Use of nursery areas by the extinct megatooth shark Otodus megalodon (Chondrichthyes: Lamniformes)

1. Introduction

Otodus megalodon is the largest macropredatory shark to have ever lived, with body length estimates of up to 15–18 m for the largest adult individuals [1–4]. This species inhabited the warm and temperate waters of all major ocean basins, spanning a range of almost 20 Myr, from the early Miocene to the Pliocene [5–7]. Most of the studies assessing the causes of the cosmopolitan distribution, the evolution of gigantism, and the extinction of otodontid sharks have focused on the impact of climatic factors and/or the abundance and migration patterns of their potential prey [1,8–11], competition with other macropredatory taxa [7,12], the availability of suitable habitats [13] and the presence of regional endothermy [6,14–16]. However, much less attention has been paid to important aspects of their reproductive biology [17].

Nursery areas are crucial for several marine groups, as they improve the probability of survival of juveniles and allow a greater persistence and recruitment to the adult populations [18,19]. Shark nurseries are usually located in geographically discrete zones with high primary productivity, being defined by a comparatively high abundance of juveniles that tend to remain or return for extended periods of time [20,21]. The use of these areas has been documented in a number of living species [22], and may vary depending on the habitat (coastal, pelagic or deep-water nurseries) [22], the temporal patterns of occupation (permanent or seasonal) [23], or the number of taxa involved [24]. Some authors have also distinguished between primary nurseries, where gravid females give birth and neonates spend the first stages of their life, and secondary nurseries, with slightly older but still immature individuals [25].

Despite the utility of more recently proposed criteria in defining shark nursery areas, these are usually difficult to apply to extinct species and the recognition of nurseries in the fossil record is therefore challenging (see Heupel et al. [20,22] for a detailed discussion on the evolution of the concept for modern taxa). The existence of palaeonurseries has been recently supported by the presence of either abundant fossilized egg cases [26], biased representations of juveniles [8,27–35], or both [36,37]. Accordingly, the late Miocene Gatún Formation (Panama) has been proposed as a possible nursery area for O. megalodon based on the high proportion of juvenile individuals [27]. A second possible nursery area for O. megalodon could have been present in Borneo Island [38], but this observation, based only on three isolated juvenile teeth, is questionable. Although relevant, these sporadic findings are insufficient to evaluate the prevalence and the impact of such a reproductive strategy in the evolutionary history of O. megalodon and, as consequence, a more comprehensive body of evidence on nursery use patterns is required to reliably assess these questions. This is especially evident when considering the wide temporal and palaeobiogeographic distribution of this species.

In the present study, a new potential nursery of O. megalodon is described from the Langhian (middle Miocene) of Spain. Moreover, we analyse a comprehensive database with records of this species revealing the plausible existence of multiple additional nursery areas, widely distributed in time and space. The broader implications of these findings are then contextualized in the light of some previous hypotheses about the success and later demise of this iconic extinct top predator.

2. Material and methods

The O. megalodon teeth studied first-hand in the present work were collected from Langhian outcrops (late middle Miocene) in the Reverté and Vidal quarries, which are situated in the southwestern part of the Vallès-Penedès Basin, close to the lowest stratigraphic layers of the El Camp de Tarragona Basin (see electronic supplementary material, figure S1a and supplementary methods for a detailed description of the geological context of the area). These quarries are located in the municipality of Castellet i La Gornal, in the province of Tarragona (Northeastern Spain) (electronic supplementary material, figure S1b). A total of 32 teeth, deposited in the Museu del Cau del Tauró de l'Arboç (Tarragona, Spain), have been collected over the past 20 years. Of these, 25 were considered in the present study owing to their excellent preservation (figure 1; electronic supplementary material, table S1). Their assignment to O. megalodon is well supported based on several morphological characteristics [27,39] (see electronic supplementary material, supplementary methods).

• Download figure
• Open in new tab
• Download PowerPoint

Teeth were assigned into jaw positions based on comparison with images in Pimiento & Balk [2], as well as an artificial tooth set figured in Applegate & Espinosa-Arrubarrena [40]. The population size-class structure of O. megalodon from the Reverté and Vidal quarries was inferred from dental parameters. According to the methodology described in Pimiento et al. [27], crown height was used to calculate total body lengths from the position-specific regressions established in the great white shark (Carcharodon carcharias) by Shimada [41] (electronic supplementary material, table S2 and supplementary methods).

