The late Middle Devonian fauna of Red Hill I, Nevada, and its paleobiogeographic implications

The fauna of the Middle Devonian Red Hill I locality, Nevada, is unusual in the cooccurrence of a rich fish assemblage with a rich invertebrate one. Sponges are second in abundance of specimens and number of species only to the fishes and occur together with other invertebrates (conodonts, conulariids, dacryoconarid tentaculites, gastropods, bivalves, brachiopods, arthropods, and unidentifiable ammonoids and echinoderms). The invertebrates indicate a marine depositional paleoenvironment. The conodonts indicate a placement within the lower disparalis Zone, late Givetian. The fish assemblage is dominated by the antiarch Asterolepis. All the other fishes, acanthodians, actinopterygians and sarcopterygians, are less common. The closest biogeographic relationship of the fish fauna is with the Middle/Late Devonian fish fauna of the Baltic Region, followed by that of eastern Canada (Miguasha), Scotland and Iran. This distribution corresponds to the Devonian Euramerica faunal province with connection to eastern Gondwana (Iran and Australia). Localities with the same genera as Red Hill I are interpreted as marine with the exception of the Scottish localities. Asterolepis is the most widely distributed vertebrate genus, mostly marine, but it may be able to enter freshwater like Eusthenopteron if one accepts a freshwater depositional paleoenvironment for the Scottish localities.

stones are intercalated within the finely bedded sequence. Two bone beds occur in layer 14. The layers were numbered from the top down as the excavation progressed. On the right side of the section (Fig. 1), the symbols show the distribution of the fauna. Here we have the unusual situation that a rich fish fauna occurs together with a rich marine invertebrate fauna.
In the literature, different biostratigraphic ages are given for the fish-bearing section, and this has historical reasons. A consensus on the position of the Middle/ Late Devonian boundary was reached in the 1980s. Therefore the placement of the Red Hill I fish beds into the earliest Late Devonian (Johnson 1977: Leiorhynchus hippocastanea Zone [brachiopod interval 25] corresponding to earliest Polygnathus asymmetricus Zone [conodont]) corresponds to the placement of the boundary below the P. asymmetricus Zone. Gregory et al. (1977) and Reed (1985Reed ( , 1986) supported the early Late Devonian age by comparison with Baltic fish occurrences in the Gauja and Amata beds. With the placement of the Middle/Late Devonian boundary above the earliest P. asymmetricus Zone, the fish occurrences were moved into the latest Givetian (Johnson et al. 1988). Murphy et al. (1976) had already given a late Middle Devonian age (upper Schmidtognathus hermanni-Polygnathus cristatus Zone). The conodonts occurring with the fish indicate a placement within the early Klapperina disparalis Zone, latest Givetian (Johnson Figure 1. Section of the fish horizons in the thin-bedded lime mudstones alternating with thick-bedded limestones of the Red Hill Beds, Middle Devonian at Red Hill I, Nevada (Middle/Upper Devonian boundary above the section). Symbols from left to right: plants, invertebrates (sponges, conulariids, dacryoconarids, gastropods, bivalves, ammonoids, lingulids, articulate brachiopods, echinoderms) and fishes. 1990, fig. 52; Elliott & Johnson 1997;Elliott et al. 2000), which corresponds to the present day placement of the Middle/Late Devonian boundary.

Fossil assemblage
The fossil assemblage is unusual among rich fossil fish localities as also being rich in sponges, conulariids, articulate brachiopods, and echinoderms. Plants, gastropods, ammonoids, inarticulate brachiopods and echinoderms are so poorly preserved that they are unidentifiable. Sponges are the most common fossils next to the fishes; the sponge fauna is diverse with 16 different forms. Sponges and conulariids, like most invertebrates and fishes, are flattened; they are preserved, at least in part, with hematite. The change to hematite probably occurred during a widespread hydrothermal event within the region during the Tertiary (Johnson 1960;McKee 1996;Tingley & LaPointe 1997); that event is responsible for the red color of the sediments. The locality lies within the Battle Mountain -Eureka Trend (French et al. 1996, fig. 3) northeast of the Tonkin Springs mine. Ore deposits and hydrothermal changes of sediments are connected with the intensive fault systems. The age for the hydrothermal event is given as after Miocene basaltic andesites (16.9 AE 0.2 Ma) for the Buckhorn district (Benedict 1996) northwest of Red Hill I or after the late Oligocene magmatic event (24-25 Ma) for the Gold Bar district in the Roberts Mountains (French et al. 1996) southeast of Red Hill I or even earlier (36-38 Ma) after Ressel (2005;Cellura personal comm.).
The co-occurrence of abundant sponges and fishes is most unusual. J. K. Rigby visited the quarry during the 1987 excavation by the Kansas University crew and later described many new sponge genera and species together with D. Mehl (Rigby & Mehl 1994). The articulate brachiopods were described by Johnson (1978), and the conulariid was described by Babcock & Feldman (1986a).

