Devonian trilobites from the Amazonas Basin: systematic diversity, taphonomy, and biogeographic significance Trilobitas

Although the Devonian invertebrate biota of the Amazonas Basin is incompletely known, its unusual mixture of supposedly Malvinokaffric and Appalachian taxa is of considerable biogeographical interest. The history of research on Devonian trilobites from the Amazonas Basin is summarized. These fossils occur in the Maecuru (latest Emsian – early Eifelian) and Ererê (latest Eifelian – early Givetian) formations, representing the families Homalonotidae, Dalmanitidae, and Calmoniidae. Trilobite diversity is higher in the Maecuru (nine named plus six indeterminate taxa) than the Ererê Formation (with only two named taxa), which displays faunal impoverishment generally. None of the Devonian Amazonas Basin trilobite species is known from other Malvinokaffric sites, suggesting very high endemicity within this basin. A taxonomic revision of “Homalonotus” derbyi (Clarke, 1890) is presented and a lectotype is designated. A lectotype and paralectotype are also designated for Eldredgeia paituna (Hartt & Rathbun, 1875). Some taphonomic observations of fossils from the Maecuru Formation are made, and a brief biogeographic synopsis of the Devonian Amazonas biota is also given.


INTRODUCTION
Of all the Paleozoic sedimentary basins in Brazil, the remoteness of Amazonas Basin, coupled with the paucity of exposures, has imposed severe limitations upon research on its Devonian fossils. Despite this, past investigations nevertheless suggest that the Amazonas Basin was extremely important biogeographically during the Devonian, as a transitional "biogeographic boundary mixing area" (Boucot & Racheboeuf, 1993) characterized by the co-occurrence of Eastern Americas brachiopods and Malvinokaffric trilobites. Unfortunately, the literature dealing with these fossils (especially trilobites) is widely scattered. In an attempt to rectify this situation, the present paper reviews the history of research on, and the systematic diversity of Devonian trilobites from the Amazonas Basin. Additionally, taphonomic observations are provided in order to improve understanding of the depositional environments in which these fossils occur, especially regarding the Maecuru Formation. Finally, a brief biogeographic synopsis of the Amazonas Devonian biota is presented, with emphasis on similarities and differences from Andean and Appalachian biotas.
The first Devonian trilobites described from the Amazonas Basin were Dalmania paituna Hartt & Rathbun (1875) and Homalonotus oiara Hartt & Rathbun (1875), collected during the Morgan Expeditions, in 1870 and 1871, around the village of Ererê, in the State of Pará, Brazil. At that time, the Ererê sandstone was considered to be of Middle Devonian age and was correlated with the Hamilton Group of New York. Modern palynological data indicate that the Ererê Formation is latest Eifelian to early Givetian (Grahn & Melo, 2004).
Another expedition, undertaken by the "Comissão Geológica do Império do Brazil" (1876), carried out a further examination of the Ererê district and resulted in the discovery of richly fossiliferous Devonian deposits along the Maecuru and Curuá rivers.
Marine invertebrate macrofossils occur in the upper part of the Maecuru Formation [(sensu stricto after Derby (1878) and Cunha et al. (2007)] and in the lower level of the Ererê Formation. Clarke (1890) described the trilobites from Maecuru and Ererê formations. For many years afterward, no taxonomic revision was made for the Amazonian trilobites and subsequent papers used only Clarke's original designations. Clarke (1913) coined the generic name Phacopina with the type species Phacops braziliensis Clarke, 1890. Other supposed Phacopina species from North America included by him, are no longer considered to belong to this genus (Eldredge & Branisa, 1980, p. 240). Eldredge & Ormiston (1979) presented a "Master List" with all the known genera and subgenera of Malvinokaffric trilobites, and discussed their phylogenetic affinity, including species from the Devonian of the Amazonas Basin. Tomczykowa (1975), Copper (1977), Melo (1988), Lieberman (1993), Lieberman et al. (1991), Sandford (2005) and Rustan & Vaccari (2012) also discussed Devonian Amazonian trilobites in their works. Carvalho & Fonseca (2007) reviewed the taxon "Dalmanites" maecurua and erected the new genus Amazonaspis.

