Unraveling the Strange Case of the First Canarian Land Fauna (Lower Pliocene)

Abstract

Geological data of the region indicate that the Canary Islands have not been connected to the mainland before. However, fossil evidence suggests some kind of faunal exchange with Africa during the late Neogene. After extensive field work during past years, a re-evaluation of the fossil remains of the first terrestrial vertebrates that settled and thrived on the Canary Islands is presented, with special attention to the long-debated identity of birds that laid large-sized eggs, reported some decades ago on Lanzarote Island. The age of the eggshell-bearing deposits has been recently updated as Early Pliocene (ca. 4 Ma). The dispersal mode of these terrestrial birds to reach the island was an unsolvable challenge in previous studies because the regional geography of the sea bottom was neglected, as well as the chronological succession of events in the formation of the Canary Eastern Ridge, which increased attention to a unique case of arrival of ratites on an island never before united with the mainland. The few animals found in northern Lanzarote (ratites, snakes, turtles, terrestrial snails and bite marks on eggshells pointing to a jagged and unknown large predator) probably made the sea crossing from the mainland in different ways. Two scenarios are contemplated. In both, the circumstances facilitating the faunal transit from Africa to the Canaries ceased after the early Pliocene, around 4 Ma, since these animals have never managed to cross the Canary Channel again.

1. Introduction

In the mid-twentieth century, an unusual find was revealed in Neogene deposits then dated as from the upper Miocene in northern Lanzarote Island, Canary Islands. Rothe (1964, 1974) [1,2] and Sauer & Rothe (1972) [3] found many eggshells, which they had attributed to two ratite groups, Struthio and an indeterminate aepyornithoid type. The authors did not provide any hypothesis in accordance with the regional geological history to explain the arrival of these animals to the island. A subsequent study [4] did not confirm the previous results and fueled the controversy over the taxonomical identity of these findings, although renowned specialists accepted the first taxonomic determinations [5], which were incorporated into the body of the specialized literature. Three major paleontological sites have provided hundreds of fragments of ratite eggshells and several complete or nearly complete eggshells. The most abundant fossils are terrestrial gastropods. Hundreds of turtle eggshells and internal casts of insect cells have also been collected. Two vertebrae of a small snake, one of them previously known, and an egg of a neognath bird have been collected at the Valle Chico site [6].

The Canary Islands are located in the North Atlantic, off the west coast of Africa, between the coordinates 27°37′ to 29°25′ N and 13°20′ to 18°10′ W. The archipelago is made up of seven main islands and several islets. Fuerteventura is the closest island to the African continent, about 100 km from mainland, while the northern massif of Lanzarote is located about 140 km from the current coast of the continent. Fuerteventura is separated from the current Lanzarote by a narrow strait 11 km wide, although during the time span under study, the central part of Lanzarote had not yet been formed, and there were two emerged volcanic massifs, one in the north (Famara) and one in the south (Ajaches). The Canary current is a part of the North Atlantic gyre and flows southwest currently, although water circulation of the deep Atlantic Ocean appears to have experienced changes in the past, in accordance with climatic fluctuations [7].

The monophyly of the ratites has become a classic question of study. Morphological studies support monophyly of ratites, with the South American tinamous as its closest relatives [8,9]. Conversely, phylogenetic studies using nuclear DNA sustain ratite paraphyly [10]. Ratites includes the extant ostrich (presently restricted to Africa), emu (Australia), cassowary (Australia and New Guinea) and rhea (South America) and also the recently extinct moas (New Zealand) and elephant bird (Madagascar). The oldest forms attributed to ratites are found in the Paleocene of France, the Eocene of Brazil and Germany and the Oligocene of Australia [11,12,13,14,15,16]. North America and Antarctica are the only continents where apparently these birds were not present during the Neogene.

The islands have been considered laboratories of biological evolution partly because the isolation to which the animals are subjected eliminates some of the variables that affect the continental populations. This fact, together with the large number of ratite eggshells that have been unearthed from the paleontological sites here studied, has allowed us to reevaluate the results of previous publications, bearing in mind that some of the first seminal studies have obscured subsequent works, mostly due to insufficient comparative material.

Dozens of tortoise eggshell fragments have also been found in these sites in the north of Lanzarote. The first reports of fossil tortoise in Canary Island date to the 20th century and were based on shell and eggshell remains from the Adeje Quarry site (Tenerife). They were originally discovered by O. Burchad but finally studied in detail by Ahl (1926) [17], who erected the species Testudo burchardi. Loveridge & Williams (1957) [18] proposed that all the European fossil giant tortoises as well as Canarian forms should be transferred into the genus Geochelone. In fact, this proposal was adopted by López-Jurado & Mateo (1993) [19], who erected a second species from Gran Canaria Island named Geochelone vulcanica. According to Hutterer et al. (1997) [20], four different taxa were present in the Canary Islands. Two of them were described on skeletal elements, Geochelone burchardi (Tenerife Island) and Geochelone vulcanica (Gran Canaria Island), and two species within Geochelone were identified on eggshell material from Lanzarote and Fuerteventura. The attribution to Geochelone was accepted by most subsequent workers until recently, when Rhodin et al. (2015) [21] suggested that fossil tortoises from the Canary Islands belong to Centrochelys. However, the authors did not provide a discussion for either of these two taxa, and consequently the taxonomic assignment is unjustified. Therefore, the taxonomy of Canary giant tortoises is still a subject of debate. Given the absence of skeletal material in Lanzarote and Fuerteventura islands, the taxonomic assignment was based on the size of the eggs. The Fuerteventura tortoise would be related to those of Gran Canaria Island [22]. In contrast, Lanzarote tortoises had smaller eggs and formed a different group of tortoises closely related to northern African extant tortoises, such as Centrochelys and Stigmochelys [20].

A snake vertebra attributed to an indeterminate boid (Boidae in the traditional sense, i.e., including pythonines, boines and erycines) was described in the Neogene of Lanzarote [6]. This vertebra is included by these authors in the family Boidae. However, the recent molecular data show that this set constitutes a paraphyletic group whose relationships are still discussed [23,24,25,26,27]. In this work, we use the term “boid-like” to indicate a homogeneous vertebral morphology found in Pythonidae, Boidae and Calabariidae [25].

Pore pattern descriptions (morphotypes) are used in this paper because they have been used in previous literature. Morphotypes are structural descriptions and should not be confused with taxa or parataxa because they do not belong to either of these two classification systems. Until now, a significant portion of ratite eggshells—particularly those of the aepyornithoid type—have not been classified in ootaxonomic systematics. Including all eggshells within a single conceptual framework would certainly help understand their genesis and relationships. That is not the task addressed in this report. First, it is essential to clarify errors in the description of eggshell types from Lanzarote because they have influenced subsequent specialized literature and have led to new ones.

Geological Framework and Age of the Localities

The Canary Archipelago was formed during the Neogene and Quaternary [28], and its geological constitution is complex. All of the islands’ edifices have been formed exclusively as a result of eruptive processes, even though sedimentary processes have also been present in their geological history, in the form of intercalated deposits. On each island, different construction phases are distinguished, as well as stages with predominance of erosive phenomena. The seafloor depth in the region is about 3000 m. This seabed is one of the oldest in the Atlantic because it was formed between about 175 and 150 Ma. The Canary Island Seamount Province may be the oldest hotspot in the Atlantic Ocean and the most long-lived mantle anomaly on earth [29].

The two easternmost islands, Lanzarote and Fuerteventura, share the same marine platform, which was also the first volcanic structure that appeared in the region, about 60 Ma ago. Both islands, together with La Concepción bank, constitute the East Canary Ridge, which extends 400 km along a NNE-SSW direction, sub-parallel to the continental margin. The archipelago’s first subaerial volcanism is recorded on Fuerteventura Island, with dates of 25 Ma, but the first submarine eruptions in the region offer a dating of about 60 Ma. Some models have been proposed to explain the Canarian volcanic dynamics. The formation of Fuerteventura Island predates that of Lanzarote. Fuerteventura had reached almost its current form and extent just before the first magmatic emissions started to build the emerged foundations of Lanzarote [30,31]. Two huge shield stratovolcanoes began to emerge in the south and north of what is now the island of Lanzarote at about, respectively, 15.5 and 10.2 Ma [30]. The current Los Ajaches and Famara massifs are the results of an intense and prolonged erosion of these previous volcanic structures.

Most of the fossil items come from three sites: Valle Grande, Valle Chico and Fuente de Gusa. The fossiliferous beds are sedimentary deposits intercalated between basalt lava flows of the northern area of the Famara massif. A specific study devoted to the stratigraphy of the sites [32] has shown their spatial correlation and age interval, and a sedimentological interpretation of the original environments. The K/Ar age interval ranges from ca. 4.3 Ma to 3.8 Ma, within the Early Pliocene (Figure 1). The principal component of the deposits is a bioclastic calcarenite of aeolian origin (sand sheet deposits), which is present in all three sites and constitutes 65% of the beds. The remaining 35% is of fluvial-aeolian origin, mainly occurring as stream deposits. The local paleogeography consisted of a flat plain, slightly inclined to the E and NE, over which aeolian sands moved freely with a prevailing NNE-WSW wind direction.

