An endemic microphytoplankton assemblage from Middle Devonian Iberia and its paleogeographical significance

Diverse microphytoplankton assemblages, including 72 taxa belonging to 27 genera of acritarchs and prasinophyte phycomata, have been recovered from Middle Devonian rocks in northern Spain, revealing an endemic flora dissimilar to coeval assemblages. These deposits are of early Givetian age and consist of the laterally equivalent Naranco, Huergas and Gustalapiedra formations of Asturias, León and Palencia provinces. At the time, Iberia was part of the Armorican Terrane Assemblage, a comparatively isolated island chain positioned between Laurussia and Gondwana. The studied formations represent a marine transect across a nearshore-offshore gradient and consist of a large clastic unit sandwiched between extensive carbonate deposits. This clastic unit incorporates the Kačák Event, an important global extinction ev ent associated with marine anoxia. Herein, the suite of generally well-preserved microphytoplankton assemblages is described and their stratigraphical and biogeographical importance are considered. The microphytoplankton represents a single assemblage deposited in a short interval and is interpreted as being endemic. The assemblage is only moderately similar to contemporary assemblages from Laurussia and Gondwana and, although certain characteristic Middle Devonian taxa are present, other common species such as Arkonites bilixus and Tyligmasoma alargada are absent. While no unique taxa are found here, the taxa which are present represent a particular combination of species not seen elsewhere. Certain taxa appear which may have discordant temporal ranges, though no major inferences can be made from them as only two taxa both occur in significant numbers and have a confident identification. This assemblage adds to our knowledge of phytoplankton palaeobiogeography, representing a significantly endemic assemblage within the generally cosmopolitan microphytoplankton flora of the Middle Devonian.


Introduction
Prasinophyte phycomata and acritarchs (hereafter collectively termed microphytoplankton) from the Middle Devonian deposits of northern Spain have received very limited study, having only brief mentions in two papers predominantly focused on land-plant spores (Cramer, 1969;Fombella Blanco, 1988). This is in spite of the rock formations themselves being otherwise well studied and presenting a number of interesting features. They comprise a clastic unit positioned between thick carbonate deposits, representing a completely marine onshore-offshore transect, well age-constrained by conodonts and dispersed land-plant spores, and well characterised sedimentologically. The sequence merits particular attention owing to: (1) its location around a group of relatively small islands isolated from the large supercontinents of Laurussia and Gondwana Cocks, 2013, 2016) and (2) the clastic unit incorporating the Kačák

A C C E P T E D M A N U S C R I P T
Event (García-Alcalde, 1998;Askew & Wellman, 2018), a major global extinction event of uncertain cause. This study comprehensively describes the composition of the microphytoplankton assemblage from this sequence for the first time, identifies differences between this community and contemporary assemblages from Laurussia and Gondwana and discusses the palaeobiogeographical implications of this assemblage within the Middle Devonian world.

Geological setting
This study utilises the same samples as Askew & Wellman (2018) and their geological setting is described in detail therein. Briefly, the studied Naranco, Huergas and Gustalapiedra formations are lateral equivalents, found in Asturias, León and Palencia provinces, respectively.
They are bounded by limestone formations, the older Moniello, Santa Lucía and Polentinos formations below, and the younger Candás, Portilla and Cardaño formations above, all similarly laterally equivalent (Fig. 1). The boundaries between the formations are described as conformable (IGME, 2015). The Naranco, Huergas and Gustalapiedra formations represent a clastic interruption in carbonate deposition and consist of a coarse, sandstone lower unit and a more mixed upper unit containing alternating sandstones and siltstones (with occasional limestone beds), indicating somewhat reduced clastic input compared with the lower unit.
Detailed descriptions of the various facies present and their environmental interpretations can be found in García-Ramos (1978).

A C C E P T E D M A N U S C R I P T
The marine macrofauna from the Naranco, Huergas and Gustalapiedra formations indicate an Eifelian-Givetian age , refined by conodont faunas found in the surrounding limestones. These give an age range from the Polygnathus costatus costatus zone (middle Eifelian;ca. 391-392 Ma (Becker et al., 2012)) to the lower Polygnathus varcus zone (early, but not earliest, Givetian;ca. 386-387 Ma (Becker et al., 2012)) . The dispersed spore assemblage described in Askew & Wellman (2018) supports this, though suggests a short period of deposition for the palynomorph-bearing deposits of slightly more than 1.5 million years (ca. 386-387.7 Ma), entirely within the early Givetian lemurata-langii Assemblage Zone (Geminospora lemurata Interval Zone) (Breuer and Steemans, 2013). This places the assemblage outside the stratigraphic range of the Kačák Event as traditionally defined, though see Askew & Wellman (2018) for a fuller explanation.