Fish assemblage
The most common fish in the Red Hill I Lagerståtte (concentration Lagerståtte in the sense of Seilacher 1970) is the antiarch placoderm Asterolepis (Murphy et al. 1976), which occurs disarticulated throughout the section (J. W. Reed described single bones in his Ph.D. thesis 1979). Sometimes complete heads and/or armor, and separate tails of Asterolepis occur; one complete specimen with tail ( Fig. 4) has been discovered. Os-teichthyans occur frequently; in descending order of abundance they include lungfish (Reed 1985), rhipidistians and actinopterygians (Reed 1992;Arratia & Cloutier 2004). Arthrodires, onychodonts, and acanthodians (Reed 1986) are rare.
The Red Hill I fish fauna (Gregory et al. 1977) is diverse. Asterolepis, the most common fish in the Red Hill I Lagerståtte, is the most widely distributed fish genus of the fauna (Schultze & Cloutier 1996). Upeniece (2002) showed a frequent occurrence of Asterolepis in Givetian to lower Frasnian localities of the Baltic region, NW-Russia, Timan and Scotland. She also listed occurrences in Belgium, Greenland, Spitsbergen, Iran and the eastern USA. In addition, Denison (1978) cited Asterolepis radiata Rohon, 1900 from Kazakhstan and Asterolepis sp. from Australia; Young (1993) added one occurrence from the lower Frasnian of New South Wales, Australia; and Janvier & Villarroel (2000) described one from Colombia, South America. Such a worldwide distribution indicates a marine form, which is congruent with the interpretation of the depositional environment of most occurrences (Upeniece 2002) except those in Scotland, Greenland, Spitsbergen and New York State, which are interpreted as freshwater deposits by most authors.

Placodermi
Asterolepis sp. (levels 3, 5-21): one complete specimen (Fig. 4), heads, trunks, tails, and mainly single plates (Reed 1979) arthrodire (  The new tristichopterid (Fig. 6) is quite closely related to Eusthenopteron, if it is placed in the phylogenetic diagram of Ahlberg & Johanson (1997). Shape of lacrimal and jugal, presence of only one supraorbital and the longest preorbital region compared to the skull roof length of all tristichopterids are unique features of the new form. The new tristichopterid has enlarged premaxillary and anterior dentary fangs, a feature typical for advanced (above Eusthenopteron) tristichopterids (manuscript in preparation).

Taphonomy
Besides the rare co-occurrence of abundant fish and conulariids, the co-occurrence of abundant fish and glass sponges is exceptional (Rigby & Mehl 1994; e.g. beds 10-12). Fishes, conulariids, and sponges show similar preservation in the Red Hill I Lagerståtte. All have been flattened; it is unusual that all conulariids from a locality are crushed to this extent. Crushing in some conulariid specimens evidently occurred follow-  ing initial exoskeletal collapse because some specimens show older, postmortem infolding of the relatively weak exoskeleton in the apertural region. These taphonomically produced structures have been called apertural lappets in the earlier literature (Babcock 1993;Babcock et al. 1987).
Fishes, sponges and conulariids are preserved in various states of disarticulation. Isolated parts of fishes are most common, but sometimes nearly complete specimens can be found (Fig. 4). In addition, fish bone beds have been found in level 14. The sponges are generally fragmented (Rigby & Mehl 1994). However, the presence of complete specimens (Fig. 2) and the presence of seemingly undisturbed thin bedding, together suggests that benthic scavengers and sediment bioturbators are largely excluded from that environment. Rapid and deep burial, dysaerobic conditions, and perhaps hypersaline conditions, are examples of situations that would tend to exclude the scavengers and therefore be conducive to their preservation as fossils. Most conulariids are relatively complete, articulated exoskeletons ( Fig. 3) but disarticulated exoskeletal pieces, including individual rods, do occur. The relatively large number of articulated, but not fully complete conulariid exoskeletons contrast with mostly disarticulated specimens of fishes and sponges. The fact that tiny exoskeletal pieces are preserved adjacent to articulated specimens of conulariids and fishes, instead of being washed away, suggests that current agitation and bioturbation of the sediment surface was quite limited except in the bonebeds of level 14. By analogy with the occurrence of disarticulated conulariids in a Mississippian black shale from Ohio (Babcock & Feldmann 1986b), it is inferred that some Red Hill I conulariids were lying on the sediment surface long enough for some disarticulation to occur without interference by currents, benthic scavengers, or sediment bioturbators. Scavengers of conulariids and fishes as well as burrowing organisms, may have been largely excluded from the site of deposition.
There is no indication of disturbance of the specimens after deposition. Thus one has to conclude that the specimens arrived as pieces (disarticulation through decay while floating) or complete at the depositional site. Marginal disturbance has to be explained by crushing or last forms of decay.