2. Materials and Methods

The material studied for this article is new. It comes from nine excavation campaigns carried out in three paleontological sites (Valle Grande 1, Valle Chico and Fuente de Gusa) (Figure 1). The material is kept in the Cabildo of Lanzarote. None of the finds have an inventory number or code. However, the codes we assign here will be incorporated into the Cabildo of Lanzarote’s inventories. All material used directly for this paper, except for the whole eggs, is kept in a box labeled “Valle Grande, Valle Chico, Fuente de Gusa—Special Material.” Several whole eggs have been unearthed. They appear crushed and deformed to a greater or lesser extent as a consequence of the sediment pressure. The eggs’ volume and weight can be estimated quite accurately from their linear dimensions. Schönwetter (1930) [33] used a formula for calculating the volume of current ostrich eggs, which was later tested [34]. Hoyt (1979) [35] evaluated the accuracy with which egg volume and weight of bird species can be calculated from linear dimensions, using the following equations: $volume = k_v·L·B^2$ and $weight ={ k}_w·L·B^2$, where L is the length of the longitudinal axis, and B is the diameter of the equatorial circumference (the maximum circumference perpendicular to the longitudinal axis). $k_v$ and $k_w$ are proportionality constants that depend on the particular species. Their respective values for the extant Struthio camelus are $k_v = 0.521$ and $k_w = 0.597$ [35].

The integrity of the complete eggs unearthed in Lanzarote has allowed their almost perfect ellipsoidal or oval shape to be measured, so that the maximum circumference perpendicular to the longitudinal axis is located in the center of the egg. To find the length of the longitudinal and equatorial axis, the meridional and equatorial contours have been measured. The formula of the ellipse perimeter has been applied: $ellipseperimeter \approx 2\pi \sqrt{{\displaystyle \frac{a^2 +{ b}^2}{2}}}$ (a is the major semiaxis, and b is the minor one) to the longitudinal axis, and the circumference length formula has been applied to the equatorial axis. Using axial lengths, the above formulas have also been applied to find the weights and volumes of ratite eggs found either at various fossil sites or collected in the literature [34,36,37,38,39,40,41,42,43,44,45,46,47], with the aim of comparing them with the data from Lanzarote (Supplementary Table S1).

This volume calculation implies that fossil eggs from Lanzarote resemble those of the extant African ostriches and so are acceptable to take the corresponding species-dependent constants to calculate volumes and weights. Although the eggs are not spherical, this has been done to compare with the results of previous studies. In parallel, the egg volumes have also been calculated from the images taken using a Siemens Somaten Definition 128 dual energy tomograph. The volumes have been obtained with the SYNGO CT VA48A. Egg volumes obtained from computed tomography images provide larger values and are given in

Table 1. Dimensions of the complete ratite eggs found in the sites from the north of Lanzarote and estimation of the body mass of the females. ⁽¹⁾ In millimeters, ⁽²⁾ measurements taken directly on the eggs, ⁽³⁾ obtained from the meridian length, ⁽⁴⁾ obtained from the equatorial length, ⁽⁵⁾ obtained by computed axial tomography, ⁽⁶⁾ in cubic centimeters, ⁽⁷⁾ calculated (see Methods), ⁽⁸⁾ in grams, ⁽⁹⁾ in kilograms.

Complete Eggs (Overall Shape)	Longitudinal Axis A (1,2)	Meridian Lenght (1)	Longitudinal Axis B (1,3)	Equatorial Lenght (1)	Equatorial Axis (1,4)	Egg Volume (5,6)	Egg Volume (6,7)	Egg Weight (7,8)	Female Body Mass (9)
Egg 1 (oval)	183	450	157.6	400	127.4	2051.6	1332	1527	-
Egg 2 (oval)	167	477	118.8	421	134.0	1790.6	1111	1273	-
Egg 5 (oval)	159	420	147.8	370	117.8	-	1069	1224	110.1
Egg 7 (oval)	174	450	164.8	370	117.4	1450.6	1183	1356	128.8
Egg 8 (oval)	178	455	166.4	375	119.4	1315.0	1236	1416	137.8
Eg 10 (oval)	178	470	171.2	391	124.4	1536.4	1380	1582	163.5
Ranges	159–183	420–477	118.8–171.2	370–421	117.4–134.0	1315.0–2051.6	1069–1380	1224–1582	110.1–163.5

as additional information. It seems advisable not to compare these results with those obtained using the above formulas. Consequently, they have not been used to calculate egg weights or to estimate female weights.

The body mass of the females has been estimated from the egg mass calculated from the above formulae (Table 1), according to Dyke & Kaiser (2010) [48], and assuming that the eggs belong to a precocial species.

Egg 5 has been subjected to computer tomography using a Philips Brilliance 64 CT Scan with the following parameters: pixel size of 0.2369 mm, slice thickness of 0.67 mm, and an inter-slice spacing of 0.30 mm, which generated a matrix size of 768 × 768 pixels. Scanner energy was 120 kV and 800 mA. The slices obtained were imported into the image processing software program 3D Slicer 4.5.0.

For comparison purposes, eggshell pieces of Aepyornis maximus deposited (no inventory code) in the Museo de la Naturaleza y el Hombre (Santa Cruz de Tenerife, Spain) have been used, as well as fossil Struthio camelus and Psammornis eggshells collected by two of the authors (A.S.-M. and M.A.P.-B.) in several locations in Western Sahara.

Neognath eggshells are housed in the Faculty of Biology (ULPGC), still without an inventory code.

Some eggshell fragments exhibiting the aepyornithoid 1 morphotype were subjected to artificial mechanical wear simulating the friction eggs experience against the ground during incubation. The goal was to determine whether removing the most superficial part of the shell would reveal a pattern that could be interpreted as aepyornithoid 2.

Historically, fifteen species of aepyornithids have been described in Madagascar. The statistical study performed by Mlíkovský (2003) [49] of the available complete eggs of these ratites shows that they all belong to a single species, Aepyornis maximus Geoffroy-Saint-Hilaire, 1851, the name that holds the priority according to the International Code of Zoological Nomenclature.

3. Results

3.1. Ratite Eggshells

The nomenclature of ratite eggshells in the specialized literature lacks homogeneity. Some of the first studies [43,50,51,52] played an influential role thenceforward. The morphotypes that were described in the first studies, based on the configuration of the levels of the eggshells in radial section and the pattern of pore openings, have continued to be used in subsequent literature, as well as an ootaxonomic approach and a mixture of the previous two [5]. Mikhailov’s (1997) [5] attempt to synonymize all Struthio species based on eggshells within the oogenus Struthiolithus, including Namornis oshanai (originally described as Struthio oshanai) and Diamantornis, did not solve the problem and has created some confusion. Additionally, some supposed apparent correspondence of eggshells with fossil bones deserve to be reviewed. It is far from the aim of the present paper to review and transfer the taxa previously described according to the zoological taxonomy to an ootaxonomical nomenclature. At this point, it seems clearer to refer to previous names, treating the sets of features of ratite eggshells as morphotypes.

Fossil ratite eggshells are frequently found, mostly in Neogene deposits of Eurasia and Africa (Supplementary Table S2), both in stratigraphic contexts and in extensive land surfaces [39,50,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67]. In a radial section, the eggshell of this avian group shows from the outside to inside three mineralized layers—cuticle, palisade and mammillary layers— all of them characterized by distinct structures [5,51,68], in addition to the membrana testacea, of proteic nature, which is not noticed in fossil ratite remains [69]. The description made by Sauer (1972) [51] of the eggshells of Aepyornis maximus found in Madagascar has been very influential in subsequent works when identifying eggshells at sites throughout Africa and Eurasia. For Sauer (1972) [51], the eggshells of aepyornithoid type are characterized by “…the alignment of the linear, here and there bent and forked pore grooves, the short ‘dagger-point’ or ‘comma’ pores, and the oval ‘sting’ pores…”. And on the same page, “It is the pronounced linear pattern that marks the aepyornithid eggshells as unmistakingly different from the struthionid eggshells with their diffusely scattered micropores and the irregularly curved and frequently branched grooves,” as well as “The linear arrangement of the aepyornithid pore openings remains recognizable in the different shell types which range from those with a pronounced presence of linear grooves to those in which the number of grooves is variously reduced in favor of the short dagger-point and sting pores” [51].

