Several modern and archaeological otolith lots of Cilus gilberti and Sciaena deliciosa were selected for the present study. Only sagittae that are the biggest otoliths in these sciaenids are considered here. Modern material was collected along the Pacific coasts of Peru, between Callao (12°S) and Ilo (18°S), and Chile, at Antofagasta (23°S). The archaeological otoliths come from the archaeological site of ‘Quebrada de los Burros’, located in southern Peru (18°00′S, 70°50′W), close to the Chilean border. They were collected in level 3 of the excavation, which has been dated between 7000 and 8000 BP [28,29].

Sagittae of both species exhibit a moderately convex inner face showing a typical tadpole shaped sciaenid sulcus (Fig. 1A and E) and an outer face exhibiting a prominent, expanded, post central umbo (Fig. 1B and C). Most of the modern otoliths display an opaline colour and show a smooth, botryoidal surface. Nevertheless, in some rare cases (2.5% in Cilus gilberti, 2% in Sciaena deliciosa), one or both otoliths display a layer of translucent, euhedral crystals over their outer face. In that case, these anomalous otoliths have a rough surface. Their umbos volumetrically increase and are decorated by irregular, finger-like overgrowths (Fig. 1B and D). Although much rarer, such anomalous otoliths of C. gilberti and S. deliciosa also exist in archaeological material (Fig. 1F). In the archaeological material from southern Peru, only one anomalous sagitta of each species was discovered among the 86 C. gilberti and 891 S. deliciosa examined otoliths. However, in the field, anomalous otoliths could be overlooked because staining and preservation conditions in the sediment alter the translucent colour of their crystalline outer layer.

Compared to the most common otoliths, the general shape of anomalous sagittae is preserved, but a loss of weight, varying between 2 and 19% (Table 1: Cl-21 and Pe-5077), was observed between right and left otoliths, whereas there is no weight difference among normal specimens of S. deliciosa [26].

A pair of modern otoliths of Cilus gilberti comprising a typical and an anomalous specimen, along with an archaeological anomalous right otolith of Sciaena deliciosa, was selected for X-ray diffraction analysis and scanning electron microscope (MEB) investigation.

For X-ray powder diffraction analysis, a micro-sample (<0.5mm3) of otolith rough surface was crushed in an agate mortar and pestled on an X-ray diffractometer. The data were collected on INEL XRG 3000 diffractometer coupled to CPS 120 multichannel detector with Cu Kα radiation and Ni filter at the ‘Muséum national d'histoire naturelle’ (‘Minéralogie’), Paris, France. The operating conditions were 40-kV voltage and 1800–2400-s counting time.

The JEOL JSM 840 A scanning electron microscope of the UFR ‘Sciences de la Terre’ of the ‘Université Pierre-et-Marie-Curie’ (Paris-6), France, was used in the backscattered-electron (BSE) and secondary-electron (SE) modes. The otoliths (either bulk material or polished transverse section of modern otoliths, crosscutting the nucleus) were pasted on a graphite-coated adhesive and carbon-coated. Cathodoluminescence (CL) spectra were recorded between 200 and 900 nm, with a CL spectrometer coupled to the SEM with an aluminium-plated parabolic mirror collector, a specific cooled stage, a silica lens UV quality, a Jobin–Yvon H10.UV grating spectrometer and a Hamamatsu R636 AsGa photomultiplier [30–32]. Operating conditions were 25-kV voltage, with 1×107-A beam intensity, 1-nm scan step and an integrating time of 0.2 s. Such conditions provide a convenient signal for high-resolution spectral measurements, while minimizing defect production in analysed crystals.

According to X-ray diffraction data, aragonite is the only CaCO₃ polymorph in normal otoliths, in agreement with the only data we know for sciaenids [33]. By contrast, both calcite, aragonite and vaterite, were identified in anomalous otoliths (Table 2). Calcite and vaterite were identified in the translucent crystalline outer layer. Euhedral crystals covering the umbo of both modern and archaeological sagittae are clearly identified as calcite. Vaterite and aragonite were only detected in micro-sampling of inner face of the modern anomalous otolith.

Scanning electron microscope images provide further identification clues. The aragonitic otoliths show a botryoidal surface (Fig. 2B1 and B2), which, at higher magnification, appears to be composed of successive concentric, thin layers displaying an orange skin-like surface (Fig. 2B3 and B4). By contrast, anomalous otoliths have a much more complex morphology. Calcite in the umbo of the modern and archaeological otoliths occurs as hexagonal prisms (500–700 μm) ended by a pyramid, a typical morphology of calcite from the Ge group, Phe+ subgroup [34] (Fig. 2A1, A3, and A4). The inner face of anomalous otoliths differs from that of normal otoliths by two different classes of euhedral crystals (Fig. 2A2): (1) hexagonal calcite crystals similar to those of the outer face, but of much smaller size (100–200 μm) and (2) small (50 μm), euhedral, lamellar prisms of vaterite arranged as an ‘open book’. In polished section, aragonitic normal otholiths show radial epitaxial crystal growth (Fig. 1D), as previously observed in broken sections of otoliths [10]. The cores of the anomalous otoliths preserve a similar structure that characterizes the early growth stages. Only the peripheral layers corresponding to the latest growth stages display translucent calcite (± vaterite) crystals, instead of aragonite (Fig. 1D). These observations are in agreement with those of Tomás and Geffen [13].

