim away from predators, a trait that is rare among other epifaunal bivalves (Joll Reference Joll1989). For comparison of drilling on chemosymbiotic and siphonate bivalves, the chemosymbiotic group includes all the lucinid bivalves; the siphonate group includes families of suspension-feeding or deposit-feeding bivalves with siphons long enough to facilitate relatively deep burrowing. The latter include members of the families Semelidae, Solecurtidae, Tellinidae, Thraciidae, and Veneridae.
Gastropods
The abundance of predators (Naticidae and Muricidae gastropods) or changes in their relative abundance (ratio of Muricidae to Natidicae) might also influence drilling frequency. To investigate this, we binned faunules according to age—11–6.35 Ma (before major restriction of shallow-water connections between the Caribbean and eastern Pacific), 4.25–3.05 Ma (transitional period of increasing restriction between the oceans), and 2.6–0.007 Ma (modern Caribbean)—and performed Kruskal-Wallis tests to determine whether there were significant differences in the ratio of predators to prey or shifts in the dominant drilling predators through time. In cases in which Kruskal-Wallis tests showed significant differences, Wilcoxon rank-sum tests were performed a posteriori. We also performed G-tests of independence to test for significant differences in relative abundance of muricid and naticid gastropods in soft-sediment and biogenic habitats
Results
We examined a total of 109,202 bivalve valves from 145 genera and found 11,405 valves with unambiguous drill holes. We also counted 34,931 gastropods and identified 1563 naticid gastropods and 761 muricid gastropods. Analysis of this data shows that the drilling frequency of bivalves significantly increases through time despite high variability among faunules (ρ=0.33, p<0.05) (Fig. 2). However, when faunules are grouped by habitat, the percentage of drilled bivalves is uncorrelated with faunule age (soft-sediment habitats: ρ=0.04, p=0.44; biogenic habitats: ρ=−0.19, p=0.26). Assemblage-level drilling frequencies of biogenic (seagrass meadow, coral reef, or mixed reef–seagrass) habitats are nearly double those of assemblage-level drilling frequencies in soft-sediment faunules (W=159, p<0.01; Fig. 3A). Drilling frequencies for bivalve genera characteristic of biogenic habitats experience significantly higher drilling frequencies than genera characteristic of soft-sediment habitats (W=1299.5, p<0.01; Fig. 3B). Furthermore, bivalve functional groups typically associated with biogenic habitats experience more drilling than bivalve functional groups associated with soft-sediment habitats. The drilling frequency of epifaunal bivalves characteristic of reef-associated habitats and chemosymbiotic bivalves characteristic of seagrasses is 2 to 4 times higher than the drilling frequency of infaunal bivalves characteristic of soft sediments (G=322.89, p<0.0001 and G=680.59, p<0.0001, respectively; Fig. 3C,D). However, drilling frequencies for individual genera that occur in both soft-sediment and biogenic habitats show only one case of a significant difference between the two habitat groups (Fig. 4).
Figure 2 Percentage of bivalves drilled for each faunule. The percentage of drilled bivalves significantly increased through time (ρ=0.33, p<0.05), but there is no correlation between percentage of drilled bivalves and time when faunules are analyzed by habitat (soft-sediment habitats (squares): ρ=0.04, p=0.44; biogenic habitats (circles): ρ=−0.19, p=0.26). This contradicts the hypothesis that predation is positively correlated with primary productivity.
Figure 3 Differences in predation rates between habitats and bivalve functional groups. W-values are for Wilcoxon rank-sum tests; G-values are for G-tests. A, Biogenic faunules (n=14) have significantly higher drilling frequencies than do soft-sediment faunules (n=14). B, Genera characteristic of biogenic habitats (n=36) display higher drilling frequencies than do genera characteristic of soft-sediment habitats (n=54). C, The percentage of epifaunal bivalves (n=13592 valves) drilled is more than twice that of infaunal bivalves (n=44011 valves) in all 28 faunules combined. D, The percentage of chemosymbiotic bivalves (n=4646 valves) drilled is more than triple that of siphonate bivalves (n=31050 valves) in all 28 faunules combined.