A key milestone in the investigation of the ratite eggshells was the discovery of two almost complete eggs and 302 fragments in two paleontological sites in the north of Lanzarote [1], at that time assigned to the Miocene [51]. Rothe (1964) [1] identified two pore patterns in this material, the one observed in the Madagascan Aepyornis maximus and the one in the extant African Struthio camelus camelus, respectively. Next, Sauer [3,51,70] supported such observations, especially in his seminal paper of 1972, where the author described the pore pattern of Aepyornis type (or A type)—the presence of pores grouped in linear parallel grooves, dagger-point pores and sting pores—and the so-called Struthio type (or S type), consisting of “…tiny round pores,… scattered diffusely over the shell surface… the type S pattern resembles most closely that of the S. c. camelus egg.” [51]. However, neither this description of the pattern of pores of type S nor the corresponding figures of the fossil pieces [51] match the pattern of pores characteristic of the eggs of S. c. camelus. In fact, the images of the Lanzarote S type (Sauer & Rothe, 1972: Figure 1a; Sauer, 1972: Figure 14) [3,51] show eggshell fragments that are heavily eroded, where no pore pattern is visible. During incubation, the surface of the ratite egg is subjected to intense friction against the ground, which can erase all or part of the pore pattern (personal observation by ASM at ostrich farms and zoos). Occasionally, only a few widely scattered single pores remain. Eggshell fragments with this appearance are very numerous in the Lanzarote record, indicating that such fragments originate from incubated eggs. However, the appearance they acquire after rubbing against the ground does not resemble the pore pattern of Struthio eggs or any other ratite. It appears that worn eggshell fragments were wrongly assigned to type S (Struthio) by Sauer (Sauer & Rothe, 1972: Figure 1a; Sauer, 1972: Figure 14) [3,51]. Incidentally, he made the same mistake with eggshell fragments from Tsagan (Mongolia) (Sauer, 1972: Figure 23 and Figure 24) [51]. It is also possible that varying degrees of erosion have led some scholars to believe that the pore pattern varied across the egg’s surface. This has not been observed in any complete ratite egg, modern or fossil. In extant S. c. camelus, the pore openings form rounded clusters in small depressions, evenly distributed over the entire surface. As depicted by Schönwetter (1927) [46], the different subspecies have their own pattern. Pore openings appear uniformly distributed on the eggshell, both solitary and grouped in the form of rounded rosettes. The pore canals of Struthio branch extensively. The pore openings, which vary in size according to the subspecies, are clustered in saucer-shaped depressions [71]. These aggregations of pores are, with the exception of Struthio camelus spatzi, in individual depressions, although such depressions may be connected by irregular grooves [46,51]. Henceforward, we refer to the variety of pore patterns exhibited by the subspecies within Struthio camelus as struthioide type. Mikhailov & Zelenkov (2020) [72] have reviewed the ratite eggshells recorded so far throughout the world and attributed them, respectively, to morphotypes A, S and A-S (combination of traits from morphotypes A and S). Other authors have referred to fossil eggshells displaying a similar pore pattern as Psammornis and Struthiolithus [5,43], even though specimens with different pore patterns have been identified as Struthiolithus in the literature [51]. The round shape of the pore openings of the Struthiolithus eggs from the Pleistocene loess deposits of north China emphasizes the similarity to the current common ostrich.

Psammornis pore-pattern eggshells have been considered similar to those of the extant Struthio camelus, mainly after Sauer’s (1969) [43]. However, this work does not mention an earlier one [73], where the conclusion is radically different, pointing out that the pore pattern of Psammornis is more like that of Aepyornithidae. However, the Psammornis subject should be addressed in more detail along with an increase in the amount of sampled material, since it has been erected on an incredibly small amount of remains, although this issue is outside the scope of the present paper. Recent works review and update the stratigraphy of Psammornis finds, their description and taxonomy [74,75].

However, the examination of several dozen fragments of Aepyornis maximus eggshells at the Museo de la Naturaleza y el Hombre (Santa Cruz de Tenerife) has made it possible to identify two pore-opening patterns. Figure 2A shows the above-mentioned pattern consisting of aligned slots homogeneously distributed and parallel oriented dagger-point or comma-like grooves (so-called Aepyornis type, morphotype A and aepyornithoid type in the specialized literature), and in Figure 2B, a very distinct pattern is patent that consists of round or oval sting pore openings, sometimes paired, which has previously been called Struthio type, morphotype S and struthioid type. We did not find any eggshell pieces containing both patterns. All the fragments of shells had similar thicknesses, regardless of their pore pattern. Since these two pore patterns are found in Aepyornis eggs, hereinafter we refer to them as aepyornithoid type 1, for the pore pattern of slit-like pore openings longitudinally oriented, and aepyornithoid type 2, for the pattern of round pore openings, not parallel to each other, although a certain orientation is appreciated.

3.1.1. Stable Isotopes of Eggshells and the Bird Diet in Northern Lanzarote

The carbon isotope composition of eggshell calcite reflects the composition of the bird diet modified from the sum of minor trophic and large metabolic effects [76]. Depending on the basal metabolism of birds and the status of these consumers within the trophic chain, the carbon isotope fractionation between eggshell calcite and bird diet (∆eggshell-diet) may vary between +8 and +16‰ [76,77]. Lazzerini et al. (2016) [78] analyzed three ratite eggshell fragments. Those eggshell fragments have a mean δ13Ccalc value of −10.3‰ ± 0.6 (V-PDB). According to Schaffner & Swart (1991) [76], this mean eggshell δ13Ccalc value indicates that these birds foraged in freshwater (non-marine) environments. Assuming a mean ∆eggshell-diet of +12 ± 4‰, the δ13C value of the avian diet was close to −22 ± 4‰, most likely reflecting a mixture of C3 and CAM-C4 plants either in their own diet if they were herbivorous or in the diet of their preys if they were carnivorous. Indeed, the mean δ 13C values of C3 and CAM-C4 plants in Lanzarote are −25.3‰ and −16.9‰, respectively, according to Yanes et al. (2008) [79]. At La Graciosa Islet, located next to the Famara massif, plant δ 13C values range from −29 to −13‰ [79]; their high variability is interpreted as reflecting extensive water stress. It is worth noting that these isotopic end-members for present-day Lanzarote plants bracket the calculated bird diet of −22 ± 4‰.

The oxygen isotope composition of eggshell calcite reflects the composition of water drunk by the females just before and during egg formation [80]. Lazzerini et al. (2016) [78] defined an isotopic fractionation equation for terrestrial birds:

δ18Ow = 1.077 (±0.063) δ18Ocalc − 34.607 (±1.554)

Thus, the avian eggshell fragments have a mean δ18Ocalc value of 29.3‰ ± 0.2 (V-SMOW). Using the previous equation, the δ18Ow of water drunk by these birds was −3.1 ± 2.4‰ (‰ V-SMOW). This oxygen isotope composition corresponds to equatorial or tropical precipitations (Figure 3) such as those documented today in the Canary Archipelago with δ18Ocalc values of −1.9 ± 1.1‰ (IAEA/WMO 2015) or in the present-day freshwater aquifer (δ18Ow = −2.7 ± 0.1‰) located at Fuente de Gusa [78].

Consequently, the carbon and oxygen isotope compositions of the three studied eggshell calcite fragments unearthed from Valle Grande strongly suggest that the eggs were laid by a terrestrial bird.

3.1.2. Ratite Eggshells from Northern Lanzarote

Classical studies of eggshells from the north of Lanzarote [3,51,70] led to the distinction of the two morphotypes, A and S, mentioned above. Subsequently, the Lanzarote eggshells were provisionally and without description attributed to the new oospecies Struthiolithus saueri [5]. Recently, Mikhailov & Zelenkov (2020) [72] found in Lanzarote the morphotype A and a new intermediate morphotype Aepyornithoide-Struthioide.

Several thousands of fragments of ratite eggshells have been collected mainly in three locations of the Famara massif, north of Lanzarote: Valle Grande 1, Valle Chico and Fuente de Gusa. Likewise, seven complete eggs, two external casts and one half-eggshell have been found. Table 1 (see also Supplementary Table S1) contains some dimensions of those eggs whose deformation by diagenetic processes prevents taking standard measurements. The six eggs measured have an almost perfect oval shape. Another egg is also oval, and the two external casts seem to have a fairly spherical shape. Table 1 shows the maximum lengths taken directly on the eggs. They are offered to illustrate the approximate sizes of each one. Since, except for one, the eggs have appeared more or less crushed by the sediment pressure, these measures are not reliable enough to calculate their respective volumes and weights. For obtaining these parameters, the length of the curve that passes over both apices (meridian length), and the length of the maximum curve perpendicular to the major axis (equatorial length) have been preferred. The calculated weights of the eggs from Lanzarote match those of the current North African ostrich (Supplementary Table S1). The calculated body mass of the females that would have laid three out the four eggs (Table 1) indicates that they reached quite large sizes if compared with current wild ostriches. Even the female that laid the egg 10 would have had a body mass comparable to that of a large current ostrich male bred in the wild [81,82,83]. The high values of egg volumes provided by the analysis program by means of computerized axial tomography are very striking. They have been included in Table 1, but the weights (masses) of the females have been calculated from the weights and volumes of the eggs calculated according to the measurements taken directly (see Materials and Methods) in order to compare with published data.

The thickness of the shells and the length of pore slots do not vary along the stratigraphic profile. The pore openings on the surface of eggshells produce two patterns, as already pointed out by Sauer (1972:22) [51]. One of them occurs in the vast majority of eggshell pieces and in all complete eggs and here has been called aepyornithoid type 1 (Figure 4A). In this type, the pore openings are grouped in long grooves, parallel to each other, oriented longitudinally and evenly distributed over the surface of the shell. Between these large grooves also occur pores that end singly on the surface and appear as small round dots and other groups in the form of a comma or small segment formed by the union of two or three pores. In some areas of the shells, these small and slightly parallel segments, as well as the round grouping of pores, are dominant. Even in some complete eggs, as in number 5 and the egg kept at the Museo de la Naturaleza y el Hombre of Tenerife, a pattern of short segments and rounded clusters is the only observed pattern. The type of pattern called aepyornithoid 2 shows a higher proportion of small, round pores, irregularly distributed on the surface of the shell (Figure 4B). Contrary to what has been reported in previous publications, no eggshells with a pore pattern similar to those of Struthio eggs have been found in Lanzarote. The error that was initially made in Sauer’s work [1,2,3,51], probably by confusing fragments eroded by friction during incubation, has persisted in later works, when the Lanzarote eggshells were attributed to Struthiolithus saueri [5] and when they were included in an intermediate A-S morphotype [72]. In Lanzarote eggs, as in those of the other ratites, the pore patterns remain unchanged along the surface (Supplementary Figure S1).