The population size-class structure of O. megalodon was analysed by estimating kernel probability densities of the inferred body lengths, with violin plots generated using the package ‘ggplot2' [42] implemented in R [43]. Unimodality was checked by computing Hartigans' dip statistic and its statistical significance in the R package ‘diptest' [44]. Additionally, different modal parameters (mode(s), bimodality ratio and bimodality coefficient) were calculated using the R package ‘modes' [45]. The bimodality ratio captures the difference in magnitude between the two identified peaks as the ratio between the amplitudes of the right and left peaks, respectively. The bimodality coefficient can vary between 0 and 1, where a value greater than 0.56 suggests the existence of bimodality and the maximum value of 1 indicates a fully bimodal distribution. The population size-class structure of O. megalodon was also evaluated with Gaussian mixture models using the R package ‘Mclust' [46]. The optimal model and number of clusters were determined by the maximum Bayesian information criteria (BIC).

The population size-class structure of O. megalodon was assessed in eight additional formations covering a wide geographical area, including sites from the Atlantic, Caribbean and Pacific basins, which range from the middle Miocene to the Pliocene (i.e. the Temblor, Calvert, Yorktown and Bone Valley formations of the USA, the Pisco Formation of Peru, the Gatún and Chucunaque formations of Panama, and the Bahía Inglesa Formation of Chile; figure 2). We analysed the dataset compiled in Pimiento & Balk [2], selecting only the formations therein containing more than 20 tooth records (see electronic supplementary material). Analyses were first performed at formation level because this provided higher sample sizes and more statistically robust results. To obtain higher temporal and spatial resolutions, all the analyses were then repeated at locality level when more than 15 teeth were available (i.e. Shark Tooth Hill, Parkers Creek, Las Lomas, Lago Bayano, Mina Fosforita and Lee Creek Mine from Temblor, Calvert, Pisco, Chucunaque, Bahía Inglesa and Yorktown formations, respectively). All the R scripts are available in the electronic supplementary material.

• Download figure
• Open in new tab
• Download PowerPoint

3. Results

Body length estimates derived from our teeth sample ranged from 2.61 m (MCTA04205, upper anterior tooth) to 13.40 m (MCTA04212, lower lateral tooth) (electronic supplementary material, table S1). Kernel probability density analysis and Hartigans' dip test (HDT) indicate that the population size-class structure of O. megalodon from Reverté and Vidal quarries shows a bimodal distribution, as suggested by estimated density graphs (figure 2), the high bimodality coefficient (0.73) and the comparatively low unimodality test p-value (0.22; although no statistical significance at 0.05 level is found). Higher densities of individuals are estimated within small and large body sizes (modes equal to 5.45 and 10.52 m, respectively) (table 1). Similarly, the Gaussian mixture model analysis (GMM) supports bimodality within the optimal model, with modes equal to 5.28 and 11.10 m, respectively (table 1).

The analysis of additional formations reveals a variety of size-class structures that can be grouped into two major types (figure 2 and table 1). The Temblor, Pisco, Bahía Inglesa and Yorktown formations show maximum density peaks in occurrences of large body sizes (modes from the HDT equal to 11.07, 11.05, 11.95 and 9.82 m, respectively; modes from the GMM equal to 11.68, 11.40, 12.47 and 11.04 m, respectively). Unimodal distributions are well supported by comparatively low bimodality coefficients (below 0.45), high unimodality test p-values (above 0.87) and BIC model selection. By contrast, the Calvert, Gatún, Chucunaque and Bone Valley formations show maximum density peaks within smaller body sizes (modes from the HDT equal to 4.62, 6.56, 6.25, 5.62 m, respectively; modes from the GMM equal to 4.29, 6.66, 7.69 and 6.71 m, respectively). Bimodality coefficients are comparatively higher (above 0.53) and BIC values provide similar support for both unimodal and bimodal models in some instances. A second smaller peak can be recognized in the Gatún Formation (in both the HDT and the GMM), corresponding to large body sizes (modes equal to 17.15 and 14.79 m, respectively), but is negligible when compared to the main left peak (bimodality ratio equals to 0.16). A second smaller peak is also detected in the Calvert Formation (only in the GMM), corresponding to comparatively small body sizes (mode equal to 7.54 m). Despite this, unimodality is well supported in all these formations based on the unimodality test p-values (above 0.6). Results from the analyses performed at locality level are largely equivalent to those of the formations to which they belong (electronic supplementary material, figure S2 and table S3).

