Looking for the key to preservation of fossil marine vertebrates in the Pisco Formation of Peru: new insights from a small dolphin skeleton

: e upper Neogene Pisco Formation of Peru is known worldwide as one of the most signiﬁcant Cenozoic marine vertebrate Konservatt-Lagerstätten, even featuring cetacean specimens that retain remains of so tissues or stomach contents. Previous works showed that biomediated precipitation of dolomite


Introduction
Cropping out along the southern coast of Peru from Pisco to Yauca, the Pisco Formation is a late Neogene, shallow-marine sedimentary unit, which is known worldwide for its outstanding paleontological content.In particular, the record of fossil marine vertebrates of the Pisco Formation is exceptional in terms of abundance of specimens, taxonomic diversity, ecomorphological disparity, and quality of preservation.Examples of remarkably preserved fossil remains featured in the Pisco Formation include phosphatized baleen plates and bristles of mysticete whales (Pilleri and Pilleri, 1989;Brand et al., 2004;Esperante et al., 2008;Bisconti, 2012;Gioncada et al., 2016;Marx and Kohno, 2016;Marx et al., 2017), stomach contents and regurgitations of cetaceans and sharks (Collareta et al., 2015(Collareta et al., , 2017;;Lambert et al., 2015), and articulated skeletons of cartilaginous fish (Ehret et al., 2009(Ehret et al., , 2012;;Collareta et al., 2017).Consequently, the Pisco Formation is now recognized as one of the most important Cenozoic marine Fossil-Lagerstätten worldwide (Esperante et al., 2015;Collareta et al., 2015Collareta et al., , 2017;;Gariboldi et al., 2015;Bianucci et al., 2016a, b;Gioncada et al., 2016).Such an outstanding accumulation of fossil vertebrates is currently regarded as the result of a complex interplay of different biotic and abiotic factors and processes, which concurred in creating an environmental setting where post-mortem preservation processes were, locally and at times, greatly facilitated.In particular, the widespread presence of dolomite concretions enclosing many fossil vertebrates of the Pisco Formation has been recently recognized.e formation of mineral concretions during early diagenesis as a consequence of biomediated processes related to organic matter decay has been described elsewhere (McCoy et al., 2015;Yoshida et al., 2015).In the Pisco Formation, early precipitation of dolomite around vertebrate carcasses, although not always present, has been pointed out for its major contribution in allowing the preservation of complete and articulated skeletons, delicate anatomical structures, and even so organic tissues (Gariboldi et al., 2015;Gioncada et al., 2016Gioncada et al., , 2018)).Field evidence and petrographic observations indicate that this precipitation process affected especially large vertebrate carcasses (e.g., large-sized baleen whales), whereas it appears to have been less important (up to almost negligible) in correspondence of smaller carcasses (e.g., small-sized delphinidans and pinnipeds) whose oen well-preserved remains constitute a large part of the fossil assemblages of the Pisco Formation (Gariboldi et al., 2015;Bianucci et al., 2016a, b;Di Celma et al., 2017).Consequently, providing a detailed and integrated characterization of the modes of fossilization of small-sized (i.e., less than 2.5 m in total length) vertebrates preserved in absence of dolomite concretions is pivotal for understanding the genesis of the Pisco Formation Fossil-Lagerstätte.For this purpose, we investigated a well-preserved skeleton of a small pontoporiid dolphin from Pampa Corre Viento, a locality in the Ica Desert where upper Miocene beds of the Pisco Formation are exposed.is specimen is kept in the Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos of Lima, Peru (hereinaer: MUSM) with accession number MUSM 887.Although this specimen is not enclosed within a dolomite concretion, it is surrounded by an evident dark boundary, suggesting that the processes involved in fossilization le their traces in the form of diagenetic minerals around the cetacean skeleton.Examples of similar dark boundaries around remains of larger-sized vertebrates have been observed in the field within the Pisco Formation at the localities of Cerro Los Quesos and Cerro Colorado (Gariboldi et al., 2015).With the aim of shedding light on the fossilization processes that affected well-preserved skeletons of small-sized vertebrates in the Pisco Formation, we undertook a taphonomic, petrographic, and mineralogical study of MUSM 887 and of its "nodule", thus adding a new piece to the comprehension of this renowned Fossil-Lagerstätte.