In radial sections, both morphotypes show the same three mineralized layers (Figure 4C). No thickness differences were found between the two types of eggshells. Type 1 is between 1.90 and 2.96 mm in thickness, with a mean of 2.36 mm (s.d. = 0.19; n = 230), while type 2 is between 2.01 and 2.49 mm (n = 6). Many fragments could not be measured due to their intense deterioration.

Postdepositional processes of chemical erosion [84] on the upper part of the fossil produce circular depressions of varying extent, generally a few millimeters in diameter, which are characterized by densities comprised between 30 and 80 per cm² (Figure 5A). On the lower part of the fossil, a smooth surface is produced. When the eggshell fragment comes from an incubated egg, it may have suffered more or less severe wear. The first thing to disappear on the outer surface are some pore furrows. If the erosion has been more intense, the thickness decreases, and only single, scattered pores are observed. When the chemical-made depressions have not yet reached a large size and are not very numerous, seen together with parallel pore grooves, they could be mistakenly taken as an intermediate pattern between Aepyornis and Struthio or even Struthio (see Supplementary Figure S1). The same could have happened with fragments eroded by friction, as noted above.

In order to exclude the possibility that the areas without pores or with very small pores characteristic of type 2 shells could in fact be type 1 eggshells albeit physically eroded, some fragments with type 1 pore pattern were subjected to mechanical abrasion. The typical slots of the aepyornithoid type 1 shell disappeared when the wear had eliminated the cuticular layer and part of the spongy one. In a shell fragment with an initial thickness of 2.6 mm, the pore pattern disappeared when its thickness was reduced to 2.0 mm. In another fragment of 2.5 mm in thickness, the slots of the pores disappeared when a thickness of 1.6 mm was reached. This prevents pieces assigned to the aepyornithoid 2 morphotype from actually being fragments of eroded eggshells of morphotype 1.

Along layer 3 of the stratigraphic profile of the Fuente de Gusa outcrop, several concentrations of eggshell fragments appear, located vertically, one above the other. These concentrations are clumps of pieces about 5 cm thick and between 1 and 2 m in length. They look like the remains of broken eggs of various clutches. The nests seem to be undisturbed or slightly disturbed as if they had been buried shortly after the hatching of the chicks.

Predation Evidence

Two percussion plugs dug up at levels 2 and 3 of VG 1 and one peck mark on an eggshell fragment from level 3 of VC constitute evidence of contemporary predation on the ratite eggs. A percussion plug is a rounded splinter that occurs occasionally as a result of the pressure of tooth on the eggshell. The punctual pressure exerted by the tooth can produce a hole in the shell, a peck mark, and a splinter is detached. These splinters or percussion plugs have a very open cone shape (Figure 5B,C). They are 8.0 and 10.3 in length, respectively. In lateral view, the layers of the eggshell are observed. Same predation marks are reported on one ornithoid egg from the lower Eocene of Colorado [71]. The predation marks from Lanzarote are the only evidence of a quite large-toothed predator (crocodile?) on the island, which possessed jaws large enough to prey on ostrich eggs.

Computed Tomography Scan

Among the eggs unearthed so far, number 5 (Figure 6, Supplementary Video S1, measurements in Table 1) is the one with the shell with the greatest integrity—albeit chemically altered—and whose content, apparently, is free from materials from the outer matrix. After processing the slides, it was observed that the egg was filled with a homogeneous sandy-like content, lacking any traces of embryo.

3.2. Ratite Eggshells Remarks

The record of material presented in the main text and tables is not exhaustive, and the comments do not constitute a taxonomic revision, which we consider should be carried out by describing ootaxa. For an expanded record, see Mikhailov & Zelenkov (2020) [72].

3.2.1. Eurosiberian Plate

Several sets of ratite eggshells were collected from 20 deposits of the Turkish locality of Ģandır. One of the deposits, site C5, yielded an incomplete pedal phalanx that was attributed to Struthio cf. brachydactylus [70]. This author reports to have found eggshell fragments at the same C5 site. Most of them possess a smooth outer surface, and few others have a furrowed one. Pieces of the first type were also associated with Struthio cf. brachydactylus [70]. Two photographs show the typical aepyornithoid 1 pore pits pattern. Sauer (1979) [70] indicated that the curvatures of these eggshell fragments point to an ovate shape of the original eggs, while the other type is not illustrated. Sauer (1979) [70] calculated the lengths of both types of eggs and assumed that the smooth-surface type, of similar size to the extant Struthio camelus, corresponds to Struthio cf. brachydactylus. However, the single pedal phalanx mentioned above, which was identified with uncertainty, constitutes a very small basis to describe the size range of the species that laid the eggs. For Sauer (1979) [70], the furrowed-surface fragments would correspond to eggs significantly larger in size than the other type. The description of the smoothed eggshells with furrows found in Ģandır matches aepyornithoid type 2. Sauer & Sauer (1978) [52], in their study of Ouarzazate shells, emphasized that the two types of eggshells found in Anatolia are the same as those found in Lanzarote and Ouarzazate. The present paper supports the occurrence of two eggshell types in the three localities.

Wang et al. (2011) [67] reported ratite eggshell fragments from two Inner Mongolia sites chronologically separated by more than 10 Ma, Gashunyinadege (about 17.5 Ma) and Baogedawula (about 7.1 Ma). Materials from both sites are considered of aepyornithoid-type by these authors (Supplementary Table S2). Description and figures of the outer surface of the pieces from Gashunyinadege show some paired and single pores, not clearly arranged in parallel. Additionally, the pores do not coalesce in short and slit-like grooves as in the Aepyornis eggshells from Madagascar. The fragments from Baogedawula have a somewhat different pattern. Here, the outer surface is not smooth as at Gashunyinadege but is covered by wide and winding furrows. The pore openings are round or oval shaped, and some occur paired. Thus, the material from both localities would be ranked in aepyornithoid type 2. The pore openings’ density is higher at Gashunyinadege. In particular, the eggshells from Baogedawula are more similar to the morphotype 2 found in Lanzarote.

Eggshells from Torrellano locality (Messinian; in the eastern part of the Iberian Peninsula) have been assigned to the Aepyornis type, with a thickness from 2.1 to 3.0 mm with an average of 2.6 mm (Supplementary Table S1). The shells show openings of rounded pores, sometimes isolated, sometimes grouped in pairs (Figure 7) and sometimes in lax aggregations of a few pores [55]. The authors of the present paper have had access to new fossil material from this site. In our sample of 18 pieces, between 1 and 5 cm in length, the thickness of the shells varies between 2.4 and 2.8 mm, with an average of 2.6 mm (Supplementary Table S2). The number of measurements was 32, with two measurements performed on each fragment, except on the smallest one, for which only one was taken. Furrows around 1.5 mm wide, which can reach more than 1 cm in length, are observed in some pieces. They result from the action of roots. The exterior surface of the shell pieces is sometimes abraded or partially covered by a carbonated crust, but the abrasion has not eliminated the outer or cuticular layer in any piece. It is possible to observe the arrangement of the pore openings in some of the fragments. Some pore openings form small, rounded depressions in the outer surface of the shell, from 0.1 to 0.2 mm in diameter. Most of them are not grouped, although there are some small aggregates. Some short parallel-oriented slots can also be observed, although they are not homogeneously distributed along the surface. When several openings end up close to each other, their depressions occur together and give rise to small shapes, such as commas and dashes. This pattern would correspond to the aepyornithoid type 2 found in Lanzarote.

A few pieces from La Gloria 4 (early Pliocene of Spain) were ascribed to the same aepyornithoid type described by Sauer (1972) [51] on eggshells from Madagascar [85]. The description of La Gloria items is not precise enough in the text, nor are there any figures in the paper; however, these pieces seem to be most likely referable to aepyornithoid type 1.

Little has been published on the upper Pliocene eggshells from Azerbaijan associated with the pelvis of Struthio transcaucasicus. Their thickness ranges from 2.5 to 2.9 mm. Additionally, “…the distribution and shape of small pores are different…” than in the current Struthio [56].

Numerous names have been given to large eggs found in Plio-Pleistocene sediments from Asia and Eastern Europe and assignable to the group of the current African ostriches. At Kisláng (Hungary, upper Pliocene to lower Pleistocene), some shell fragments received the name of Struthio pannonicus [41,86]. The shell thickness is 2.6 to 3.4 mm and averages 3.2 mm. Kretzoi (1955) [86] estimated the major and minor diameters as 220 and 180 mm. The ootaxon Struthiolithus chersonensis is the type-species of its genus. This species was erected on a complete fossil egg found in the Cherson district of Ukraine [49] (Supplementary Table S1). The author considered that the egg most likely belonged to the Struthio group, but in the description of the egg, there is no allusion to the shell pattern of pores, but only to its oval overall shape, although less marked than in Aepyornis, and to its major and minor axes of 18 cm and 15 cm in diameter, respectively (Supplementary Table S1). The shell thickness is 2.6 to 2.7 mm [34]. Bones and eggshells of great size, coming from several Chinese upper Pleistocene sites, have received the denomination of Struthio anderssoni [34,39,53,87] (Supplementary Table S1). The corresponding shell thickness ranges from 2.1 to 2.3 mm, with an average of 2.2 mm. Struthio wimani is probably found in upper Miocene sediments, likewise at several Chinese localities (Supplementary Table S1). Its average thickness is 2.7 mm. The eggs assigned to Struthio mongolicus, from the lower Pliocene of Mongolia, with a thickness of 1.9 to 2.35 mm, would be the smallest, with sizes similar to the current Struthio camelus eggs.