Figure 4 Box plots of eight common bivalve genera that occur in both soft-sediment and biogenic faunules. W-values are for Wilcoxon rank-sum tests; n denotes the number of faunules. The percentage of valves drilled (drilling frequency) generally remains constant within a genus, with the single exception of Macrocallista.
The ratio of predatory gastropods to bivalve prey was unchanged over the entire interval (Fig. 5A). However, this apparent stability belies a profound shift in the composition of drilling gastropod assemblages. The ratio of epifaunal muricid gastropods to infaunal naticid gastropods shifts among time bins (χ2=9.73, p<0.01), with significantly lower proportions of naticids in younger time bins (W=11, p<0.01; W=4, p<0.01, respectively; Fig. 5B). Muricids comprise more than half the drilling gastropod fauna in biogenic habitats, whereas naticids are nearly 5 times more abundant than muricids in soft-sediment faunules (G=473.61, p<0.0001, df=1, n=2324; see Fig. 6).
Figure 5 Faunules binned according to age; n is the number of faunules for each bin. A, The relative abundance of gastropod predators and bivalve prey remained constant over the past 11 Myr (χ2=1.97, p=0.37). B, The ratio of muricid gastropods to naticid gastropods increased significantly through time.
Figure 6 Naticid gastropods as a proportion of total drilling gastropods varies significantly among different habitat types (χ2=10.99, p<0.01) and is significantly lower in seagrass and reef habitats than soft-sediment habitats. Faunules binned according to habitat type; n is the number of faunules for each bin.
Discussion
The results strongly support the hypothesis that frequency of predation is determined by habitat (H2) and not by the regional decrease in planktonic productivity (H1). Bivalves living in and characteristic of biogenic habitats are subject to higher predation intensities than bivalves living in soft-sediment habitats (Fig. 3). There is no evidence, however, that drilling frequencies increase uniformly in biogenic habitats. Within a genus, drilling frequency is not significantly higher in biogenic environments in seven out of eight genera tested (Fig. 4). We therefore conclude that the regional change in drilling frequency through time (Fig. 1) is due to the increase in the extent of biogenic habitats and the bivalve genera common to these habitats. Consequently, changes in the frequency of drilling predation can ultimately be linked to the final closure of the CAS ca. 3.5 Ma (Keigwin Reference Keigwin1982; Haug et al. Reference Haug, Tiedemann, Zahn and Ravelo2001; Coates and Stallard Reference Coates and Stallard2013; Jackson and O’Dea Reference Jackson and O’Dea2013) and the restructuring of benthic communities associated with closure of the CAS.
Closure of the CAS and Oceanographic Changes
Closure of the CAS led to a variety of environmental changes in the Caribbean, including increased salinity (Keigwin Reference Keigwin1982; Cronin and Dowsett Reference Cronin and Dowsett1996), a decrease in seasonality (Teranes et al. Reference Teranes, Geary and Bemis1996; O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007), and a decrease in regional planktonic productivity as Caribbean upwelling shut down (Allmon Reference Allmon2001; Kirby and Jackson Reference Kirby and Jackson2004; O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007). Isotopic evidence indicates areas of upwelling in the Caribbean in the Miocene and Pliocene and a decrease in upwelling and productivity from the late Pliocene to Pleistocene (Cronin Reference Cronin1991; Cronin and Dowsett Reference Cronin and Dowsett1993, Reference Cronin and Dowsett1996; Jones and Allmon Reference Jones and Allmon1995; Allmon et al. Reference Allmon, Emslie, Jones and Morgan1996). Further evidence for upwelling comes from large Miocene phosphorite deposits in the southeastern United States, Cuba, and Venezuela (Riggs Reference Riggs1984). Vertebrate and invertebrate fossil assemblages also indicate areas of upwelling and high biological productivity throughout the western Atlantic in the Pliocene (Allmon 1993; Allmon et al. Reference Allmon, Emslie, Jones and Morgan1996). Extinction of organisms that required high planktonic productivity in the late Pliocene provides additional evidence for regional productivity declines (e.g., Kirby and Jackson Reference Kirby and Jackson2004; O’Dea and Jackson Reference O’Dea and Jackson2009). While we have good proxy evidence for a regional decrease in planktonic productivity after closure of the CAS, we lack proxies for productivity among individual faunules. Our analysis of the productivity hypothesis (H1) is therefore restricted to a regional scale.