Geological and stratigraphic background
From a tectonic viewpoint, the Peruvian margin is recognized as a long-lived convergent margin with the oceanic Nazca/Farallon Plate subducting approximately eastward below the western margin of the South America plate, in the Peru-Chile Trench.e subduction of normal oceanic crust and collision of the aseismic Nazca Ridge beneath this part of the forearc have been associated with a wide range of effects on the overriding continent, including long-term subsidence driven by basal tectonic erosion and the development of a composite transform-convergent margin dominated by extensional and strike-slip tectonic activity (Kulm et al., 1982;Dunbar et al., 1990;Cli et al., 2003;Hampel et al., 2004, León et al., 2008;Zúñiga-Rivero et al., 2010).According to ornburg and Kulm (1981), two trench-parallel structural highs were formed on the continental shelf and upper slope of the Peruvian margin during Late Cretaceous-early Paleogene time, namely the Outer Shelf High and the Upper Slope Ridge.ese two ridges subdivided the Peruvian offshore into a series of basins, which may be separated into an inner set of shelf basins and a seaward set of slope basins.e East Pisco Basin in southern Peru formed in the forearc between 13° and 15° S and is separated from the adjacent West Pisco Basin, which is located on the slope, by the Outer Shelf basement high.e sedimentary fill of the East Pisco Basin includes the middle Eocene Caballas Formation, the middle to upper Eocene Paracas Group, the uppermost Eocene to lower Oligocene Otuma Formation, the uppermost Oligocene to middle Miocene Chilcatay Formation, and the upper Miocene to Pliocene Pisco Formation (Dunbar et al., 1990;DeVries, 1998).ese formations are bounded by regionally beveling, conglomerate-mantled unconformities that are locally accompanied by angular discordances.
Due to onlap and loss of successively younger beds at the base, in the study area the Pisco Formation forms a northeastward-thinning wedge which is little deformed and with a general dip of strata between 4° and 7° to the northeast.Strata of the Pisco Formation can be subdivided into three allomembers or depositional sequences (designated P0, P1 and P2 from oldest to youngest) that are tens to hundreds of meters thick and broadly fining-upward in aspect (Di Celma et al., 2016a, b, 2017, 2018a).A typical sequence comprises a lower, sandstone-rich nearshore succession passing upsection, through intermediary facies, into a monotonous succession dominated by diatomaceous mudstones.e allomember boundaries are taken at intraformational erosion surfaces of basinal extent and indicate abrupt seaward shi in facies.
e excellent agreement between biostratigraphic data and a few well-constrained 40Ar/39Ar radiometric ages from volcanic ash layers provides useful constraints on the depositional age of the three Pisco sequences.At present, due to the lack of direct biostratigraphic and radiometric data, the age of P0 is constrained between by the age of the upper portion of the Chilcatay Formation (18.02±0.07Ma from 40Ar/39Ar radiometric ages on a volcanic ash layer from the top of the Chilcatay strata at Cerro Submarino, Di Celma et al., 2018b) and the onset of deposition of the overlying P1 allomember strata at Cerro Colorado (dated between 9.5 and 9.10±0.04Ma by means of diatom biostratigraphy and 40Ar/39Ar dating); the base of P2 allomember is younger than 8.5 Ma (Gariboldi et al., 2017).
e cetacean specimen described herein was found by one of the authors (M.U.) at Pampa Corre Viento (indicative geographic coordinates: 14°27' S, 75°45' W), a locality ca.40 km south of Ica and about 7.5 km north-northeast of Cerro Los Quesos (a fossil marine vertebrate-rich locality of the Pisco Formation, Bianucci et al., 2016a) (Fig. 1A).e sediment succession exposed at Corre Viento belongs to the lower portion of the P2 depositional sequence of the Pisco Formation and, therefore, an age of less than 8.5 Ma can be inferred for the specimen described in this study (Fig. 1B).