Ignoring the morphological variability of eggshells of extinct ratite species, it is not possible to establish biunivocal relationships between zoospecies—based on skeletal elements on the one hand—and oospecies–based on eggshells on the other hand—even if eggshells are undoubtedly associated with fossil bones, since several biological avian species can produce similar eggshells.

Skeletal Record

As mentioned above, an incomplete pedal phalanx attributed to Struthio cf. brachydactylus was documented at the site C5 of the locality of Ģandır. This bone clearly corresponds to a ratite possessing a foot shorter than in the extant Struthio camelus [70]. In several locations around the Caucasus mountain range, the finding of eggshells together with Struthio fossil bones has been reported. A high number of Neogene localities in Europe and Asia bearing struthionid bones have been discovered. More than 30 species and subspecies of Struthio have been named, most of the Neogene forms coming from the norther peri-Pontic region [88]. However, the systematics of these taxa are very confusing. Although the osteological record encompasses numerous localities and a broad span of time (Supplementary Table S3), the inflation of its taxonomic nomenclature partly results from the scarcity of fossil remains, which makes the assessment of reliable ranges of morphological and size variations impossible, thus precluding any reliable comparisons. Boev & Spassov (2009) [88] summarized the previous attempts to untangle the taxonomic nomenclature of the Eurasian struthionids. Recently, Zelenkov et al. (2019) [89] have highlighted morphological differences in a new finding of a femur (Taurida Cave, Crimea, early Pleistocene) and in some fossil bones of very large ostriches previously assigned to Struthio. Most of the localities from which those fossils come are found around the Black Sea, from Pliocene to early Pleistocene in age. The prior subgenus Pachystruthio, erected by Kretzoi (1954) [41] to classify a phalanx from Kisláng (Hungary) as Struthio (Pachystruthio) pannonicus, is elevated to genus and encompasses Pachystruthio pannonicus (Kisláng, early Pleistocene) [41], P. dmanisensis (Dmanisi, Georgia, and Taurida, Crimea, early Pleistocene) [90] and P. transcaucasicus (Kvabebi, Georgia, late Pliocene). Zelenkov et al. (2019) [89] also tentatively assign findings from other sites to Pachystruthio: Çalta [91], Samos [92], Odessa catacombs [93], Livenstovka [94], the egg attributed to Struthiolithus chersonensis [36] and the eggshell fragments from Çandir [70]. Buffetaut and Angst (2021) [87] have subsequently attributed a femur of large size to Pachystruthio sp. (Nihewan formation, China, early Pleistocene).

Pachystruthio transcaucasicus, described on an incomplete pelvis, was associated with very thick (2.9–3.3 mm) eggshell fragments of aepyornithoid type collected in different localities of similar age from Azerbaijan [56,59]. The eggshells associated with P. pannonicus have thicknesses between 2.6 and 3.4 mm [41].

The most recent studies highlight the presence of struthionids in this region. Struthio sp. has been found in Çalta, early Pliocene of Anatolia, of similar size to Struthio asiaticus and intermediate between Struthio dmanisensis and the extant Struthio camelus [91] and two incomplete bones from the late Miocene of Bulgaria, assigned to Struthio cf. karatheodoris, showing that the corresponding animal had a smaller size than Struthio novorossicus and was also a foot more robust than the current ostrich [88].

3.2.2. Indian Plate

An interesting aspect of the record of ratite eggshells is the geographic distribution of the items over the Indian plate. More than forty Pleistocene localities, the Anjar site among them, are scattered along the Indian Peninsula, some of them including paleolithic implements. The eggshells show a pore pattern consisting of the presence of numerous small circular depressions where several pores end. The thickness range is similar to that of the current Struthio molybdophanes [63,95]. Findings from the Siwalik region, on the border between the Eurosiberian and Indian plates, deserve careful attention. They correspond to two of the basic morphotypes, struthioid and aepyornithoid. Sauer (1972) [51] studied several eggshell fragments collected in 1935 near Hasnot, Upper Siwalik, from the Dhok Pathan Fm., within a horizon attributed to the Pliocene, although they are probably Miocene in age. These fragments had originally been identified as Struthio (?). Sauer (1972) [51] assigned these specimens to the aepyornithoid type. The description of the pieces and the corresponding figure offer no doubts on such an assignation. Several eggshell pieces from Dharamsala, middle Siwalik, coming probably from a single egg, were assigned to the ootaxon Struthiolithus sp. [61]. The site was palaeomagnetically dated to 10.1 Ma. The outer surface of the eggshells reveals tiny pores (0.01–0.1 mm in diameter) as needle-points, sparsely distributed. However, pores also occur as aggregate complexes in very shallow depressions [61]. The Dharamsala eggshell thickness averages around 3 mm. The pore depression observed in Figure 5 of Patnaik et al. (2009) [61] is approximately 9 mm in diameter. These values of thickness and pore depressions far exceed the values obtained in the current subspecies of Struthio camelus [46,51]. In fact, the pore complexes are wider than in Struthio daberasensis and Struthio karingarabensis [96,97]. Both parameters fall within the variation range of some oospecies of Diamantornis [96], while the pore depression is much smoother. Thus, the concerned specimens are “ostrich-like eggshells,” as indicated by Patnaik et al. (2009) [61], and more particularly have a pore pattern similar to those of Struthio molybdophanes, S. camelus massaicus and S. c. australis.

As a result of a study on the carbon isotope record in the Siwalik Himalayan foothills, Stern et al. (1994) [98] collected ratite eggshells attributed to the aepyornithoid and struthioid morphotypes from about ten locations in the north of India and Pakistan, both morphotypes coinciding in some of the sites. Aepyornithoid type occurred in strata ranging in age from 11.3 to 1.25 Ma. This type showed elongate slits where pore pits unite, which is the most notorious feature of what in the present paper is called aepyornithoid type 1. Pieces regarded by such authors as of struthioid morphotype came from strata of 2.2 to 0.6 Ma. The outer surface of these eggshells had “circular surface depressions… containing many pores that coalesce at depth.” [98]. This description fits again with the pore pattern found in the eggs laid by Struthio molybdophanes, S. c. massaicus and S. c. australis. Stern et al. (1994) [98] provided a single thickness measurement of one piece of aepyornithoid type (2.84 mm) as well as one of struthioid type (2.30 mm). It may be inferred that in the Indian plate, there were ratites that laid eggs with a pore pattern similar to those of the current subspecies Struthios molybdophanes, S. c. massaicus and S. c. australis from about ten million years up to the latest Pleistocene. Other ratite populations laid pore pattern eggs similar to that of Aepyornis with a distribution close to the border between the Indian and the Eurosiberian plates, from about eleven million years ago to the early Pleistocene. Both types of pore patterns co-occur in this region during the early Pleistocene and, probably, at some point during the early Miocene.

Skeletal Record

Milne-Edwards (1869–1871) [99] erected the species Struthio asiaticus on some ostrich fossil bones coming from unspecified Siwalik localities (Supplementary Table S3). From Sauer (1968) [34] onwards, it has been assumed that these skeletal elements were associated with eggshells found in the Siwaliks as well, although later stratigraphic studies indicated a time lapse of more than 6 Ma between these horizons [100]. Therefore, no clear association can be established between bones and eggshells coming from the Siwaliks [63].

3.2.3. African and Arabian Plates

Ratite eggshells attributed to the oldest age have been uncovered from the Tsondab Sandstone Formation, in the Namib Desert and other areas in Namibia and South Africa [65,96,97]. In these fossil sites, different forms of eggshells have been described as genera and species within the zoological systematics (Supplementary Table S2). Eggshells with a pore pattern of the previously known aepyornithoid type appear in the lowest stratigraphic layers, roughly attributed to the early Miocene (20 to 16 Ma) on the basis of its mammal record [65]. This paper [65] (Figure 3) supports the likelihood of such an assignation. In fact, looking at the morphology of the pore-openings aggregates, the eggshell fragments would be assigned to the aepyornithoid type 1, although their thickness is extraordinarily reduced, around 1.2 mm. Bones of Struthio coppensi are allegedly associated with these eggshells, but its association degree (i.e., same layer, close layers of same site, correlated layers from different sites, etc.) is not indicated. Overlying these layers in the regional stratigraphic profile, eggshells of Namornis oshanai, Diamantornis corbetti, D. spaggiarii, D. wardi, D. laini, Struthio karingarabensis, S. daberasensis and the extant Struthio camelus occur. A progressive decrease in the eggshell thickness in this series of taxa is noticed. Additionally, with the exception of Namornis, which has pore aggregations of a shape not observed elsewhere, all the above-mentioned Namibian eggshell taxa have uniformly distributed rounded pore clusters, ranging from the largest sizes in the oldest forms to the smallest in the most modern ones, ending with the pattern of the extant Struthio camelus camelus [65,96,97]. Pickford (2014) [101] has described some new taxa of eggshells from the Tsondab sequence. Eggshells coming from the lowest levels, all of them belonging to aepyornithoid type 1, have been classified within two taxa, Tsondabornis minor, which is characterized by thinner shells, and Tsondabornis psammoides, having a somewhat thicker shell. The biostratigraphic ages of both taxa are ca. 21 and 18 Ma, respectively. Tsondabornis diagnosis matches aepyornithoid type 1. In the same publication, the Namornis oshanai eggshells with thicknesses comprised between 2.5 and 3.7 mm (Supplementary Table S1) are classified within the new species Namornis elimensis [101].

