The sponge genus Ephydatia from the high-latitude middle Eocene: environmental and evolutionary significance

The freshwater sponge species Ephydatia cf. facunda Weltner, 1895 (Spongillida, Spongillidae) is reported for the first time as a fossil from middle Eocene lake sediments of the Giraffe kimberlite maar in northern Canada. The sponge is represented by birotule gemmuloscleres as well as oxea megascleres. Today, E. facunda inhabits warm-water bodies, so its presence in the Giraffe locality provides evidence of a warm climate at high latitudes during the middle Eocene. The morphological similarity of the birotules to modern conspecific forms suggests protracted morphological stasis, comparable to that reported for other siliceous microfossils from the same locality.


Introduction
Freshwater sponges (Porifera, Spongillida) are common in many modern continental waters, but their fossil record is somewhat sparse. Although marine sponges have existed since the Proterozoic (Pisera 2006;Van Soest et al. 2012), the earliest freshwater sponges do not appear in the fossil record until the Permo-Carboniferous of Europe (Schindler et al. 2008). Mesozoic occurrences of Spongillida are known from the USA during the Late Jurassic, and from England and Patagonia during the Lower Cretaceous (Pisera and Saez 2003). The oldest preserved gemmules (resting bodies of sponges) were found in the latter region (Chubut Valley) (Ott and Volkheimer 1972;Volkmer-Ribeiro and Reitner 1991). More common freshwater sponges have been discovered in Paleogene and Neogene deposits, with records originating from Germany, Siberia, Japan, Chile, and South Africa (see Pisera and Saez 2003). Recently, middle Eocene (*40 Ma) lake sediments within a kimberlite diatreme in northern Canada, referred to as the Giraffe locality, have yielded a rich assemblage of Handling editor: Mike Reich.

Geological setting
The Giraffe fossil locality (Wolfe et al. 2006;Siver and Wolfe 2009;Doria et al. 2011) is a kimberlite diatreme that was intruded into the Slave Craton of the Canadian Shield 47.8 ± 1.4 Ma ago and infilled with an organic sediment sequence that accumulated post-eruptively ( Fig. 1). Over 60 m of lacustrine sediments (laminated shales and mudstones) comprise the lower maar facies, which is succeeded by *40 m of terrestrial sediments that include significant amounts of Metasequoia foliage and wood. The transition between lacustrine and terrestrial sedimentation occurred *38 Ma ago, based on glass fission-track ages (Doria et al. 2011), implying that the entire lake sequence is middle Eocene (Lutetian Stage) in age. The thermal and tectonic stability of the locality post kimberlite emplacement has resulted in excellent preservation of both siliceous and nonsiliceous fossils (Wolfe et al. 2006;Doria et al. 2011).