4. Discussion

Our results reveal, for the first time, that nursery areas were commonly used by O. megalodon over large temporal and spatial scales. Five of the nine formations examined in this study (i.e. Reverté and Vidal quarries, Calvert, Gatún, Chucunaque and Bone Valley formations) show size class structures that are consistent with the presence of nurseries, with the main size modes placed among body lengths typical of neonates and young juveniles (neonates approx. 4 m and juveniles less than 10.5 m [1]; figure 2 and table 1). Among these, the Calvert, Gatún, Chucunaque and Bone Valley formations show unimodal distributions that are comparable to the age distributions present in most modern shark nursery areas ([48–55] among others). The population size-class structures estimated from the Reverté and Vidal quarries have bimodal distributions, with a second-density peak among typical adult body sizes (greater than 10.5 m [1]) that might represent gravid females. This distributional pattern is present in some living shark populations occupying areas that serve as primary nurseries [56–58]; however, a change in the use of this locality over time could also explain this size distribution, with juvenile and adult populations occupying this region at different times. In any case, the high proportion of neonates and juveniles supports the hypothesis that this area acted as a nursery ground for O. megalodon at some point during the Langhian. This interpretation is further supported by the fact that, during this interval, the Vallès-Penedès Basin was part of a shallow protected bay [59,60] with several taxa that could act as potential prey (i.e. marine mammals and a number fish species) and high primary productivity (as highlighted by the numerous fossil remains found in association with the teeth; electronic supplementary material, table S4). The remaining four formations (i.e. Temblor, Pisco, Bahía Inglesa and Yorktown) demonstrate size-class structures typical of populations dominated by adults (greater than 10.5 m [1]; figure 2 and table 1), suggesting that these regions might correspond to feeding or mating areas [23].

Overall, our analysis based on kernel probability density and Gaussian mixture model also supports the interpretation of the Gatún Formation as a nursery area [27] and reveals the presence of at least four additional nurseries for O. megalodon (i.e. Reverté and Vidal quarries (considered together as a single area), Calvert, Chucunaque and Bone Valley formations). We further validate the results from the size-class structure analysis with a more integrative approach including stratigraphic, sedimentological and palaeontological evidence (electronic supplementary material, supplementary information), which is a fundamental procedure for the reliable recognition of new nursery areas [34]. Moreover, these findings are reinforced by analyses performed at locality level, with higher temporal and spatial resolutions (electronic supplementary material, figure S2 and table S3). This high proportion of potential nursery areas among the analysed localities may not be surprising as nurseries are zones with a high abundance of sharks and, as consequence, huge numbers of teeth can be shed, increasing the chances of subsequent fossil discoveries.

The use of nursery areas is likely to play a key role in the evolutionary history of some shark species [20,61]. This seems to be particularly true for large, slow-growing taxa with low fecundity and high age of maturity, which can benefit from this strategy of maximizing juvenile survival [22]. Evidence from analysis of vertebral growth rings suggests that the slowing or cessation of somatic growth in O. megalodon occurred around 25 years of age, denoting an extremely delayed sexual maturity [14]. It therefore seems plausible that the use of nursery areas could have been essential for O. megalodon, in order to reduce neonate and juvenile mortality and to provide maximum recruitment, thus maintaining viable populations on a long-term temporal scale. The widespread presence of O. megalodon nurseries supports this idea, demonstrating that this species benefited from this strategy at least from the Langhian (middle Miocene) to the Zanclean (Pliocene) in geographically separate regions (figure 2; electronic supplementary material, figure S2). The presumed reliance of O. megalodon on the presence of suitable nursery grounds should be considered in future work as a potential disadvantage in light of the increasing competition of juveniles with the great white shark [6,7] and the drastic reduction in coastal habitats in the Pliocene [13]. More broadly, this work advances our understanding of the impact of reproductive strategies on the evolution of gigantic marine top predators and sheds light on the underlying factors of their evolutionary dynamics.

Ethics

This study does not concern live animals and therefore was not subject to an ethics assessment.

Data accessibility

All raw data and code are included as electronic supplementary material.

Authors' contributions

H.G.F. and C.M.-P. conceived and coordinated the study. C.M.-P., H.G.F., H.B. and J.L.H. participated in the design of the study. J.R. collected and provided the specimens. J.L.H. collected the data. H.G.F. performed statistical analyses. J.L.H. and H.G.F. wrote the paper with input from all co-authors. All authors approved the final version of the manuscript and agree to be held accountable for the content therein.

Competing interests

We declare we have no competing interests.

Funding

H.G.F. is a recipient of a Marie Skłodowska-Curie Individual Fellowship (H2020-MSCA-IF-2018-839636).