Wadi Moghra, an early Miocene (c. 17 Ma) fossil site located in northern Egypt, has yielded one small eggshell fragment, which was identified as aepyornithoid-type [102]. The openings are rounded and simple and are not arranged in parallel. The thickness of the piece is 2.1 mm, and it presents a very high density of small or simple pores. In the figure of the eggshell fragment, it is clear that some of the pores are paired. Thus, it bears a remarkable similarity to the second pattern that we have identified among the Aepyornis shells, the aepyornithoid type 2.

Eggshells from Lothagam, Kenya, revealed the occurrence of two types of ratite pore patterns in some geological layers. The eggshells from the lower member of the Nawata formation (7.4–6.5 Ma) were first referred as to Aepyornithidae (tentatively) and cf. Struthio sp. [57]. The description of the aepyornithoid pieces indicates the presence of point and sting pores and linear grooves, even though the corresponding figure does not document such linear grooves. Thus, the pattern is of type aepyornithoid 2. In a review by Stidham (2004) [66], the remains of cf. Struthio are considered to belong to Diamantornis wardi. These eggshells were later assigned to Diamantornis cf. laini [58]. In the Upper Nawata member (6.5–5.0 Ma) of Lothagam, Harris & Leakey (2003) [57] only recognized cf. Struthio sp. Nevertheless, Stidham (2004) [66] identified the taxa Struthio daberasensis and Diamantornis laini on the same material. For Harrison & Msuya (2005) [58], these fossils belonged to Struthio cf. karingarabensis and Diamantornis sp. Likewise, the eggshells from the Apak member of the Nachukui formation were attributed to cf. Struthio sp. [57]. Stidham (2004) [66] assigned these eggshells to Struthio daberasensis while Harrison & Msuya (2005) [58] considered them related to Struthio cf. karingarabensis.

Bibi et al. (2006) [54] attributed some eggshell fragments from several localities of the Baynunah formation to Diamantornis laini and to aepyornithoid-type. The Baynunah formation, in the United Arab Emirates, is biostratigraphically constrained from 8 to 6 Ma [103]. Both the photograph and the description given by Bibi et al. (2006) [54] of the pieces they assigned to the aepyornithoid-type fit with the pore opening patterns observed in Aepyornis. Some of the pieces show aligned linear grooves, where pores coalesce (aepyornithoid type 1), while other fragments present oval and round pores (aepyornithoid type 2). The thickness of the 25 shells measured by the authors (both aepyornithoid types) varies between 1.65 and 2.29 mm, with an average of 1.93 mm. Pickford et al. (2023) [104] studied 17 small eggshell fragments from the site 25 RAK, in Rub Al Khali, Marsawdad Fm., Oman. The specimens range in thickness from 2.0 to 3.0 mm, and all but one are badly abraded. The two eggshell fragments that did not suffer such intense surface erosion have been ascribed to Diamantornis laini [104], but the small outer surface exhibited by such small fragments and the range of thicknesses (Supplementary Table S1) are compatible with both Diamantornis laini and Struthio karingarabensis. The remaining specimens could not be attributed to any taxonomic group.

Sauer & Sauer (1978) [52] found two types of eggshell fragments at various sites located in the Moroccan district of Ouarzazate that they assigned to aepyornithoid, struthious and intermediate aepyornithoid-struthious types, such types being supposedly of the Mio-Pliocene age. However, the age of these eggshells is doubtful, as it is only based on the geological chart of the area. The thickness range is 1.40 to 2.70 mm, although the lowest values include heavily abraded fragments. The so-called struthious pattern of Sauer & Sauer (1978) [52] is concentrated in the lower half of the thickness range. These authors mention the occurrence of the classical aepyornithoid morphotype, where “the openings of the pore canals are arranged parallel to one another” [52] and fragments attributed to struthious and intermediate morphotypes, exhibiting “…circular needle-point and needle-prick pores irregularly distributed, singly and occasionally in rows” [52]. However, the previous pore pattern is not observed in the eggshells of the current Struthio camelus, where the pore openings are uniformly distributed over its surface. According to the descriptions, the two types of eggshells previously referred to as aepyornithoid type 1 and aepyornithoid type 2 occur at Ouarzazate, while none of the extant ostrich patterns were found.

In the Moroccan deposit of Ahl al Oughlam, some eggshells attributed to struthioid type were found along with several skeletal remains attributed to Struthio asiaticus, although a recent study does not support identifications of Struthio asiaticus outside India [105]. Such eggshell fragments have similarities to those of the current subspecies Struthio molybdophanes [60]. The pore openings in Ahl al Oughlam eggshells form round clusters of 0.5 to 1 mm in diameter, with a density of 10 to 20 pores per cm². The shell thickness ranges from 2.3 to 2.7 mm, which are values comparable to Struthio daberasensis. In the Namibian material on which this ichnospecies was described, the diameter of the pore complexes ranges from 0.5 to 2.2 mm, and the thickness ranges from 1.7 to 2.6 mm [95]. The eggshells of Struthio daberasensis from Mwimbi North (Malawi) have pore complex diameters from 0.5 to 1.5 mm and have a thickness range from 2.0 to 2.7 mm [66]. Nevertheless, the pore groups of Struthio daberasensis converge at rounded pits, not at smooth depressions such as in Struthio molybdophanes and in the eggshell from Ahl al Oughlam.

Skeletal Record

Struthio coppensi, a species smaller than the extant ostrich, was described by Mourer-Chauviré et al. (1996) [106] on hind limbs found in the lower Miocene of Namibia. These findings were associated with eggshells of aepyornithoid type [65,106]. Two distal ends of didactyl tarsometatarsi, similar in size to Struthio camelus, were found in Bled ed Douarah (late Miocene, Tunisia) and labeled as Struthio sp. [107]. Some hind limb bones from several sites of western Kenya, dated from the middle Miocene (between 14 and 12.5 Ma), belong to an indeterminate form of Struthio, which was of intermediate size between Struthio coppensi (the smallest known ostrich) and the extant subspecies Struthio camelus camelus and the same size as Struthio orlovi, from the late Miocene of Moldova [108] (Supplementary Table S3). In the same Baynunah Formation of the United Arab Emirates where Diamantornis laini and aepyornithoid eggshells were found, an almost complete pelvis has appeared, attributed to Struthio cf. karatheodoris [109].

3.3. Other Faunal Records from Northern Lanzarote

3.3.1. Terrestrial Gastropods

All the fossil land gastropods unearthed belong to three species: Zootecus insularis, Leptaxis orzolae and Theba orzolae. The specimens show different states of preservation: some of them are well preserved, others appear burned by overlying lava flows or hydrothermal fluids, others are significantly deformed and others are simply internal fossil molds. Among them, some specimens of Zootecus insularis are completely flattened along their anterior-posterior plane, which suggests a loading of lava flows or stones. This species lives in muddy soils, over rocky and vegetated escarpments or in lands aside small ravines and on gravel flood plains and terraces [110]. L. orzolae and T. orzolae are restricted to Pliocene deposits from the north of Lanzarote, so confident ecological information is not available. However, from the geological formation of the deposits studied here, we can deduce the affinity of these two species with sandy soils. Species of the same genus occur currently in Atlantic islands on various kinds of vegetation.

3.3.2. Tortoise Eggshells

The paleontological sites of Famara massif have yielded the oldest testudinid eggshells of the archipelago. Level 3 of Valle Grande 1 [32], the oldest one, records almost complete eggs [20]. Previously, complete eggs had also been recovered from Fuente de Gusa [111]. Even if the tortoise egg material from Lanzarote (Figure 8) is similar to giant and semi-giant extant taxa from Africa, they cannot be referred to Centrochelys or Stigmochelys on the basis of their egg sizes alone. More detailed histological studies including tortoise eggshells from other Canarian localities and extant taxa from Africa and Europe are required to clarify the taxonomy of Lanzarote testudinids. Thus, the eggshells from Famara massif are identified as Testudinidae indet.

3.3.3. Boid-like Vertebra

A new snake vertebra has been recovered in recent years at the Valle Chico site (Figure 9). It is a medium-sized medial-posterior truncal vertebra (centrum length, CL = 6 mm). The specimen has a weathered surface. Its zygosphene, zygapophysis and neural spine are broken, and its condyle and paradiapophyses are eroded.

Even though it is deteriorated, this vertebra shows a marked “boid-like” pattern: vertebra massively built, with a short and wide centrum (relation centrum length (CL)/width of the interzygapophyseal constriction (WIC) = 0.95), neural spine and hemal keel (but not hypapophysis) present and paradiapophysis weakly divided into para- and diapophyseal portions [112,113,114].

In anterior view, the neural canal is reduced and subtriangular, with a base wider than high; the zygosphene, partially broken, is moderately thick, and the cotyle is large and slightly dorso-ventrally flattened. A shallow fossa is present between the cotyle and the prezygapophyses, but no paracotylar foramina are visible. In dorsal view, the vertebra is short and wide, and the interzygapophyseal constriction is weakly expressed (however, this could be an artefact related to the partial fragmentation of the postzygapophyses). The neural spine is broken; however, it seems to start just behind the zygosphene. Posteriorly, the median notch of the neural arch is not deep. In ventral view, the centrum is short and widens anteriorly, with the subcentral ridges not strongly marked. The hemal keel is large and well-marked off from the centrum and with lateral margins slightly medially concave. The ventral surface of the centrum is flat, and the subcentral foramina are present on each side of the hemal keel. In lateral view, the vertebra had to be higher than long; the paradiapophyses, badly eroded, are protruding, with a subrectagular outline, and weakly divided into para- and diapophyses. In posterior view, the neural arch is weakly vaulted, and the postzygapopyses are slightly inclined above the horizontal. The condyle is partially eroded.

