Comprehensive analysis and reinterpretation of Cenozoic mesofossils reveals ancient origin of the snapping claw of alpheid shrimps

Alpheid snapping shrimps (Decapoda: Caridea: Alpheidae) constitute one of the model groups for inferences aimed at understanding the evolution of complex structural, behavioural, and ecological traits among benthic marine invertebrates. Despite being a super-diverse taxon with a broad geographical distribution, the alpheid fossil record is still poorly known. However, data presented herein show that the strongly calcified fingertips of alpheid snapping claws are not uncommon in the fossil record and should be considered a novel type of mesofossil. The Cenozoic remains analysed here represent a compelling structural match with extant species of Alpheus. Based on the presence of several distinct snapping claw-fingertip morphotypes, the major radiation of Alpheus lineages is estimated to have occurred as early as 18 mya. In addition, the oldest fossil record of alpheids in general can now be confirmed for the Late Oligocene (27–28 mya), thus providing a novel minimum age for the entire group as well as the first reliable calibration point for deep phylogenetic inferences.

Using a set of invasive and non-invasive techniques, the present study provides the first comprehensive structural examination of these enigmatic remains. We demonstrate here that previously described as well as newly collected fossil samples not only represent remains of alpheids, but can in many cases even be identified as the fingertips of the snapping claws of early representatives of the genus Alpheus. These findings lead to the assignment of the oldest fossil record of alpheid shrimps to the Late Oligocene, almost 30 million years ago.  Virtual sagittal section through the same µCT dataset illustrating differences in X-ray absorption caused by different degrees of cuticular calcification. Scale bars equal 1 mm. cut = cuticle, dac = dactylus, epi = epidermis, fit = fingertip, mus = muscle, plu = plunger, pol = pollex, sca = strong calcification, soc = socket, wca = weak calcification.

Results
Isolated, small (<5 mm) fossils from a broad geographical range, including North America, Europe, Africa, and Asia ( Fig. 2) and from various stratigraphic settings (Table 1)  The following aspects strongly argue for an interpretation of these fossils as remains of alpheid shrimps.
All fossil specimens exhibited a consistent preservation pattern: they appeared to be broken off along a similar line of structural weakness. In decapods, such a line is known to mark differences in the calcification pattern between the distal tip and the remainder of the original structure 21,22 . The particular preservation pattern observed is highly reminiscent of the claw fingertips of alpheids. For example, in all species of Alpheus, the fingertip of the snapping claw is always more strongly calcified than the remainder of the claw (Fig. 4), resulting in an externally identifiable boundary between these two areas (Fig. 5a). This boundary is particularly conspicuous in living individuals, in which the fingertips of the snapping claw are pale pinkish, reddish, or purplish, thereby markedly contrasting with the different colour of the rest of the claw (Fig. 1a). In contrast to Alpheus, all species of Synalpheus as well as some species of other alpheid genera (e.g., Alpheopsis Coutière, 1897 or Nennalpheus Banner & Banner, 1981) possess claw fingertips that are not calcified, but instead are corneous (i.e. proteinaceous), semi-transparent, and amber yellow in colour (Fig. 5b). The claw fingertips of the remaining alpheid taxa are similar to the rest of the claw: here, sometimes only the distal-most portion of the fingertips may be slightly more calcified or corneous than the rest of the finger. Further decapod taxa possessing a functional snapping claw, such as some representatives of the palaemonid shrimps (Decapoda: Caridea: Palaemonidae), have uncalcified fingertips that do not display a clear boundary between the fingertip and the remainder of the claw (Fig. 5c).
All fossil specimens exhibited rows of pores along their crests (Fig. 6). The arrangement of these pores was symmetrical on both sides of the fossil structures (Fig. 6b,c,e,f). Such an arrangement of pores can also be found in claw fingertips of extant representatives of Alpheus (Fig. 6a). Based on their size, location, and pattern of distribution, these structures were identified as setal pores. Another similarity between most of the fossil material and extant specimens of Alpheus was the presence of basal pits on the flattened side of the fossil structure (Fig. 6d,g). In Alpheus, these pits bear stamen-shaped sensillae that serve as sensory structures 33 .
Analyses of the internal structure of the fossil samples showed a distinct layering reminiscent of the cuticle found in decapods (Fig. 7a,b) 21 . SEM micrographs of the external layer revealed the presence of numerous pores (Fig. 7c), whilst thin sections showed elongated structures penetrating the different layers ( Fig. 7d-e). In decapods, these so-called tegumental canals are associated with the transportation and deposition of cuticular material 21,34 .
Further analyses of the internal structure of extant as well as fossil claw fingertips revealed significant differences in X-ray attenuation between the proximal and distal portions of the snapping claw dactylus (Fig. 8). In the distal part, an increase in X-ray attenuation from the inside to the outside was observed that corresponded to differences in cuticular density resulting from differing degrees of calcification (Fig. 4b). The pattern of cuticular layering seen in the extant sample (Fig. 8a) was very similar to that seen in the fossil specimens (Fig. 8b-e).
With regard to their chemical composition, the claw fingertips of extant alpheids and those of fossil specimens were largely identical, with calcite being the principal constituent (Fig. 9a,b). The fossil samples contained a secondary substrate admixture composed of muscovite, chlorite, and quartz deriving from attached sediment particles (Fig. 9b). Results obtained using Raman spectroscopy revealed three characteristic bands of calcite at 1088/1087, 714/713, and 283/282 cm −1 (Fig. 9c-d). Fluorescence levels were relatively high, implying the presence of organic and inorganic impurities. Furthermore, in the proximal portion of the dactylus of the extant specimen (Fig. 9c), a major broadening of the 1088/1087 cm −1 calcite band was observed, which can be explained by the lower degree of calcification of the proximal portion of the alpheid snapping claw.