Paleoecology
The invertebrates indicate a marine depositional paleoenvironment. The presence of conulariids in the Red Hill I fauna yields limited paleoecological information because these animals seem to have been rather ubiquitous in many Paleozoic marine habitats (Babcock & Feldmann 1986b). Conulariids are known only from marine strata. Paraconularia recurvatus may have been an opportunistic species, as judged by its abundance (conulariids occur usually single or as few specimens), and by the lack of other conulariid species in the fauna. Conulariids as a general rule were evidently capable of tolerating conditions ranging from well-oxygenated to poorly oxygenated (Babcock 1991), and were not restricted by water depth, at least not in shelf seas. Murphy (1977) considered the paleoenvironment of the Northern Simpson Park Range as open marine at the time (also see Rigby & Mehl 1994); Johnson et al. (1988) placed the sedimentation of the limestones on the outer shelf. The trace fossils, which indicate a near shore paleoenvironment, occur only in one horizon above the section with the fish-sponge-conulariid assemblage (Fig. 1).
According to Denison (1978, p. 108), antiarchs are predominantly freshwater stream and lake dwellers, though a few (Gerdalepis, Lepadolepis and Grossaspis) became adapted to marine life. Asterolepis is common in coastal marine sediments of the Middle and Upper Devonian of the Baltic region (Gauja beds, Burtniecki, Upper Narova, Arukçla, Armata beds) and in sediments interpreted as freshwater deposits of Scotland (Nairn sandstone, Upper Ronsay series and Brindister flags) and Greenland (Tab. 1). Asterolepis juveniles occur in coastal marine waters (Lode, Latvia; Upeniece & Upenieks 1992), and by comparison with extant (or Pennsylvanian of the North American mid-continent) marine fish behavior (Gunter 1961) this is a further indication that Asterolepis lived in marine environments. The  (Campbell & Barwick 1988). The actinopterygian Cheirolepis appears to have been adapted to a marine environment, except for the occurrences in Scotland, which are interpreted as "freshwater" deposits (see Schultze 1999, p. 380). The occurrence of Soederberghia at Red Hill I is the only clear marine occurrence of the genus, even though Ahlberg et al. (2001) suggested the habitat of the genus as non-marine to shallow near-shore marine. The figured tristichopterid (there seem to be two different tristichopterids at Red Hill I) is very similar to Eusthenopteron and has been identified as Eusthenopteron by Reed (1979) and by Elliot & Johnson (1997) in biostratigraphic comparisons and by Schultze & Cloutier (1996) in paleoecological comparisons. In Table 1, the Red Hill I tristichopterid is recorded together with other tristichopterids to discuss the paleogeographic distribution of the group (see below). The fishes at Red Hill I occur in a definitive marine environment as indicated by the invertebrate fauna. There is no indication that the fishes have been washed in; the bone beds are reworked zones closer to shore. Laterally the Red Hill I beds continue in limestones of marine origin. That habitat indicates marine adaptation of these fishes; their occurrences in putative freshwater deposits require an explanation if one accepts the sedimentological interpretations of the Scottish, Greenland and Australian localities. Either different species of these genera had different adaptations, or they could enter freshwater without too much difficulties as some marine fishes can today (in contrast: freshwater fishes except catadromous species cannot enter marine environments).