This new vertebra shows a very similar morphology to that described by Barahona et al. (1998) [6] and could therefore belong to the same taxon. Although slightly larger, the observed morphological differences in this new vertebra (i.e., narrower neural canal, flatter neural arch and prominent hemal kell) may be related to the more posterior position.

Compared to the current “boid-like” taxa present in continental Africa (Python, Eryx and Calabaria), the Lanzarote vertebra has an original set of characters that prevents a precise attribution. As already indicated by Barahona et al. (1998) [6], the presence of a weakly vaulted neural arch belongs to the Eryx (and we can also add Calabaria). A large hemal keel is also present in Eryx, while in Calabaria, it is narrower. However, the fossil vertebra is larger than those of Calabaria or the current and fossil known erycines, the zygosphene is thicker and the interzygapophyseal constriction is less marked. For these characters, the fossil vertebra is reminiscent of Python. Among the known fossil taxa, the above-mentioned set of characters shows similarities with some members of the Bavarioboa genus—an European Boinae known from the Middle Oligocene to the Middle Miocene [113,114,115]—and thus a relationship between the Lanzarote fossil and Bavarioboa should not be ruled out. However, no Bavarioboa fossils were mentioned in Africa, and its extinction in Europe occurred well before (Middle Miocene, MN5/MN16, around 15 Ma [114]) the estimated date for the arrival of the first terrestrial vertebrates in Lanzarote (Upper Miocene, around 10Ma).

The paradiapophyses and the condyle of the new vertebra are eroded with strongly spongious bone exposition. Additionally, a loss of bone material is observed on the posterior wings of the neural arch, just above the zygantrum place (an oval perforation on the right place and a deep notch on the left place). This alteration pattern is reminiscent of that observed on vertebrae that have undergone chemical dissolution by the digestive juices of a predator.

3.3.4. Neognath Eggshells

A group of neognath eggshell fragments occur in layer 3 of Valle Chico (Figure 10). They come from at least three eggs, located next to each other, and they probably correspond to a single clutch. They constitute the sole record of neognaths in this outcrops complex of northern Lanzarote. None of the eggs are complete, but the size and contour are well visible. They are oval in shape, not conical at all, including their upper apex. The lengths of major and minor axes are approximately 30.5 and 20.5 mm (Figure 9). Their shape and size are compatible with eggs of the European Storm-petrel, Hydrobates pelagicus. Procellariiform eggs occur frequently in eolianites from the Pleistocene of Lanzarote and Fuerteventura [116,117].

4. Discussion

The thickness of Ratitae eggshells is quite variable, and this does not seem to be closely related to the various pore patterns. The fossil record suggests that when a pore pattern has reached vast geographical and chronological distributions, eggshells with varying thicknesses are found, which is evident in the case of aepyornithoid morphotypes (Aepyornis, aepyornithoid type, Tsondabornis) (Supplementary Table S1). In the Namibian record, different pore patterns exhibit the same rate of variation of the eggshell thickness over time. Thus, the Namornis pattern shows the same thickness variation as Diamantornis and aepyornithoids. It is apparent that eggshell thickness should not be discarded as an expression of plastic adaptation to environmental conditions, disregarding the particular pore pattern morphotype that the eggshell bears.

The finding of eggshells with aepyornithid-like pore patterns has been considered by some authors [54,69] as a counterargument, which means an indication of the presence of ratites of the elephant bird lineage outside of Madagascar. The osteological record seems to indicate that the ostriches that inhabited Eurasia and Africa during the Neogene were didactylous ratites of the family Struthionidae (Supplementary Table S3). The absence outside of Madagascar of skeletal elements attributable to the aepyornithid lineage may indicate that the separation between Africa and Madagascar was prior to the arrival of the first ratites in Madagascar. The attribution of aepyornithoid eggshells, found in Lanzarote and other paleontological localities, to the Struthio type, to the intermediate aepyornithoid-struthiode type or to the oogenus Struthiolithus, has from the beginning blurred the understanding of the geographical and chronological distribution of ratites in Africa and Eurasia. It has been tacitly assumed that the Aepyronithidae were a group endemic to Madagascar and that, consequently, African and Eurasian ratites could only be referred to the present-day Struthio lineage. This premise may not be correct.

Two reports devoted to the oldest finds of eggshells of aepyontithid type are doubtful because some issues concerning such identifications have not been sufficiently documented. One is the paleontological site of Tashgain Bel, in Mongolia, attributed with doubts to the Oligocene [54], and the other is the outcrop of Elisabethfeld, in Namibia, the type-locality of Struthio coppensi, with an age of 20 Ma estimated by biostratigraphic methods [95]. However, these geological ages are doubtful in both cases. Also in both cases, the occurrence of aepyornithoid-type eggshells is mentioned, but the eggshells are neither described nor figured. In the paper on Struthio coppensi, it is indicated that the fossil bones are associated with eggshells [106], but it is not clearly stated whether or not this association occurs in the same layer, the same locality or even the same geological formation.

The oldest finds of ratite eggshells occur in Africa and Asia, but it is in southern Africa, mainly in Namibia, where the greatest diversity of pore patterns is documented. These old findings on both continents correspond to the aepyornithoid morphotypes. The pore patterns here called aepyornithoid types 1 and 2 are registered with certainty from the early Miocene of Egypt and China, both aged around 17 Ma [67,102], to the lower Pliocene of Lanzarote and the Holocene of Madagascar [118], where both patterns disappeared with the vanishing of the Canarian ratites and the Aepyornis extinction, respectively (Table 1). Thus, these two aepyornithoid morphotypes are shared by both the late Holocene elephant birds from Madagascar, birds exhibiting a tridactyl foot [119], a less derived character state than the characteristic didactyly of African and Eurasian ostriches, and by the eggshells from the Famara massif, which implies that the aepyornithoid morphotypes precede the split between Struthionidae and Aepyornithidae. Although we do not have a record of bones and eggshells of Aepyornis before the Holocene, the Miocene is the terminus ante quem for the Aepyornis lineage to have been separated from the African ratite stock.

The pore patterns of the taxa Struthio karingarabensis, S. kakesiensis and S. daberasensis, in this order, are viewed by some specialists as transitional forms between Diamantornis corbetti, D. spaggiaii, D. wardi and D. laini and the extant Struthio camelus [58,65,96]. In this supposed sequence of transformations, the clusters of pore openings diminish in size and, consequently, increase their density on the egg surface. It is plausible that this chronological sequence of forms (species, oospecies or morphotypes) corresponds to an image of a real transformation over time of some features of the eggshells within one ratite lineage. However, this succession of forms, contrarily, may be a construct, that is, not a real succession of forms, but an expression of different taxa adapting to changing environmental conditions. It cannot be excluded that it could be the result of an adaptive response of more than one ratite lineage to a climatic fluctuation. The absence of skeletal remains prevents the rejection of either hypothesis. In addition, the fact that several of these morphotypes have been classified within the genus Struthio without being associated with skeletal elements, instead of receiving morphotypic names or being described according to parataxonomic rules, creates prejudices and hinders the discussion.

In the Siwalik region, on the northern edge of the Indian plate, eggshells with a pore-opening pattern similar to the current one of Struthio molybdophanes, camelus massaicus and c. australis is recorded from 10.1 Ma [61] until the middle Pleistocene [98]. In such a region, the aepyornithoid 1 morphotype is documented at least between 11.3 and 1.25 Ma [98]. However, only the morphotype of S. molybdophanes-c. massaicus-c. australis has been found during the Pleistocene inside the Indian Plate [63].

The microstructure of pore-opening arrangements is similar in the eggshells from Ahl al Oughlam (early Pleistocene, Morocco) and Anjar (late Pleistocene, India). In both eggshells, the pore openings are small and mostly rounded and are grouped in circular or oval shallow pits. The pores are branched and have an average diameter of 1 mm. Every depression or pit contains clusters of around 50 pores. The shells have an average thickness of 2.54 mm at Ahl al Oughlam and 2.2 mm at Anjar [60,63].

The ratites whose eggshells have been found on Lanzarote lived during the lower Pliocene on a relatively small island, one third the size of present-day Lanzarote, which today constitutes the north of the island. The occurrence of two eggshell morphotypes of pore pattern is an argument against the fact that both morphotypes correspond to two different ostrich species, since it is difficult to admit that such a small island could have yielded food resources and water to two populations of single large herbivorous or omnivorous avian species. One viable population of a single species maintained over time would already constitute a great challenge. Thus, it seems more likely that a single species of ratite would lay eggs with two pore patterns, instead of both patterns appearing together in some eggs, the few cases that mention in the literature the latter condition being doubtful. In the extensive collection of Lanzarote eggshells, no single item has been seen displaying the two pore patterns. Thus, the different configurations of both pore opening configurations do not discriminate zoological species of ratites from Lanzarote. Apparently, they could be a consequence of genetic variations having different proportions in populations through time and space. That is, in Lanzarote, there would be a single polymorphic species regarding the pore pattern of the eggshells.