Discussion
Alpheid shrimps are model organisms for studying morphological variation, one of the principal causes for adaptive radiation. In this sense, the key innovation of alpheids is the snapping claw 3 . This specialised organ is a multifunctional tool used for various inter-and intraspecific behaviours, such as aggression, warning, or defence, as well as for stunning and killing prey 9,14 . The functional morphology of the snapping claw has been studied extensively 3, 10, 35 and several physical phenomena associated with snapping, including water jets, light production, and cavitation bubbles have received considerable attention 11,12,36 . However, reliable fossil material that would permit studying the evolution of the alpheid snapping claw more comprehensively or to calibrate molecular clock estimates for alpheids in general had previously not been identified.
The poor fossil record of Alpheidae is a direct consequence of two principal factors: low fossilisation potential of small-sized decapods 19 and difficulties in attributing fossil remains to alpheids and not to other shrimps. The latter impediment is illustrated by the interpretation of some of the fossil samples studied herein as cutting edge fragments of the claws of swimming crabs 32 or as the rostra and beaks of cephalopods [29][30][31] . However, the chelipeds of derived alpheids differ morphologically from those of all other decapods and exhibit a unique combination of morphological characters not present in any other decapod taxon 3 . Therefore, based on the results presented above, the entire fossil material listed in Table 1 is here identified as the remains of the strongly calcified claw fingertips of alpheids. The material comprises several morphotypes of snapping claw fingertips, including four morphotypes attributable to the tip of the dactylus (Figs 1e and 3a-l) and one morphotype that can be assigned to the tip of the pollex (Figs 1f and 3m-o). Our data show that fingertips of alpheid snapping claws are not uncommon in the fossil record and should therefore be considered a novel type of mesofossil.
In taphonomic experiments using extant decapod material, strongly calcified claw fingertips were always retained, irrespective of the time of deployment 20, 22 . Due to their increased level of calcification, claw fingertips are therefore present in most fossil decapod assemblages 21 . Correspondingly, we observed a fundamental difference in the microstructure of alpheid claw fingertips in comparison to the remainder of the claw (Figs 7 and 8), as well as in the chemical composition of the respective parts (Fig. 9). These structural differences are conducive to the preferential preservation of the distal-most parts of the claw. However, a positive bias towards fossilisation of larger-sized alpheid species that possess distally strongly calcified snapping claws -primarily species of Alpheus -can be expected.
In fact, some of these fossil forms have morphological analogues among extant species of Alpheus. For instance, the large triangular dactylus with a long hook and a convex margin documented from the Pleistocene of Egypt (Fig. 3d) as well as the Miocene of Slovakia (Fig. 3e) and Austria (Fig. 3f) is an almost perfect match for the homologous structure in the extant species Alpheus armatus Rathbun, 1901 (Fig. 5a). In addition, the fossil triangular morphotype with a short hook (Fig. 3a-c) can be matched with the dactylus of extant representatives of the Alpheus brevirostris (Olivier, 1811) species group, such as Alpheus bellulus Miya & Miyake, 1969 36 . The co-occurrence of these forms at some localities suggests that the group was already diversifying in the Middle Miocene. As was previously shown, the co-occurrence of several, more or less closely related species of Alpheus at a single site is very common in modern shallow marine habitats, especially in the tropics 4,37,38 .