Description of MUSM 887
e fossil specimen consists of an almost complete skeleton, which is partially articulated and partially enclosed in two distinct blocks of indurated sediment (Fig. 2).e main block contains the cranium, both mandibles, parts of the pectoral girdle and forelimbs, an anterior section of the vertebral column, and parts of the rib cage.e short-snouted cranium is complete, lying in le dorsolateral view, and the mandibles are substantially preserved in situ, only slightly shied with respect to each other at the level of the mandibular symphysis.Both the maxilla and the mandibles support several sharp, tiny teeth.Posterior to the skull, a wellarticulated segment of the vertebral column is preserved, with the atlas still connected to the skull (via the right occipital condyle) and to the axis.e other cervical vertebrae are strongly anteroposteriorly flattened and possibly fused to each other; they are followed in anatomic connection by at least the first thoracic vertebra.e rib cage is collapsed, and the ribs, most of which are complete, rest horizontally around and above the articulated vertebrae.Among ribs, a detached vertebral epiphysis has been detected.e almost complete le scapula is superimposed on some ribs and, partially, on the anteriormost section of the vertebral column and on the first sternebra (i.e., the manubrium).Two ribs lying under the le scapula have induced on the latter two systems of load fracturing.e le humerus is still articulated to the corresponding scapula.Additionally, two isolated fragmentary vertebrae (including a posterior thoracic) have been identified close to the collapsed remains of the rib cage.e second block contains thirteen fully articulated lumbar to caudal vertebrae, some of which (the anteriormost) being covered in part by the fragments of at least five ribs (Appendix Fig. 1).Moreover, a single isolated vertebral epiphysis has been detected.At a variable distance ranging from ca. 1 to 10 cm of the pontoporiid bones, a thin but conspicuous layer of blackish, stiffened sediment runs all around the cetacean remains, completely wrapping the partial skeleton (Fig. 2).

Systematic attribution of MUSM 887
Based on the observation of i) anteroposteriorly elongated nasals; ii) a proportionally low skull vertex; iii) the presence of maxillary crests on the antorbital processes; iv) the presence of premaxillary eminences (i.e., inflated premaxillary sac fossae region), this specimen is unambiguously attributed to the family Pontoporiidae (e.g., De Muizon, 1988;Lambert and Post, 2005).Due to the observation of i) a dorsoventrally and transversely expanded lacrimal, which constitutes the major part of the enlarged antorbital process; ii) a deep and narrow antorbital notch, MUSM 887 is assigned to the genus Brachydelphis De Muizon, 1988, as redefined by Lambert and De Muizon (2013).is attribution is further corroborated by the observation of a weak transverse constriction of the premaxilla at the level of the base of the rostrum, a character observed in Brachydelphis, but not in the other pontoporiid genera known to date (Gutstein et al., 2009;Lambert and De Muizon, 2013).Brachydelphis is a late Miocene pontoporiid genus that is rather common both in the Pisco Formation and in the roughly coeval Bahía Inglesa Formation of northern Chile (De Muizon, 1988;Gutstein et al., 2008Gutstein et al., , 2009;;Bianucci et al., 2016b;Di Celma et al., 2017).On the basis of i) the short, triangular, and upwardly projecting rostrum; ii) the narrow separation between the maxillae across the skull vertex, MUSM 887 can be assigned to Brachydelphis mazeasi De Muizon, 1988 rather than to the other, longer-snouted species B. jahuayensis Lambert and De Muizon, 2013.For what concerns the Pisco Formation, specimens of B. mazeasi were found in the stratal package of the P1 sequence (notably including the Tortonian deposits that form the lowest part of the sedimentary succession exposed at Cerro Colorado, referred to by Di Celma et al., 2016b as the "lower allomember") and the basal beds of the P2 sequence (which in turn correspond to the "Cerro La Bruja vertebrate level" of De Muizon andDeVries, 1985 andDe Muizon, 1988) (Bianucci et al., 2016b;Di Celma et al., 2017).Based on comparison with extant Pontoporia, the total body length of the Brachydelphis mazeasi specimen MUSM 887 can be estimated at about 175 cm; therefore, it is here regarded as well-representative of those small-sized vertebrates (shorter than 2.5 m in total length, as defined previously) whose fossilized skeletons comprise a large fraction of the marine vertebrate assemblages of the Pisco Formation.

Petrography and chemistry of the MUSM 887 bones
e bony remains have been firstly inspected at the MUSM with a hand lens.e compact tissue of the bone is hard and has a dark amber color (Fig. 2), the external surface being blackened by late Mn patinae, while the trabecular tissue exhibits late gypsum/anhydrite crystals partially infilling the intertrabecular cavities.Macroscopically, no traces of abrasion were observed and fragmentation is scarce.
A 1 cm-sized fragment of rib (consisting of both cortical and spongy bone) was broken in smaller fragments for stereoscopic observation and then mounted in resin, sectioned, and polished for subsequent analyses (for analytical methods see Appendix).Reflected-light optical microscopy and scanning electron microscopy-backscattered electron imaging (SEM-BSEI) indicates a well-preserved compact bone, without evidence of low density zones, replacement, or microborings, and without any carbonate concretion (Fig. 3A).e composition of the bone tissue is homogeneous and corresponds to calcium phosphate.Electron microprobe analyses (reported in Table 1) indicate a chemistry corresponding to fluorapatite, with Ca/P ratio of 1.65-1.79and MgO and Na2O both averaging 1.1 wt%, SO3 3.3 wt%, and FeO around 0.1 wt%.Analytical totals for the electron probe analyses are in the range 90-93 wt%.