When comparing to samples from the wild North African ostrich, the dimensions of the whole eggs and the corresponding body masses of Lanzarote females point to the highest values of the size range of the extant Struthio camelus (Table 1, Supplementary Table S1).

The concentrations of eggshell fragments in the Fuente de Gusa site are most likely evidence of a ratites nesting area. The extant African ostrich usually nests in the same place, and their nests are formed with eggs of around six females owing to their polygynous family structure. The small dimensions of the nests found in Lanzarote would be a result of the laying of only one or two females.

One peck mark and two percussion plugs on eggshell fragments are the only evidence of the existence of an egg-eating predator. Considering that the size range of the eggs found in Famara was similar to that of extant ostriches and that the thickness of the shell was quite greater, the size of the jaws and the power this predator could exert were considerable. So far, no other data on the existence of such a predator is available.

Arrival Modes

The date marking the beginning of subaerial volcanic emissions in the massif of Famara is 10.2 Ma [30], and consequently such a date could be admitted as the terminus post quem for the arrival of land animals to the island. However, there is no fossil record of terrestrial animals close to this date. The 4.3 Ma dating for the beginning of the faunal record that we report here was obtained by the current team from basalts at the present sea level [32]. It is possible that before such a date, there was not enough space for sustaining the life of these animals. There is also no later record of any terrestrial vertebrate until the late Pleistocene, which, in turn, could mean that the conditions for the continuity of populations of these animals were impossible after 3.8 Ma.

Although it is easily accepted that various animals can be dragged to the islands by natural phenomena, such as river flash flooding, tropical storms, tsunamis, etc., even if they survived their time adrift, it could be considered unlikely that punctual events would yield a high enough flow of individuals to get viable populations over time. Anyhow, the case of Lanzarote shows that several species of land vertebrates arrived in sufficient numbers to form viable populations and, likewise, found food resources to sustain themselves for at least several millennia. There is a consensus that at least 50 individuals are necessary to prevent fatal inbreeding, and a minimum of 500 individuals is needed to minimize genetic drift over generations [120,121]. More recently, Reed & Bryant (2000) [122] have considered that these minimal threshold values have been underestimated and that the minimal number to avoid lethal inbreeding is closer to 100 individuals. If the arrival of large land birds to a volcanic island, which has not been linked to the continent, is unusual, the conditions for these animals to constitute viable populations involved many punctual rafting events over a very short time (century scale).

By contrast, Zootecus insularis is currently distributed from the Atlantic coast of Sahara to Egypt, Arabia and India. Even though it was cited in Cape Verde Archipelago as a subspecies [123], Z. insularis subdiaphanus, it differs from the Canary Islands fossil specimens, the latter being more similar to the form of North Africa [124]. It is worth highlighting the absence of this species from the current fauna of the Canary Islands and other Atlantic archipelagos. This fact suggests easier communication between Lanzarote and the coast of Africa during the early Pliocene or previously than today.

There are two basic ways for any dispersal system to explain the arrival of the precursors of Lanzarote’s first terrestrial fauna from Africa, by land and by sea. The land route was proposed by the discoverers of the ratite eggshells [1,2,3,34,51,70]. Based on Dietz & Sproll (1970) [125]—results that have not been confirmed since—Sauer & Rothe (1972) [3] and Sauer (1972) [51] suggest that the eastern Canary Islands (Fuerteventura and Lanzarote) originated from a continental fragment broken off from the African margin and that this process could have lasted long enough for these animals to establish themselves in Lanzarote. In addition, in the paper by Sauer & Rothe (1972) [3], the hypothesis of a bridge made of volcanoes is raised. That bridge would allow an east-to-west stepping stone model of dispersal between Africa and the eastern Canaries. Both explanation proposals were developed from a biased attention to geological studies and an incomplete analysis of the lithology of Valle Grande 1 and Valle Chico sites. Thus, the depth of the seabed in the area between the Canaries and Africa is 1300 m, and there are no remains of submerged volcanoes. On the other hand, the oceanic crust in this region is very stable, so the distance between the Canary Islands and the continent should not have changed significantly in the past.

One aspect to consider is the seemingly small number of species that were established on then-existing Famara island in an early stage of colonization. If there had been a land connection, or some intermediate islands used to progress by leaps, it would be expected to result in a richer fauna by the presence of other animals, such as mammals, as it happened in the Mediterranean islands during the Neogene. Another interesting point in this issue is the uniqueness of this faunal settlement. The same kinds of animals (snakes, ostriches, turtles), probably belonging to different species but with similar adaptations, have not crossed the Canary channel again despite being present since then, for millions of years, on the African coast, in front of the Canary Islands. Therefore, we can expect unique conditions that would result in river overflows or strong surface currents flowing from the mainland coast to the ocean and likewise, coinciding in time to the existence of a regime of marine currents that favored a rafting dispersal of some terrestrial vertebrates. A combination of particular climatic conditions on the African continent together with adequate marine currents as a prerequisite for the arrival of terrestrial vertebrates has been documented in Madagascar [126].

A possibility that seems remote at first but that should be explored is that what arrived in Famara were eggs, not adult ratites. That is, large rivers on the opposite African coast (Draa, Sus) overflowed at that time, dragging away ratite nests, whose eggs, after several days in the sea, ended up reaching the coast of the island of Famara. However, there are a number of facts that oppose this possibility. If the eggs had arrived passively by sea, they would have arrived along with marine elements. That is, it would be expected to find marine mollusks, fish bones or coral fragments, at least in the lower levels of the stratigraphic column. However, there is no trace of marine elements in any of the fossil beds. On the contrary, no sea turtle eggs have been found, and likewise all the strata are dominated by the presence of land snails and insect cells. All this points to the fact that the formation of the deposits took place in terrestrial environments, not even intertidal. Then, there is a group of vicissitudes that the eggs and juveniles would have to overcome. In the event that the eggs had floated in and that the embryos had not died from hypothermia, they would also have completed the embryonic cycle without incubation. Afterwards, the juveniles should have been able to survive without parental care until reproductive age. The few known cases of viable bird eggs after immersion in water refer to species that nest next to water and immersion times that do not exceed two days [127].

The only possibility for the animals to arrive at the Pliocene proto-Lanzarote (Famara and Los Ajaches islands) was to cross the Canary Channel by sea. However, two scenarios open up, which could be called southward and northward. The southward scenario consists of directly overcoming the 150 km distance between Africa and the then island of Famara by one or several single journeys. The northward scenario is a case of island-hopping (or stepping-stone) colonization mode, with a first big leap towards Fuerteventura, located 100 km west from mainland. So, those animals first reached this island, older than the proto-Lanzarote islands and closer to the continent. Then the animals moved northward, step by step, to the islands corresponding to the present-day Ajaches and Famara massifs. However, both journeys involve following courses between NNW and WNW, that is, against the Canary current, which runs with a NNE-SSW predominant direction, at speeds between 0.5 and 2 marine knots. Its eastern boundary along the northwest African coast consists of a surface current almost permanently directed southward, with velocities of about 35 cm/s [128].

One possibility within the first scenario that should not be ruled out is that the animals that arrived in Famara were dragged by the current of the Canary Islands. For instance, large river floods caused by heavy rains in the Atlas could drag animals from Sus and Draa valleys, which discharge their waters in front of the eastern Canaries. Passively, and with today’s velocity of the Canary current, the animals could arrive in four days. A rafting dispersal mode from the north for animals such as snakes and tortoises seems plausible. Swimming could have been likely for a “python,” though impossible for an Erycine snake because they are very bad swimmers. Conversely, extant ostrich, emu and rhea are good swimmers. Given that the snake vertebra has traces of digestion, its introduction by a bird of prey should not be ruled out.

If one or the other of the scenarios is confirmed, it would have very different implications and consequences. The underlying lava flow to the Famara outcrops is dated 4.3 Ma and is at the current sea level. The fossil deposits correspond to the first evidence of sand accumulations and soils that were formed on the island. Thus, it seems that ostriches occupied Famara Island as soon as suitable conditions existed. In any case, the southward scenario means that arriving animals would bear the same characters as their African counterparts since they went directly from the mainland. The northward scenario seems more plausible, as it allows the crossing of the Canary channel prior to the formation of Famara Island because the formation of Fuerteventura and Los Ajaches islands had concluded previously, in the middle Miocene. Since the ellipsoidal-shaped eggs (Figure 6 and Supplementary Video S1) with aepyornithoid pore patterns recorded in Lanzarote apparently had already disappeared from Africa in the late Pliocene (Supplementary Table S2), and since these ostriches had settled on Famara Island not long after suitable living conditions for the existence of land vertebrates, these birds may have arrived from the neighboring islands in the south, from larger populations that would have been previously established in Fuerteventura. This scenario also eludes the problem of the maintenance over time of one viable ostrich population on a relatively small island, as Famara must have been when it was a single island. A main population of ostriches in Fuerteventura would have constituted the irradiation core of the individuals that moved swimming towards the nearby smaller islands of Los Ajaches and Famara. Thus, the northward scenario seems to be the most inclusive. Recently, paleontological surveys have been started in Fuerteventura to search for deposits with evidence of ratites.

It can be inferred with some certainty that the circumstances that caused the faunal movements from Africa to the Canaries before or during the Lower Pliocene ceased. The insular forms to which they hypothetically gave rise became extinguished, and similar species did not cross the Canary channel again.