However, some of the morphological variation observed may also be attributable to differences in the degree of usage of the claw as well as the individuals' age or sex. In general, sexual dimorphism in alpheids can be assessed by observing the more or less pronounced differences in claw shape and size among sexually mature adults 3,38 . However, a critical evaluation of intraspecific variation in the morphology of snapping claws of extant taxa would have to be carried out first in order to determine whether or not the fossil alpheid material may be attributed to particular lineages within Alpheus. Nevertheless, it can be stated with confidence that already by 30 mya alpheids developed several of the snapping claw morphologies that can be observed today.
The tree topology resulting from a broad phylogenetic analysis of alpheid morphological characters suggested a single origin of the snapping claw 3 . However, the apparent parallel evolution of the snapping claw and orbital hoods -protrusions of the carapace which protect the shrimp's eyes from mechanical damage resulting from snapping -a hypothesis first postulated by Coutière 39 , may have resulted in an interdependence of several, possibly homoplasious characters. In general, many conspicuous features of alpheid claws appear prone to convergent  Note that in (a) the musculature inside the claw was virtually removed to allow for better comparison. Scale bars equal 100 µm. epi = epidermis, cut = cuticle. evolution 3 , and the snapping mechanism may not be an exception. Structural differences between the snapping claws of Alpheus and Synalpheus 3 (Fig. 5a,b) as well as preliminary molecular analyses 16 support multiple origins of this highly specialised appendage, a situation also occurring in palaemonid shrimps 40 .
Using a molecular clock approach, the origin of Alpheidae was previously estimated to around 150 mya 41 . Earlier investigations of a selection of American species of Synalpheus 2 found evidence for a major radiation of this taxon during the Late Miocene/Early Pliocene (5-7 mya), i.e. prior to the final closure of the Isthmus of Panama 42 . In addition, Hurt and colleagues 17 concluded that at least two transisthmian species pairs of Alpheus diverged well before the final closure of the Isthmus of Panama, one of them possibly as early as 13 mya. The split between the most divergent transisthmian pairs of Alpheus was therefore estimated to have occurred during the Early Miocene at about 18 mya 43 , which is corroborated with the present observations of several distinct claw fingertip morphotypes from Middle Miocene deposits (Table 1, Fig. 3a-f). However, as shown here, the emergence of a complex snapping claw must have taken place much earlier, at least prior to the Late Oligocene: the oldest known fossil alpheid samples originate from the Chickasawhay Limestone (Table 1; Fig. 1e,f), a unit dated at 27-28 mya 44,45 . This date is more than ten 27,28 or even more than 25 million years 23-26 older than the previous, uncertain records of alpheid fossil remains. Our data thus provide the first reliable calibration points for future phylogenetic inferences focusing on the evolution of complex behavioural and morphological traits among one of the principal model taxa of benthic marine invertebrates.

Methods
Specimens. Fossil specimens extracted from bulk samples that had been processed wet through a stack of sieves were manually picked from washed residues under a binocular. Detailed information on fossil specimens is provided in Table 1 Micro-computed tomography. The entire snapping claw of one extant specimen (Alpheus bisincisus NHMW-CR-25767) and the claw fingertips of four fossil specimens (NHMW 2016/0154/0009, NHMW 2016/0154/0010, UMJGP 211460, UMJGP 211461) were analysed using a SkyScan 1272 µCT scanner (Bruker microCT, Kontich, Belgium). The dry specimens were placed in conical plastic tubes and scanned in air. Scanning parameters were: 60 kV source voltage, 166 µA source current, 3 µm isotropic voxel resolution, 1,706 ms exposure, 0.5° rotational steps over 180°, 2 averages, 0.25 mm aluminium filter, and 56 min scan time.
Light microscopy. Selected specimens were manually ground to a thin slice. After transfer to a glass slide and fine grinding to the target plane (75 µm), sections were polished and observed under a SteREO Discovery.V20 stereomicroscope using polarising filters.