Petrography and mineralogy of the sediment enclosing MUSM 887
Two samples of the sediment enclosing MUSM 887 were collected: the first from nearby the bones and the latter from the dark boundary layer under the specimen (Fig. 2).Both samples are made of siliciclastic sand, consisting of terrigenous rock fragments and crystals (plagioclase, ilmenite, chlorite, quartz, and amphibole), volcanogenic glassy clasts, and minor biogenic (diatoms) components.e sediment is yellowish close to the bones and reddish out of the black boundary layer.
Observations under reflected-light optical microscope and SEM-both secondary electron (SE) and backscattered electron (BSE) imaging -reveal that the black layer (Fig. 3B, C, D) consists of Mn oxides cementing the sediment for an average thickness of 3-4 mm and also filling the fine pores of diatom frustules.e upper edge of the Mn-rich layer (i.e., its transition to the yellow sediment hosting the bones) is irregular, but sharp and well-defined, whereas its lower edge (i.e., its transition to the reddish sediment far from the cetacean remains) is gradual.
Between the Mn-rich layer and the skeleton, calcium phosphate rods or needles about 2-4 µm in length and smaller grains permeate the sediment (Fig. 3E).eir small size prevented from obtaining spectra on single crystals by means of Energy Dispersive X-ray Spectrometry (EDS); however, the calcium phosphate composition was determined collecting several spectra on raster areas measuring about 5×5 µm (Fig. 4).eir P-Ca composition, small size, and occurrence suggest a mineral of the apatite group and a diagenetic (rather than clastic) origin.
Below the Mn layer, Fe-oxides or hydroxides are abundant in the matrix, whereas calcium phosphate crystallites are absent.
X-rays Powder Diffraction data, aimed at detecting the presence of carbonate phases in the sediment, did not reveal calcite or dolomite either above or below the black Mn-rich layer.

A field example of Mn layer under a fossil marine vertebrate in the Pisco Formation
In several cases, the dolomite concretions enclosing fossil marine vertebrate remains of the Pisco Formation were observed being delimited outwards by a dark boundary, with a characteristic sequence of colors corresponding to concentrations of Mn and Fe (Gariboldi et al., 2015).For one of these Mn layers, sampled at Cerro Los Quesos in correspondence of a balaenopteroid (rorqual) baleen whale specimen with a moderately developed dolomite concretion (Fig. 5A, B; specimen M58 in Bianucci et al., 2016a), EDS (electron dispersion spectroscopy) elemental maps were collected and overlaid on SEM electron images (Fig. 5; Appendix Fig. 2).Single element maps of Al, Ca, Fe, Mg, Mn, P, S and Si have been run for the dark boundary layer and the overlying and underlying sediments; additionally, four combined maps have been generated: Ca+P, Fe+Mn, Fe+Mn+Ca+P, and Mg+Ca+Mn+Fe+Si+P+S.is latter map summarizes the distribution of all the elements mapped individually and shows the BSE image of the slide on the background (Fig. 5C).
Maps of Ca, P, and Mn reveal a concentration of these elements in defined millimetric layers.Ca is scattered in submillimetric spots all around the slide and concentrated in a layer at about 3.5-5 mm from the top of the slide where P, also, is concentrated (Fig. 5D, E); the scattered Ca spots are easily interpreted as gypsum, comparing the Ca map with the S map (Appendix Fig. 2), while the superposition of the Ca and P maps (Fig. 5F) suggests that the highlighted layer is a Ca-phosphate layer.e Mn map (Fig. 5G) highlights that this element is concentrated in a discontinuous layer, 1.0 to 2.5 mm thick, inside and roughly parallel to that of the apatite layer, from which it is separated by ca.1.0 to ca. 2.5 mm of sediment (Fig. 5C).