Decreased Productivity, Habitat Change, and Molluscan Assemblages
Reconstructions of paleoenvironments from the fossil record of the southwestern Caribbean demonstrate that reefs and shallow-water seagrass beds were uncommon in geologic formations older than 3.5 Ma (Jackson et al. Reference Jackson, Todd, Fortunato and Jung1999; Hendy Reference Hendy2013) and that extensive reef development and shallow seagrass beds occurred only after seaway closure and a shift to oligotrophic conditions (Jackson et al. Reference Jackson, Todd, Fortunato and Jung1999; Domning Reference Domning2001; Todd et al. Reference Todd, Jackson, Johnson, Fortunato, Heitz, Alvarez and Jung2002; Johnson et al. Reference Johnson, Todd and Jackson2007, Reference Johnson, Jackson and Budd2008; O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007; Jagadeeshan and O’Dea Reference Jagadeeshan and O’Dea2012). Bivalve assemblages from biogenic habitats are strikingly different from those in soft-sediment environments (Jackson Reference Jackson1972, Reference Jackson1973; Todd et al. Reference Todd, Jackson, Johnson, Fortunato, Heitz, Alvarez and Jung2002; Johnson et al. Reference Johnson, Todd and Jackson2007; Smith and Jackson Reference Smith and Jackson2009; Leonard-Pingel et al. Reference Leonard-Pingel, Jackson and O’Dea2012). In particular, hard substrata associated with reefs are dominated by epifaunal bivalves and their predominantly muricid gastropod predators, whereas seagrasses are commonly dominated by infaunal chemosymbiotic bivalves and their naticid gastropod predators. Previous analysis of the bivalve assemblages used in this study showed a significant increase in the abundance of epifaunal suspension feeders and chemosymbiotic feeders after closure of the Isthmian Seaway—this diversification in diets was attributed to a shift toward a detritus-based trophic ecology after closure of the CAS (Leonard-Pingel et al. Reference Leonard-Pingel, Jackson and O’Dea2012). Our results further demonstrate that the dominant drilling gastropods also shifted through time. Muricid gastropods increase in abundance and are more abundant in both reef and seagrass environments. On the other hand, naticid gastropods dominate soft-sediment communities but are less dominant (although still abundant) in seagrass and reef environments (Figs. 5B and 6). Epifaunal and chemosymbiotic bivalves from the Caribbean experience much higher predation intensities than do many other bivalve guilds (Leonard-Pingel and Jackson Reference Leonard-Pingel and Jackson2013). Independent work in Adriatic ecosystems shows a similar pattern (Sawyer and Zuschin Reference Sawyer and Zuschin2010). Increases in the abundance of guilds and taxa that are more susceptible to predation may impact the drilling frequency of the entire faunule, driving regional trends in drilling frequency.
Productivity and Predation
Our results clearly demonstrate that drilling frequency increased while regional planktonic productivity plummeted, contrary to earlier predictions (Todd et al. Reference Todd, Jackson, Johnson, Fortunato, Heitz, Alvarez and Jung2002; Johnson et al. Reference Johnson, Todd and Jackson2007). However, biogenic ecosystems on the seafloor, including coral reefs, algae, and seagrasses, exhibit very high benthic primary production that may rival or even exceed primary production of phytoplankton (Odum and Odum Reference Odum and Odum1955; Zieman and Wetzel Reference Zieman and Wetzel1980; Hatcher Reference Hatcher1988, Reference Hatcher1990; Gallegos et al. Reference Gallegos, Merino, Marba and Duarte1993). Thus the relationship between the incidence of predation and total community primary production is still unresolved.