Comparison with other faunas
All the sponges (Rigby & Mehl 1994) despite their diversity are endemic to the Red Hill I Lagerståtte at species and in cases generic level. They are the last (Cyathophycus, Protospongia) or the first (Actinodictya, Teganiella) known occurrences of the genus. Cyathophycella is restricted during its whole temporal occurrences to the Roberts Mountains/Northern Simpson Park Range area of Nevada. Actinodictya is the only genus with species of similar age (Frasnian) in New York State. This connection to the eastern United States is supported by the occurrence of Dictyospongia in the Late Devonian rocks of New York State.
"Only one conulariid species, Paraconularia recurvatus Babcock & Feldman, 1986a, is represented, and it is endemic to the Red Hill I locality. P. recurvatus is the oldest species of the genus known from North America, and shows a morphology consistent with its inferred phylogenetic position as a form intermediate between Early to Middle Devonian species of Conularia and Mississippian species of Paraconularia. The close spac-ing of rods is characteristic of a species in the presumed ancestral genus Conularia, whereas strongly reduced spines and, to a lesser extent, the recurvature of the rods, is characteristic of Paraconularia species. The only other known Paraconularia from the Devonian of North America are P. chagrinensis Babcock & Feldmann, 1986a, from the Upper Devonian of Ohio; P. wellsvillia Babcock & Feldmann, 1986a, from the Upper Devonian rocks of New York; and an undescribed species from the Devonian-Mississippian transition beds of South Dakota (Babcock & Feldmann 1986a). Worldwide, only two described species of Paraconularia, P. africana (Sharpe, 1856), and P. ulrichana (Clarke, 1913), are older than P. recurvatus. Both species are from Lower to Middle Devonian strata of the Malvinokaffric Realm in South America and southern Africa (Babcock 1993;Babcock et al. 1987). The genus Paraconularia has an Euramerican -Gondwanan distribution." (Babcock).
The fishes at Red Hill I show a worldwide distribution on a generic level (Tab. 1) as can be expected from marine fishes. Asterolepis has its widest distribution around the Middle/Late Devonian boundary, as shown by Young (1981Young ( , 1993 and Janvier & Villarroel (2000). The distribution of Asterolepis coincides with the occurrences of the widely distributed antiarch Bothriolepis, for which Young (2003) suggested a dispersal through the sea. The acanthodian Machaeracanthus was widely distributed in the Early and early Middle Devonian. The actinopterygian Cheirolepis seems to be limited to the northern continents (Laurentia and Baltica at the time), whereas the dipnoans Griphognathus and Soederberghia, and the tristichopterid rhipidistians were cosmopolitan. Young (1981Young ( , 1993Young ( , 2003 based migration pattern mainly on the "freshwater" occurrences of groups (xenacanth sharks, phyllolepid placoderms, large rhizodont crossopterygians and early tetrapod amphibians), which are not present in the Red Hill I fauna. He favors a migration of these groups from eastern Gondwana westwards. In contrast, I favor here marine occurrences of the Red Hill I taxa and like to argue that the marine adaptation of these genera (Tab. 1) gave them the freedom to move between areas. Asterolepis is known from Laurussia and northern Gondwana (South America); records from Australia (Denison 1978) are doubtful. The actinopterygian Cheirolepis is widely distributed in Laurussia, but not recorded from Gondwana. The lungfish Griphognathus is known from Laurussia and Australia from the earliest Frasnian; the Red Hill I occurrence is somewhat older as may be the occurrence in Bergisch Gladbach, Germany. Soederberghia co-occurs with early tetrapods in Australia, North America and Greenland. The occurrence in Red Hill I is the earliest and marine record of the taxon. These early occurrences of the two lungfish in Laurussia and later occurrences in Australia contradict Young's proposed direction of migration. A special group of interest are the tristichopterids. Ahlberg & Johanson (1997) argued that all primitive tristichopterids occur in Laurussia, a migration to Australia is postulated. Young (2008) countered that conclusion with establishing a new subfamily, Mandageriidae, for the advanced Australian tristichopterids. That excludes that endemic group from a migration pattern (Clement et al. 2009 suggested biogeographical barriers). The tristichopterid from Red Hill I is as old as the earliest Australian form and younger than the earliest Eusthenopteron records. In conclusion, it seems that there is no unidirectional migration, but an exchange between Laurussia and Gondwana (Young 2008).
The Red Hill I Lagerståtte was included in the comparison of the Escuminac Formation ichthyofauna with other late Givetian/early Frasnian ichthyofaunas by Schultze & Cloutier (1996), even though the Red Hill I ichthyofauna was not fully prepared and described at that time. The Red Hill I ichthyofauna appeared close to Russian and Baltic region ichthyofaunas on a generic level, and close to Scottish ichthyofaunas on family level in that analysis. The close connection to European localities is no problem for a marine fauna that could migrate along the continental margins. Red Hill I clusters with marine localities worldwide: Iran, Germany, Australia, northwest Russia, the Baltic region, Antarctica, and France. Gogo, Australia (Campbell & Barwick 1999), and Bergisch-Gladbach, Germany (Jux 1963;Schultze 1969;Jessen 1993), are marine localities of Griphognathus, and Koknese, Latvia (Gross 1956;Kuršs 1992;Lukševičs 2001), is a near-shore marine locality with Griphognathus. In contrast, the widely distributed localities of Soederberghia are interpreted as non-marine (Ahlberg et al. 2001) with the exception of Red Hill I, Nevada. Schultze & Cloutier (1996) evaluated which fishes determined the clusters of localities and which indicated the depositional environments of the localities. The analysis resulted in a big cluster (assemblages II-V) of marine forms including Griphognathus, Persacanthus and a tristichopterid, Eusthenopteron. Asterolepis is part of assemblage I in Schultze & Cloutier (1996), the most widely distributed assemblage (in 24 of 38 localities). It occurs with other widely distributed genera having a marine preference, but also occurs in environments interpreted as freshwater deposits. Schultze & Cloutier (1996) concluded that there is no clear "freshwater" assemblage. Leli vre (2004) reached the same conclusion for the Famennian using a different methodology. The interpretation of these fishes as marine is thus the most plausible one. All the "freshwater" occurrences, if correct, could represent migrations of these fishes into freshwater.