Discussion
4.1.Deposition, burial and mineralization of the bones e overall organization of MUSM 887 in two distinct blocks, each of which is comprised of anatomically coherent and rather articulated bones, suggests an early burial of the carcass by so sediment, followed by gravitational collapse of the body due to decay-related fluidization of the so tissues, and consequent loss of connectivity of the skeletal elements (e.g., Reisdorf et al., 2014).Moreover, a moderate degree of deformation and bone displacement during diagenetic compaction of the host sediment (e.g., Reisdorf et al., 2012) cannot be excluded, as possibly supported by the observation of load fracturing on the le scapula of MUSM 887 (see above).Although early burial due to the action of bottom currents (possibly concurring to bone displacement and re-clustering) or to small submarine landslides cannot be excluded, these hypotheses could not be validated in the field.Detachment of portions of the appendicular skeleton before complete burial should also be considered, as suggested by the lack of bony elements of the le forelimb distal to the humerus.e bones are very well preserved.e microscopic and microanalytical investigations indicate a good degree of mineralization of the bone tissue, now consisting of fluorapatite.Such a degree of mineralization suggests that the post-mortem degradation of collagen within bone tissues was followed, during the early stages of diagenesis, by recrystallization of the original hydroxylapatite, increase of crystallite size, and a consequent reduction of porosity (Trueman and Tuross, 2002;Keenan, 2016).e new Ca-phosphate phase, namely fluorapatite or carbonated fluorapatite (Pfretzschner, 2004;Wopenka and Pasteris, 2005), produced by recrystallization of bone hydroxylapatite has been demonstrated to be more stable, even in aqueous solutions at relatively low pH or low phosphorus concentrations (Keenan, 2016).erefore, the presence of environmental conditions facilitating bone recrystallization during early diagenesis was a predominant parameter for the preservation of this beautiful specimen.

The Mn-rich layer
e evident Mn-rich layer around the specimen delimits a volume of sediment some centimeters away from the bones.Mn occurs also as patinae at the surface of the bones, but in this case it seems to follow infiltration paths, thus suggesting a late origin (see also Pfretzschner and Tütken, 2011).Associated to a high concentration of diagenetic Ca-phosphate (Fig. 3), the Mn-rich layer is unrelated to the late Mn patinae.Indeed, would it be a late feature, it would be difficult to explain the sharp boundary with the surrounding sediment, which remained permeable as indicated by the presence of late gypsum spots.
In the marine environment, Mn is soluble in the forms Mn2+ and MnCl+, whereas it forms oxide particulate as Mn(IV) in oxic waters (e.g., Calvert and Pedersen, 1993;Tribovillard et al., 2006).Mn-rich layers may form by precipitation of Mn-minerals from seawater at a redox boundary, located below or at the sediment-water interface, as a consequence of the redox-dependent behavior of this element.e presence of reactive organic carbon at sea bottom or within sediment has a main role in governing the Mn reduction process.High concentrations of Fe may accompany Mn layers, due to the similarly redox-dependent behavior of this element.However, iron precipitation may not coincide with Mn layers since, differently from Mn, it is possible in suboxic and in sulfidic conditions.Based on this framework, the preservation of manganese and iron layers in lacustrine and marine sediments can be studied to unravel early diagenetic processes (e.g., Burdige, 1993;Van Santvoort et al., 1996;De Lange et al., 2008;Torres et al., 2014;Löwemark et al., 2014).
e Mn layer below MUSM 887 seemingly encloses the fossil specimen (Fig. 2).Several similar examples, as that of specimen CLQ-M58 (see paragraph 3.5), have been observed in the field, suggesting the same relationship (Fig. 5).erefore, it can be suggested that the formation or preservation of the Mn-cemented dark boundary is strictly related to the presence of the carcass: it delimits the volume of sediment that was affected by chemical conditions differing from the contemporary surrounding sea bottom as a consequence of the decay of the odontocete carcass.
e sedimentary palaeoenvironment of the highly fossiliferous sandstones that constitute the basal beds of the P2 sequence (i.e., the "Cerro La Bruja vertebrate level" exposed at Corre Viento) was a shallowmarine upper shelf environment (e.g., Di Celma et al., 2017) with an oxic water column and no evidence of anoxic conditions at the seafloor.erefore, the oxidation front was located below the sediment-water interface, favoring the formation of a concentration of Mn in the upper sediment layer (Fig. 6A).We can suppose that, aer the death of a marine vertebrate, aerobic remineralization of organic matter was the main, although not the sole, mechanism as long as the remains were exposed on the seafloor.In case of large-sized marine vertebrate falls, the oxygen recharge may be locally insufficient for organic matter oxidation, even if the conditions at the seafloor are generally oxic (Allison, 1988).erefore, anaerobic reduction of Mn, nitrate, and Fe, as well as other biogeochemical processes such as sulfate reduction and methanogenesis, occur at the seafloor, governed by the microbial ecosystem of whale-falls (Allison et al., 1991;Shapiro and Spangler, 2009;Danise et al., 2012).ese processes induce local geochemical gradients in the underlying sediments and in bones, e.g., in pH, alkalinity, and sulfate/sulfide equilibrium (Treude et al., 2009).In the case of small-size marine vertebrates, such as the pontoporiid specimen MUSM 887, organic matter degradation on the seafloor may be accomplished mainly by aerobic processes.However, the early burial of MUSM 887 proposed to explain the arrangement of the preserved bones could have favored the above mentioned anaerobic biogeochemical processes, fueled by the organic matter (deriving from both so tissues and bone lipids).In this locally oxygen-poor environment, the decay of organic matter could occur through biomediated processes of reduction of Mn, nitrate, Fe, and sulfate.ese processes can be fingerprinted by the consequent precipitation of specific authigenic minerals.As regards sulfate reduction, mineralogical evidence such as those found with the fossil vertebrates in dolomite nodules -relics of pyrite framboids and the dolomite itself-are absent in the MUSM 887 specimen.e Mn pattern, instead, may be interpreted in the frame of early diagenesis.
Mn may have rather high concentrations in seawater when the terrigenous input is high.In an oxic water column, it undergoes syn-sedimentary burial as Mn oxyhydroxide particulate (Tribovillard et al., 2006): the latter can be partly or completely reduced within the sediments below the oxic-anoxic boundary and then, with the upward migration of porewater, reprecipitated at the anoxic-oxic boundary, resulting in an upward migration of Mn that may or may not result in a local increase in Mn concentrations (Burdige, 1993).In the locally anoxic environment created by the decay of the Corre Viento cetacean carcass at the sea bottom, and in case of Mn availability within bottom sediment, Mn-reduction could have been one of the most prominent mechanisms for organic matter remineralization.Additionally, the lowering of pH due to bacterial decay of so tissues favored Mn solubility.As a consequence, the increased Mn concentration in porewater resulted in the precipitation of Mn minerals at the anoxic-oxic interface and/or at an alkalinity interface, representing the boundary above which porewater was chemically affected by the processes of degradation of the carcass (Fig. 6B).Indeed, this Mn-rich layer would not have developed in case of deposition of the dolphin carcass on an anoxic sea bottom.