MATERIAL AND METHODS
Specimens studied and figured herein are housed in the paleoinvertebrates collection of the Museu Nacional/UFRJ (MN) and Departamento Nacional de Produção Mineral (DNPM/RJ), Rio de Janeiro, and in the New York State Museum (NYSM), New York. They are from the Maecuru and Ererê formations, State of Pará (for locality map, see Carvalho & Fonseca, 2007). The trilobites are represented by internal and external molds of disarticulated, mostly incomplete and poorly preserved specimens. Morphological terminology follows Whittington & Kelly (1997) and Eldredge & Branisa (1980 figure 4), is designated here as the lectotype. MN 3371-I, Clarke, 1890, plate I, figure 7, small incomplete internal mold of cephalon, here designated the paralectotype.
Diagnosis: cephalon subtriangular in outline; glabella subtrapezoidal, convex (sag. and tr.), with sides weakly concave, indistinct lobation, glabellar's antero-lateral angles broadly rounded and postero-lateral angles acute; preglabellar field produced conspicuously, well below the level of the glabella. Rostral suture slightly concave. Palpebral and postocular fixigena areas bulged, depressed abruptly in front of them. Anterior branch of facial suture slightly sigmoidal. Eyes quite small, situated at the summit of the palpebral lobe, away from axial glabellar furrow. Paraglabellar area distinct. Occiptal furrow well developed, narrow, nearly transverse, curving slightly forward at the axial line. Posterior border furrow wide and very shallow. Occiptal ring moderately broad and convex (sag. and tr.), above the level of the glabella.
Remarks: the available material is represented by an external mold of an almost complete cephalon, described originally by Clarke (1890) as Homalonotus derbyi. He suggested that this species was a representative of the "subdivision Trimerus". Subsequent authors (e.g. Katzer (1897); Clarke (1913); Kozlowski (1923) followed Clarke, but Cooper (1982) considered that H. derbyi may be a species of Trimerus. Tomczykowa (1975) and Sandford (2005) tentatively referred the Amazonian species to Digonus. In reality, the allocation of this species to one or another genus of Homalonotinae is somewhat problematic, because of the incompleteness and paucity of material. Nevertheless, some cephalic characters are shared with other species of Digonus.
The specimen figured by Clarke (1890, plate I, figure  19, MN 3372-I) is not a homalonotid. It may instead belong to the family Dalmanitidae ( Figure 2C). The incompleteness of the material does not allow more accurate taxonomic placement and it is here regarded as gen. and sp. indet.
Genus Burmeisteria Salter, 1865 Type material: internal mold of a fragment of a cephalon. According to Clarke (1890), the material collected by the Morgan expeditions of 1870 and 1871 "is in the museum of the Cornell University, Ithaca" but apparently it was transferred to the New York State Museum (NYSM 4494).
Remarks: based on Clarke's description (1890), the glabella is subrectangular with a strong incurvature of the lateral margins (= urceolate) and smooth non-lobate surface, which are all characteristics of Burmeisteria. Type material: small internal and external molds of a single glabella. The material is lost and for this reason we could not provide the catalogue number of the MN (RJ) were it was originally deposited.
Remarks: according to Eldredge & Ormiston (1979, p. 163) this sp. could be considered a Dalmanitidae sensu lato, or perhaps a true dalmanitid. We did not see the original material and left the species as Dalmanitidae gen. indet.
Remarks: for complete description, as well as a diagnosis, see Carvalho & Fonseca (2007), who designated the lectotype and paralectotypes.
Remarks: Lieberman (1993) included the Amazonian species "paituna" to the new genus Eldredgeia. Eldredgeia paituna differs from E. venustus in the greater convexity (sag.) of the glabella and the greater dorsoventral expansion of the frontal lobe (see Lieberman, 1993, p. 553-554). The genus Eldredgeia is known from Amazonas and Parnaíba basins and Bolivia, but not from other Malvinokaffric areas.
Diagnosis: cephalon small, wider than long (sag.) with depressed surface. Glabella subpentagonal in outline with flattened surface; frontal glabellar lobe subrhomboidal in outline, sloping abruptly, almost vertically, to the antero-lateral margin. Glabellar furrows S3 and S2 are shallow and wider; S3 slightly convex, S2 small and weakly straight; S1 is reduced to a deep adaxial apodemal pit. S0 wide and moderately deep. L0 narrow, in profile is above of the glabella.