Cathodoluminescence (CL) easily discriminates between carbonate polymorphs, especially calcite and aragonite. The spectra obtained on Peruvian sciaenid otoliths display two wide bands (Fig. 3). The first band, observed at 363 nm for calcite and 380 nm for aragonite, is attributed to lattice defects [35–38]. The second band, located at 615 nm for calcite and 540 nm for aragonite, results from Ca substitution by Mn in mineral lattice, and clearly discriminates between both carbonate polymorphs [27,39].

The fact that calcitic overgrowths have been discovered in both modern and archaeological otoliths precludes a diagenetic origin (i.e. post mortem re-crystallization due to a diagenetic process) for calcite; single-phase calcitic otoliths have not been found and all the otholiths studied display an aragonitic core, thus indicating that otoliths invariably start growing by aragonite synthesis. Aragonite is metastable in standard geological conditions (i.e. P=1bar and T=25 °C) [40], but it does not transform into calcite over archaeological or even palaeontological timescale [41]. Thus, crystallization of calcite and vaterite in anomalous otoliths results from peculiar biological processes that occasionally occurred during the life of the fishes affected.

The scarcity of anomalous otoliths indicates that specific conditions must be met for calcite and vaterite crystals to be synthesized after a period of normal growth. Accidental and/or pathological causes are the most likely. A genetic origin can be discarded as generally only one of the two otoliths is affected, rarely both, and as, at least in our material, all otoliths have an aragonitic core. The calcification process is dependent on the endolymph chemistry and on the organic matrix [42], and it is known that alteration of the homeostasis of the endolymph may generate different forms of crystals in a single fish species [43]. This alteration could explain why the anomaly is sometimes bilateral. But the fact that an anomalous otolith can coexist with a normal one in the same fish strongly supports the assumption of an accidental cause, likely due to some dysfunction of the biomineralisation process in the inner ear. In that hypothesis, the occurrence of anomalous otoliths in both modern specimens and in archaeological populations rules out environmental modifications, for instance due to anthropic pollution, as the cause of the anomaly. Although relationships between craniofacial deformations and anomalous otoliths have been observed in reared populations of juvenile herring Clupea harengus [13], the studied fish specimens affected by malformations of the inner ear generally show no external morphological evidence. Note that this hypothetical physiological disorder mainly affects the outer faces of the anomalous otoliths and not the inner faces, which are the most important, because they are in contact with the sensory hair cells of the macula sacculi. Otoliths play an important role in hearing and balance system [44], and density differences between right and left otoliths may affect the sound perception [5]; hence a fish with a completely anomalous otolithic organ would probably not survive for long in the wild. An alternative explanation for the crystallization of calcite and/or vaterite instead of aragonite could lie in some physiological mechanisms of compensation or regeneration, similar to those involved in shell regeneration [45]. A significant weight loss characterizes all the anomalous otoliths (13.2% on average) compared to normal otoliths, whereas there is no statistical difference between right and left normal otoliths among living Sciaena deliciosa [26]. This loss is consistent with the lower density of calcite compared to that of aragonite (2.71 vs. 2.93) [46].

Another interesting observation is that anomalous otoliths have been found mainly in marine fishes; as far as we know, the only freshwater (often reared amphibiotic species) fishes displaying anomalous otoliths are a few salmonid species (Oncorhynchus mykiss, O. tshawytscha and Salvelinus namaycush) [20,47,48]. Bowen et al. [48] suggested that prevalence of anomalous otoliths might reflect stress conditions, especially among reared fishes (see also [13]). Peru is characterized by oceanic–climatic upheavals linked to the El Niño/La Niña phenomenon. Warm episodes strongly affect the Peruvian–Chilean temperate marine fauna [49], and this kind of stress might also have an impact on the otolith crystalline synthesis. It is also important to note that all the species concerned are from temperate waters, e.g., North Atlantic, South Africa, New Zealand, and Peru. In this latter country, even if the fishes mentioned here are from the intertropical zone, Peruvian waters are temperate, due to the influence of the cold Humboldt current and of the Peruvian upwelling. Hence variations in water temperature or salinity may be two controlling factors of otolith crystalline growth, although it still should be explained why the anomaly can be sometimes unilateral and sometimes bilateral. Understanding the mechanism causing this type of anomalies would help with better understanding of the biomineralisation process, and, accordingly, the species presenting these anomalies should constitute good models.

This is the first report of anomalous otoliths in Peruvian–Chilean sciaenids, both in present and archaeological specimens. The use of selected micro-sampling and SEM coupled to CL investigation reveals (1) that anomalous otoliths do preserve aragonitic cores, (2) that delicate calcite euhedral crystals cover the umbo of both modern and archaeological sagittae, (3) that vaterite is present in anomalous otoliths.

Archaeological materials provide argument against diagenetic processes, and rule out any anthropic influence. Accidental and/or pathological causes are most likely, but water temperature (or salinity) may also play a role. These parameters must be well-constrained before to use otoliths for speculating on life environment of fishes, palaeoecological environments that prevail, for example, along the south Peruvian coast 8000 years ago.