Insights from the presence of Ca-phosphate within sediment
In the marine environment, phosphorus sourced by continental weathering is uptaken by microorganisms in the zone of primary productivity and incorporated into organic matter.Upon organic matter oxidation, dissolved inorganic phosphate is re-circulated by upwelling.e degradation of organic matter is, thus, a main source of P in marine sediments (Föllmi, 1996;Filippelli, 2008;Cosmidis et al., 2013).With an oxic water column and in conditions of high primary productivity such as those of the Corre Viento shelf depositional environment, dissolved phosphate is vehiculated to the seafloor together with Fe3+ (as FeOOH•PO43-) and is then released to the porewater following Fe-reduction in anoxic conditions below the water-sediment interface (Kim et al., 1999;Cui et al., 2016).In the case of the studied pontoporiid, anoxic conditions were induced in correspondence of the carcass by decay processes.Consequently, Fe-reduction processes and, possibly, the decay of the carcass itself, released P increasing its concentration in porewater close to the pontoporiid remains, inducing authigenic calcium phosphate precipitation at the boundary with suboxic conditions (Fig. 6).e slight lowering of pH relative to normal seawater that is necessary to stabilize calcium phosphate with respect to carbonate could have been accomplished by the bacterial decay of so tissues (Briggs and Kear, 1993;Dornbos, 2011).
e presence of Ca-phosphate immediately below the specimen is, thus, an evidence of availability of P in porewater during the very early diagenesis: this is a factor of prime importance for bone apatite recrystallization and, consequently, for preservation (Keenan, 2016;Keenan and Engel, 2017).erefore, the geochemical environment in which the pontoporiid was set once the labile tissues were decayed (probably on a timescale of weeks) can explain the high degree of mineralization observed for the compact bone of the pontoporiid ribs.e sediment hosting the pontoporiid does not present evidence of a phosphogenic depositional environment, although the above mechanisms are similar to those suggested to explain phosphogenesis along the Peruvian margin (Arning et al., 2009).Accordingly, the specimen does not present borings and phosphatization crusts on the bones and evidence of microbially mediated sulfate-reduction processes such as pyrite relics.Based on the above reported evidences, we suggest that the pontoporiid cetacean carcass was partially buried in permeable sediments soon aer death, prior to the decay of most of the so tissues.is scenario is in agreement with the degree of articulation and completeness of the specimen, and allowed the establishment of a local anoxic/suboxic micro-environment within a generally oxic sea bottom, which in turn explains the development of the observed patterns of apatite and Mn-oxide authigenic minerals.