The genus Malvinella is known from the Amazonas Basin and Bolivia, but not from other Malvinokaffric areas.
Remarks: we did not make a revision of this material and have left it in open nomenclature. This material will be studied in the future.
Remarks: Lieberman et al. (1991, p. 835) informally referred this Amazonian taxon to "Palpebrops" and designated the lectotype and paralectotype. Eldredge & Ormiston (1979) placed this species within the genus Malvinella but, according to Lieberman et al. (1991), it is phylogenetically closer to Palpebrops and Vogesina than to other members of the "Malvinella group", although it could not be assigned confidently either to Palpebrops or Vogesina.
Genus Phacopina (Clarke, 1890) Eldredge & Branisa (1980, p. 165) made a revision of this taxon but did not designated lectotype or paralectotype. These will be provide in the future work. Phacopina is another genus that is known so far only from the Amazonas Basin and Bolivia.
Remarks: definitely this taxon is not a homalonotid. The specimen is represented by an internal mold of pygidium, the morphology of which suggests affinity with the Metacryphaeus Group. However, we do not assign it to a particular genus, and leave it in open nomenclature as a Calmoniidae gen. indet.
Type material: syntypes (lost), however, there is a cast of the original material in the collection of NYSM 4323 (figured by Lieberman et al., 1991, figure 8-5, 6).
Remarks: according to Lieberman et al. (1991), this species is phylogenetically closest to Vogesina, but was left in open nomenclature. This is yet another example of genus that occurs only in the Amazonas Basin and Bolivia.
Remarks: Eldredge & Ormiston (1979) included this taxon within the Metacryphaeus Group and suggested that it could be a new genus, or aff. Malvinella, along with Dalmanites gemellus Clarke, 1890 (see above), but Lieberman et al. (1991) did not comment on its affinity.
The elongated shape of this pygidium is not typical of the genus Phacops, in which the pygidium is usually short (micropygous). It is conceivable that this taxon represents a new genus. A revision of this material is required.
Remarks: the presence of small pygidial lappets definitely excludes this taxon from Phacops, but suggests calmoniid affinity.
Remarks: the identification of these pygidia is problematic and they are left as uncertain family.
From the list of the original and current nomenclature for the Devonian trilobite species from the Amazonas Basin, it is evident that diversity at genus level is far greater them was originally suspected (growing from six to eleven named genera, plus six additional indeterminate taxa) (Table 1).

TAPHONOMIC CONSIDERATIONS
In the Amazonas Basin, the taphonomic analysis of the macrofossils from Maecuru and Ererê formations is limited. First of all, the fossiliferous sandstones were not in situ when they were collected by the Petrobras expedition in 1986 (they were already moved away from the original site), while previous expeditions collected only small samples that are even more inadequate for taphonomic analysis.
Ponciano & Machado (2007) made some taphonomic observations of the Maecuru Formation and observed that many brachiopods and bivalves still had their valves articulated, but were associated with disarticulated specimens, and lacked preferred orientation. They also observed many bivalve and brachiopod shell edges without apparent modifications, suggesting possibly a low grade of transportation, although it was sufficient to result in disarticulation of trilobite exoskeletons (poorly preserved as incomplete cephalon and pygidium molds). Disarticulation of trilobites also may reveal the disturbance of the exuvium after molting (inference corroborated by the absence of in situ librigenae and hypostomes, because these sutures open during exuviation). Other delicate fossil invertebrates, as bryozoans, corals and conulariids present low to intermediate degrees of fragmentation, suggesting low energy in the marine depositional environment. According to Caputo (1984), the Maecuru Formation represents a major river-dominated fan-delta system progradation interrupted by a fast, short-lived transgression followed by another fan-delta system progradation. Much like the sedimentological studies (Caputo, 1984), the combination of taphonomic signatures also corroborate the influence of episodic events as storms or floods on the genesis of the Maecuru fossil assemblages, suggesting that the fossils are parautochthonous and were deposited in a shallow marine environment.