Materials and methods
The investigated material originates from a drill-core obtained in 1999 by BHP Billiton Diamonds Inc. during diamond exploration. The present study is based on isolated spicules obtained from core sediment samples (*200 mg) following treatment with 30 % H 2 O 2 , repeated washing with deionized water, and final suspension in reagent-grade propanol (isopropyl alcohol). The resulting suspensions were dried onto coverslips, attached with carbon glue to SEM stubs, sputter-coated with platinum, and examined with a Philips XL20 field-emission SEM (Institute of Paleobiology, Warsaw) at magnifications ranging from 1009 to 50009, using a voltage of 25 kV.  Material Approximately 100 gemmuloscleres and 100 megascleres.
Description Only disassociated spicules have been found. Gemmuloscleres are birotules ( Fig. 2) with incised margins of both rotules and spined shaft; spines are large and smooth with acute tips. The rotules are moderately smooth to strongly incised; at their center often rises a small rounded process (umbonate rotule). The number of spines on the shaft varies widely from 4 to about 20, but is usually 6-12; they are irregularly distributed and can extend in length to the rotule margin. The size of the gemmuloscleres is highly variable; length varies between about 26 and 57 lm, with an average of 41.5 lm; thickness of the shaft (not including spines) ranges from 5 to 11 lm, with an average of 7.2 lm; rotulae diameter ranges from 15 to 28 lm, with a mean of 21.4 lm. Some gemmuloscleres departing strongly from the typical morphology ( Fig. 2P-W) are interpreted here as merely malformed (ecophenotypes), because similar morphs have been observed in Recent material (see De Rosa-Barbosa 1979). Possible megascleres (Fig. 3) of the sponge skeleton (disassociated monaxial spicules from the same sample) are large oxeas measuring 162.3-307 lm (the majority between 220 and 250 lm) in length, and are 8.7-12.8 lm (usually 9-10 lm) thick in the middle. Oxeas are straight to slightly curved, microspinose, with scattered small spines or tubercles, and the ends taper to a sharp point. Some are centrotylote.
Remarks The morphology of the gemmuloscleres is highly variable in our samples, and individual specimens from morphological extremes could be mistakenly considered to represent different species. Within any given sample, there are clear transitional forms between the morphological extremes of gemmuloscleres. There is also no pattern of change in the morphology of gemmuloscleres within the stratigraphical sequence, with a similar range of variability seen across the sequence. Statistical relationships between length of gemmuloscleres and rotule diameter (Fig. 4a) also do not show any spicular groupings. Therefore, we conclude that the spicules represent a single species. We linked megascleres with the gemmuloscleres based on observations made for samples containing only Ephydatia gemmuloscleres. Our conclusion is that the slender oxeas with microspines and sharp tips belong to the same spicular complement of gemmuloscleres observed in the same samples (on the same SEM stub).
At present, only four fossil Ephydatia species are known, of which only two are well defined and described: E. guttenbergiana from the Eocene of Germany (Müller et al. 1982;Richter and Wuttke 1995) and E. chileana from the Miocene of Chile (Pisera and Saez 2003). The other two species, E. kaiseri (Rauff 1926) from the ''pre-middle Eocene'' of South Africa, and E. fossilis (Traxler 1894) from the Sarmatian (Miocene) of Hungary, are poorly circumscribed. Ampidiscs of the genus Ephydatia, without specific assignment, were also reported from the lower Oligocene of Germany (Martini and Schiller 1995). The taxon described here from the Giraffe locality differs considerably from all known fossil species. The most similar birotules in general shape and size are those of the species E. chileana from the Miocene of Chile (Pisera and Saez 2003), but the rotules of the Giraffe forms are significantly more incised. The size range and morphology of gemmuloscleres are similar in both species, but megascleres are clearly spinose in the Giraffe species, while those from the Chilean Miocene species are smooth and considerably shorter in comparison. The Giraffe fossil species is nearly identical to the recent species Ephydatia facunda as redescribed by De Rosa-Barbosa (1979) and Pinheiro et al. (2004) from Brazil (see also Nicacio and Pinheiro 2015). We have described it as E. cf. facunda due to the large time gap separating our material from the extant species. The species described as E. facunda from Cuba by The sponge genus Ephydatia from the high-latitude middle Eocene: environmental and… 675 Manconi and Pronzato (2005) has different gemmuloscleres from the type material from Brazil and represents a different species (Pinheiro et al. 2004). The megascleres that we attribute here to E. cf. facunda are morphologically very similar to those occurring in the holotype (De Rosa-Barbosa 1979), as well as newer material presented by Pinheiro et al. (2004). The megascleres from the Giraffe species are slightly smaller on average, but are within the range reported for E. facunda (De Rosa-Barbosa 1979; Pinheiro et al. 2004). In both cases the differences in spicule morphology are within the range often attributed to phenotypical plasticity relating to the environment (Poirrier 1974). The only other living species of Ephydatia known to have spines on the shaft is E. robusta (Potts 1888) from North America, but this species is often considered to be conspecific with E. fluviatilis. Very spiny shafts of birotules in the latter species are associated with harsh environmental conditions (high water temperatures, salty estuarine waters, and desiccation at gemmulation time) in Sardinian populations (RM, unpublished). E. facunda had not previously been recorded from the Nearctic Region (Manconi and Pronzato 2016), and today only two species of Ephydatia, E. fluviatilis and E. muelleri, occur in northern Canada (Ricciardi and Reiswig 1993). However, the gemmuloscleres of these latter species are easily differentiated from those of E. cf. facunda from the Giraffe locality.