Preservation of fossil vertebrates with and without concretion
Differently from several specimens in the Pisco Formation, in the case here studied, the fossil is not enclosed in a dolomite concretion.
Dolomite formation is inhibited by the high dissolved sulfate ion concentration in seawater (Baker and Kastner, 1981).e microbially mediated sulfate reduction is able to lower the sulfate ion inhibitory effect and, concomitantly, increases alkalinity (e.g., Compton, 1988).In agreement with the lack of dolomite, we did not find evidence of former pyrite framboids, typically a result of sulfate reduction, associated to the bones.We speculate that the Corre Viento sandy sediments allowed the porewater below the water-sediment interface to exchange with seawater in a rather efficient way.erefore, even with sulfate reduction, the moderate permeability of the sediment enclosing the pontoporiid carcass, coupled with the small body size of the specimen, would have prevented the concentration of dissolved sulfate (inhibiting dolomite formation) to decrease enough (or for long enough) for allowing extensive dolomite formation (Vasconcelos et al., 1995;Meister, 2013).
Interestingly, the neurocranium of MUSM 887 is well preserved and barely deformed (Fig. 2).is could be explained by hypothesizing that the inside of the brain cavity is filled by early-formed diagenetic minerals, very likely dolomite, as it is commonly observed in fossil vertebrates of the Pisco Formation, including smallsized animals (Gariboldi et al., 2015;Gioncada et al., 2016; and field observations of numerous dolomitefilled inner casts of brain cavities).Should it be confirmed (by non-invasive methods such as microdrilling), the specimen could be regarded as an example of partial concretion.e formation of early dolomite could be explained by the fact that the inside of the brain cavity is an environment with abundant organic matter (mostly consisting of easily degradable lipids) and at the same time protected from exchanges with seawatera combination of factors that facilitates sulfate reduction and the establishment of local conditions favorable to precipitation of carbonate, even before the complete burial of the skeleton.
As mentioned above, several dolomite concretions of the Pisco Formation are delimited by a dark Mn-rich boundary.e SEM elemental maps in figure 5 indicate that, also in the case of the balaenopteroid specimen M58, this boundary consists of a subtle layer permeated by Mn minerals and overlain by a layer with abundant authigenic apatite.is observation strongly suggests that the same pattern of diagenetic minerals found close to the pontoporiid specimen MUSM 887 studied in this work also developed around specimen M58, most likely as a result of similar biogeochemical processes related to the decaying organic matter.Varying development of the Mn layer in these two specimens is probably due to different environmental availability of Mn at the two sites.
e lack of dolomite in the sediment enclosing MUSM 887 suggests that the fall of this small-sized cetacean carcass did not induce a local variation of the chemical environment intense enough to reach conditions favorable to the formation of a carbonate concretion.As already suggested elsewhere (Gariboldi et al., 2015), this record identifies the amount of decaying organic matter -directly related to the body size of the dead animal -as one of the most prominent factors controlling the development of concretions linked to altered chemical environments surrounding marine vertebrate carcasses lying on the seafloor.anks to embedding in a relatively permeable sediment devoid of any dolomite nodule and to the availability of P, the recrystallization of bone apatite was favored, allowing excellent bone tissue preservation.erefore, the taphonomic scenario proposed in the present work contributes to bridge two emblematic modes of preservation of the fossil marine vertebrates of the Pisco Formation: i) preservation outside any nodule or concretion, sometimes exhibiting prominent phosphatization of the vertebrate remains (e.g., Marx et al., 2017), and ii) preservation within a fully developed, hard, and impermeable dolomite concretion (e.g., Gariboldi et al., 2015;Gioncada et al., 2016) with modest bone mineralization.More generally, different patterns in the formation of diagenetic minerals can be expected, depending on factors such as the time between death and burial, the oxygen level at the sea bottom, permeability and composition of the sediment, and the amount of so tissues and bone lipids available for decay.Interpreting these different patterns may provide significant insights into the early fossilization history of well-preserved fossil vertebrates.