PALEOBIOGEOGRAPHIC CONSIDERATIONS
The Devonian trilobites of the Amazonas Basin are important paleobiogeographically. Calmoniids were apparently endemic to the Malvinokaffric Realm, a region in the southern hemisphere characterized by cold water, siliciclastic beds and an endemic fauna. Besides calmoniids, two other more cosmopolitan groups (homalonotids and dalmanitids) also occur, but these are represented by species that were apparently restricted to the Malvinokaffric Realm.  (Clarke, 1890) Dalmanites tumiloba Clarke, 1890 Malvinella tumiloba (Clarke, 1890) Phacops goeldii Katzer, 1903 Palpebrops? goeldi (Katzer, 1903) Phacops braziliensis Clarke, 1890 Phacopina braziliensis (Clarke, 1890) Dalmanites Nevertheless, Amazonaspis maecurua shows affinity with dalmanitids from Appalachia (Carvalho & Fonseca, 2007), and the homalonotid Digonus is known from the Old World. These trilobites may have migrated into the Amazonas Basin during a widespread Eifelian transgression from the northeast. Eldredge & Ormiston (1979) divided the Malvinokaffric Realm into three Provinces, based on trilobite distribution patterns: an Andean Province, including Peru, Bolivia, Paraguay, and Argentina; a Brazilian Province, including Brazil and possibly Uruguay; and a South African-Malvinan Province, possibly including Antarctica.
The calmoniid species from the Amazonas Basin differ from those from the Paraná Basin, with no genera or species shared by these two basins. However, all the calmoniid genera recorded from the Devonian of the Amazonas Basin are also present in the Devonian of Bolivia, although they are represented by different species. As Lieberman et al. (1991) noted, representatives of the "Malvinella group" from the Maecuru Formation are most closely related to Andean taxa, suggesting that a marine connection probably existed between the Amazonas Basin and Bolivia or southern Peru during the Middle Devonian. Boucot (1988) proposed two major Devonian Units in the northern hemisphere, based on the distribution of brachiopods: the Old Word Realm and the Eastern Americas Realm. He divided the latter into four sub-provinces: Nevadan, Appohimchi, Colombian and Amazon. He also suggested that Malvinokaffric brachiopods were more closely related to early Devonian brachiopods of the Appohimchi Subprovince of the Eastern Americas Realm (except for Australocoelia). Curiously, this pattern of relationships is not observed among the trilobites in the Amazonas Basin.
The association of Malvinokaffric trilobites with Eastern Americas brachiopods in the Amazonas Basin suggests that these taxa had somewhat different physiological or environmental tolerances that rarely coincided except in this region (a "biogeographic boundary mixing area"; Boucot & Racheboeuf, 1993). This concept is also supported by paleogeographical reconstructions for the late Emsian and late Eifelian/Givetian in Brazil (Melo, 1988), as well as by reconstructions of Pangea in the late Early Devonian and Middle Devonian (Boucot, 1988).

CONCLUSIONS
Despite the paucity of Devonian fossils from the Amazonas Basin, increased systematic diversity among its trilobites has been recognized over the past century, particularly in the Maecuru Formation and especially at genus level (eleven named genera, plus six additional indeterminate taxa). This increase has resulted only from revisionary studies of existing collections, rather than from new collecting, suggesting that the true diversity of Amazonas trilobites may be seriously underrepresented and that many new forms could be revealed with renewed efforts to collect and describe fossils from this intriguing region, especially from the Maecuru Formation.
Moreover, because the Devonian biota of the Amazonas Basin seems to include biogeographically mixed taxa of Malvinokaffric trilobites and Appalachian brachiopods, renewed collecting is imperative in order to elucidate and further expand our understanding of this still largely unexplained situation.
We have attempted here to review the known occurrences of trilobites in the Maecuru and Ererê formations, to examine some taphonomic aspects of Maecuru fossils, and to summarize the biogeographical conundrum revealed by them. The point of this exercise was to provide a snapshot of current understanding rather than to offer new interpretations, and to emphasize the urgent need for further research (especially new collecting) in the Devonian of the Amazonas Basin. We are confident that such investigations will yield important new data that will greatly clarify the contemporaneous biogeographical relationships of the Malvinokaffric Realm to other regions of the world.