Implication for Drilling Frequency in the Fossil Record
The fossil record of gastropod drilling predation on prey with hard parts, primarily bivalves, provides the tantalizing potential to examine changes in predation through time and therefore draw conclusions about escalation and coevolution (Vermeij Reference Vermeij1987). However, recent studies have shown that drilling frequencies can be influenced by many environmental factors, including latitude (Kelley and Hansen Reference Kelley and Hansen2007; Martinelli et al. Reference Martinelli, Gordillo and Archuby2013), substrate type (Sawyer and Zuschin Reference Sawyer and Zuschin2010), and sedimentary regime (Huntley and Scarponi Reference Huntley and Scarponi2015), indicating that paleontologists should use caution when making interpretations about temporal changes in drilling frequencies. Our study provides additional evidence that habitat plays an important role in determining predation pressures and drilling frequency.
Regional studies of predation intensity are undoubtedly influenced by the type and diversity of habitats sampled; any trends in predation identified should, therefore, be treated with skepticism until habitat is accounted for. The influence that habitats might have on global studies of predation intensity (e.g., Huntley and Kowalewski Reference Huntley and Kowalewski2007) is less apparent. Few of these large-scale global studies have tried to account for environmental influence on drilling frequency (see Kowalewski et al. [Reference Kowalewski, Hoffmeister, Baumiller and Bambach2005] for an exception). Larger data sets are more likely to include a diversity of habitats from each time sampled, but efforts should be made to standardize these data sets with respect to the types of habits/environments represented.
Studies of predator–prey dynamics in the fossil record have, since Vermeij’s seminal paper on the Mesozoic marine revolution, become focused on the ecological factors (e.g., coevolution, escalation) driving evolutionary changes (Vermeij Reference Vermeij1977, Reference Vermeij1987; Vermeij et al. Reference Vermeij, Schindel and Zipser1981; Dietl and Alexander Reference Dietl and Alexander2000; Dietl and Kelley Reference Dietl and Kelley2002; Harper Reference Harper2006). In the absence of environmental data or stratigraphic context, the importance of ecological interactions may be overstated. Surely, both ecology and environment are important in shaping evolutionary trends (Jablonski Reference Jablonski2003), and we suggest that future studies on time series of predator–prey dynamics should investigate predation changes in more rigorous stratigraphic and environmental contexts. The court jester of environmental change may merit more attention in studies of predation.
Conclusions
Our results do not negate the importance of the collapse in planktonic productivity for ecosystem structure and function (O’Dea et al. Reference O’Dea, Jackson, Fortunato, Smith, D’Croz, Johnson and Todd2007; Todd and Johnson Reference Todd and Johnson2013), nor do they negate the potential impact productivity may have on trophic structure. However, our results do provide an example of the necessity to distinguish between proximate and ultimate factors to unravel cause and effect (Mayr Reference Mayr1961; Didham et al. Reference Didham, Tylianakis, Hutchison, Ewers and Gemmell2005; Leonard-Pingel et al. Reference Leonard-Pingel, Jackson and O’Dea2012). The decline in planktonic productivity extensively changed coastal habitats throughout the region, and these differences in habitat—rather than the changes in planktonic productivity per se—determined the kinds of bivalves present and their susceptibility to predation on smaller scales. The increasing habitat heterogeneity that occurred in the Caribbean, including habitat-level changes in productivity and types of prey, fundamentally drove larger-scale regional trends in predation.
Acknowledgments
We thank A. O’Dea, F. Rodriguez, B. Degracia, J. Pingel, and M. Ho for their assistance in the field and in the laboratory; S. Huang and J. Todd for useful discussion; and N. Bitler, A. Bush, F. Condamine, S. Edie, J. W. Huntley, D. Jablonski, S. Kidwell, S. McCoy, A. Michelson, R. Norris, T. Price, K. Roy, and an anonymous reviewer for useful feedback on the manuscript.