Conclusions
A partial skeleton referred to the extinct small dolphin species Brachydelphis mazeasi (Pontoporiidae) has been described from the late Miocene site of Pampa Corre Viento (Pisco Formation, southern Peru), and microscopical and microanalytical investigations were undertaken both on it and on the surrounding sediment.e overall excellent preservation state of the cetacean specimen and remarkable articulation degree suggest that it was buried soon aer death.
Early diagenetic processes triggered by the anaerobically degrading cetacean carcass are revealed by the types and patterns of diagenetic minerals found around the pontoporiid skeleton.Conditions favorable to precipitation of calcium phosphate, but not dolomite, around the specimen allowed increased mineralization of the bones.e presence of a dolomite nodule may be hypothesized inside the brain cavity, preventing its deformation under lithostatic pressure.
As a general consideration, the observation of regular and recurrent patterns for the distribution of diagenetic minerals in the sediment enclosing fossil vertebrates, without or with limited carbonate concretions, gives insights into the early taphonomic history of these remains, just as dolomite concretions do.In the case of the pontoporiid skeleton here investigated, this pattern reveals a situation favorable to preservation, with early burial and the establishment of local anoxic/suboxic porewater environment, even if the surrounding seafloor environment and water column were characterized by fully oxic conditions.

Fig. 1 .
Fig. 1. A. Location of Corre Viento, the fossil-bearing locality where the skeleton of the small dolphin Brachydelphis mazeasi MUSM 887 was found, in the Ica Desert (Peru); B. Schematic, dip-oriented allostratigraphic diagram for the Pisco Formation and the lower Chilcatay and Otuma Formations (not to scale), showing the indicative stratigraphic position of the strata of the late Miocene P2 sequence exposed at Corre Viento.e Pisco Formation exhibits pronounced thinning to the northeast, with the allomembers onlapping onto the basal composite surface PE0.0 and arranged in a retrogradational pattern, progressively offset to northeast.Figure 1B modified from Di Celma et al. (2017).

Fig. 2 .
Fig. 2. e block containing the remains of the skull, bony thorax, and le forelimb of the small dolphin Brachydelphis mazeasi MUSM 887, with corresponding line drawings indicating the different elements preserved.

Fig. 3
Fig. 3. A. SEM-BSE detail of a fragment of the rib of Brachydelphis mazeasi specimen MUSM 887 in transverse section.e le margin of the sample represents the inner part the rib, consisting of cancellous tissue and exhibiting late gypsum/ anhydrite crystals within the pores of the bone.e cortical tissue shows neither microborings, nor other evidence of erosion; B. SEM-BSE overview of the black layer observed in the sediment enclosing MUSM 887.e yellow sediment close to the bones corresponds to the dark-gray domain on the upper portion of the sample.e Mn layer corresponds to the wide whitish area in the central portion of the sample, in which the white cement consists of Mn oxide minerals.e boundary between the Mn layer and the yellow layer is sharp and well-defined.e red sediment corresponds to the lightgray domain in the lower portion of the sample.e boundary between the Mn layer and the red sediment is not well-defined; C. Detail of the gradual boundary between the Mn layer and the surrounding red sediment; D. Close-up of the Mn layer: Mn oxides cement terrigenous sand particles; E. SEM-SE image of pervasive authigenic apatite micro-crystals within the yellowish sediment.A diatom specimen (referred to alassionema nitzschioides) is visible in the centre of the panel (arrow).

Fig. 5 .
Fig. 5. Elemental maps of sediment below the fossil balaenopteroid whale CLQ-M58 at Cerro Los Quesos (location of specimen M58 in the map of the fossils of Cerro Los Quesos provided byBianucci et al., 2016a).A. Specimen CLQ-M58 in the field, modified fromGariboldi et al. (2015).e red rectangle highlights the portion of sediment shown in B;B.Close up of the sediment underlying the fossil.e Mn layer is clearly visible (letter "B") and separates two portions of sediments characterized by two different colors, yellow (Y) and red (R).is pattern of sediment colors is oen found below the fossil skeletons of the Pisco Fm.(Gariboldi et al., 2015); C. Composite elemental map of the yellow-black-red sequence underlying CLQ-M58, run on a polished mount of sediment embedded in resin and sectioned orthogonally to the strata; D. Ca elemental map; E. P elemental map; F. Ca and P composite elemental map (apatite); G. Mn elemental map.

Fig. 6 .
Fig. 6. A. Manganese, iron and phosphorus behavior at an oxic sea bottom (modified from Cui et al., 2016), OM: organic matter; B. Schematic model summarizing the hypothesized formation of the mineral pattern around the Corre Viento pontoporiid skeleton under study.

Fig. 2 .
Fig. 2. Combined and single elemental maps run on a polished mount of sediment embedded in resin and sectioned orthogonally to the strata.e sediment was sampled from the black layer below M58 specimen from Cerro Los Quesos (see figure 5 in the main text).A. Fe+Mn map; B. Fe+Mn+Ca+P map; C. Al map; D. Fe map; E Mg ma; F. Si map; G. S map.