Discussion
Ephydatia is a very common, cosmopolitan genus known from all over the world except Antarctica. Six species have been recorded from the Northern Hemisphere, while only three species are known from the Southern Hemisphere (Manconi and Pronzato 2008). Due to the large variability The sponge genus Ephydatia from the high-latitude middle Eocene: environmental and… 677 in gemmulosclere morphology (Poirrier 1974), there is a lack of agreement regarding both the number of Ephydatia species in existence today and the range of variability within individual species. This makes studies of fossil material based on isolated gemmuloscleres especially difficult. The fact that representatives of this genus are often reported in the fossil record follows, without doubt, from the fact that its gemmuloscleres are very characteristic and easy to determine. This also reflects the fact that Ephydatia is a common, widespread, and euryoecious genus. Such a pattern of occurrence may be explained by its long evolutionary history that allowed for diversification and wide dispersal, a hypothesis supported by our findings.
Since Ephydatia spans cold to warm climates, it is difficult to use genus-level remains to infer temperature conditions in the Giraffe locality. However, if, as we believe, our material represents the tropical species E. facunda, then it does indeed suggest that this Arctic locality had a warm climate during the middle Eocene. Today, E. facunda is restricted to warm climates in Brazil and Argentina (Nicacio and Pinheiro 2015). In addition, other sponges occur in the same fossil samples (Pisera et al. 2013, 2014, andin prep.) and warm-water diatoms and synurophytes are also present (Siver and Wolfe 2009), supporting the hypothesis that the Giraffe locality was indeed significantly warmer than today during the middle Eocene.
In modern ecosystems, Ephydatia facunda is characteristic of freshwaters that are eutrophic, circumneutral, and mesohaline with abundant macrophytes (Volkmer-Ribeiro et al. 2004Volkmer-Ribeiro and Machado 2007). However, there is no indication that the Giraffe lake ecosystem was ever a salty or brackish halobian environment, given that there are neither carbonates, evaporates, nor halophilous organisms anywhere in the sediments. Because middle Eocene forms of E. cf. facunda inhabited dilute and slightly acidic waters, as inferred from co-occurring algal microfossils (Wolfe et al. 2006;Siver and Wolfe 2009), the ecological tolerance of this sponge has either changed over time or is broader than previously envisaged. Alternatively, warm conditions may represent the dominant environmental control over the distribution of this species, with the capacity to override the influences of salinity and nutrients.
Furthermore, in some samples, there are spicules which resemble typical birotules of E. facunda but differ in size, the degree and character of rotule development, and the number of spines (Fig. 2P-W). Here, we treat them as ecophenotypes (malformed or teratological forms) of typical E. cf. facunda birotules rather than representatives of a clearly different taxon. Similar ranges of gemmulosclere morphology are observed in the type material (De Rosa-Barbosa 1979), whereas teratological Ephydatia gemmuloscleres have been reported in modern populations exposed to elevated trace metal concentrations (Poirrier 1974;Richelle-Maurer et al. 1994;Pisera and Saez 2003). One possibility is that teratological forms of E. cf. facunda in the Giraffe material have arisen because the taxon existed outside its ecological optimum with respect to salinity, as discussed above.
The well-preserved spicules of Ephydatia allow for a more precise comparison with extant species. The Giraffe gemmuloscleres fit unambiguously within the description of gemmuloscleres for E. facunda from Brazil. Our findings imply that the morphology of birotules is highly conserved and has undergone little (if any) evolutionary change since the middle Eocene. Spicular characters are also extremely well conserved in the marine sponge clades (for example : Łukowiak 2015). Similar results have been reported for euglyphid thecamoebans (Foissner and Schiller 2001;Barber et al. 2013), synurophytes (Siver and  Wolfe 2005;Siver et al. 2013), and additional sponge groups such as the potamolepids (Pisera et al. 2013(Pisera et al. , 2014.

Conclusions
Middle Eocene birotule gemmuloscleres from the Giraffe locality's lake deposits in northern Canada have been identified as being very similar if not identical to gemmuloscleres of the extant species Ephydatia facunda Weltner (1895), but due to the large time gap separating them, they are described as deriving from E. cf. facunda. This finding represents yet another example of protracted evolutionary stasis with respect to the morphology of siliceous organisms, and of the expanded distribution of warm stenothermous taxa to the northern high latitudes during the Eocene greenhouse climate episode.