Previous palynological studies
The two previous studies on the Middle Devonian palynomorph assemblages from Spain focused primarily on the dispersed spore assemblages rather than the marine palynomorphs.

A C C E P T E D M A N U S C R I P T
These taxa are figured, but with no other information given. Additionally, the spore species Acanthotriletes espinositus Cramer and A. tenuispinosus Naumova are reported, though the figured specimens more closely resemble acritarchs, possibly species of Gorgonisphaeridium, with no description provided to confirm their spore affinity.

Material and methods
One hundred and thirty rock samples were collected, covering the outcrop area of the Naranco, Huergas and Gustalapiedra formations. Samples were taken from dark, fine-grained beds of siltstone and mudstone interspersed within the layers of sandstone that dominate the formations. Sample sites were mostly either single exposures or short sections, along with three relatively long sections sampled at intervals. Sample details are given in Askew & Wellman (2018). Samples were processed using standard acid maceration techniques; dissolution in HCl, followed by removal of HCl and replacement by HF and, finally, the addition of HCl to the samples after a short time. The resulting mixture was left until sample break-down was complete, at which point some fine mineral residues were removed using a 15 µm sieve. Heavy liquid centrifugation using ZnCl 2 removed remaining mineral matter. Subsequent assessment of the A C C E P T E D M A N U S C R I P T resulting kerogen samples determined three barren samples and one with very poor preservation; these were not processed further.
The palynomorphs are generally well preserved but thermal maturity is variable, with a Thermal Alteration Index (TAI) score between 2-and 4 (Traverse, 2008). Schulze's solution was used to oxidise all the samples for between five and sixty minutes, except for four very thermally mature samples that were oxidized for around nineteen hours. The samples were spiked with tablets containing modern Lycopodium spores at a ratio of one tablet per millilitre of solid residue, to facilitate quantitative assessment of the palynomorph assemblage using an equation derived from that given by Stockmarr (1971). Two to four strew mounts were made for each sample, with 113 samples counted. Two hundred palynomorphs were counted on each slide with additional species noted as rare. In the rare cases where a slide did not contain two hundred palynomorphs, every palynomorph on the slide was counted. Such slides are evident in the count data given in Askew (2019). Slides were examined using a light microscope and specimens photographed using a Meiji Techno Infinity 1-5C camera attached to a Meiji Techno MT5300H transmitted light microscope.

Description of the microphytoplankton assemblage
Full systematic descriptions for the taxa found in this phytoplankton assemblage are given in the supplementary text and a list of taxa recorded is given in Appendix A. Plates I and II depict taxa that are either not assignable to existing species or have not been recorded from the Middle Devonian before (see section 7).

A C C E P T E D M A N U S C R I P T
The microphytoplankton assemblage consists of 72 taxa (59 identified species belonging to 27 genera) of which 10 taxa are assigned only to genera, having no previous representation in the literature at species level. Full charts of the sampled sections are presented as supplementary information, while charts of the quantitative abundances of selected taxa are given as figures 3-5. Count data is available as supplementary table S1 in Askew (2019) O. furcillatus are known to occur across the stratigraphic range of the formation, while the remaining species are previously known only from much older deposits but are mostly rare here (discussed in section 7). It is difficult to separate any of the small outcrop sites from the rest of the assemblage based on this data, given the close similarity of the assemblages as a whole. The sampled sections show few taxon inceptions during their duration, with no consistency in which taxa occur later and none becoming significant components of the assemblage.
Reworking is considered unlikely in this material. It is not evident in the dispersed spore assemblage as discussed by Askew & Wellman (2018). The phytoplankton show no apparent differences in preservation and thermal maturity and there is no obvious systematic reworking of older forms (for example those known from the underlying La Vid Group) into the assemblage.

Comparison with other microphytoplankton assemblages
Middle Devonian microphytoplankton assemblages are better known than, for example, Early Devonian ones, however the stratigraphic and geographic distribution of well-described assemblages is still uneven (Le Hérissé et al., 2000;Molyneux et al., 2013;Wicander and Playford, 2017a). Many are clustered in particular areas, particularly North America, yet many of these are not sufficiently stratigraphically constrained for valid comparisons to be made. These assemblages, as well as those of better stratigraphic constraint, are well-reviewed by Wicander andPlayford (2017a, 2017b) for North America and Molyneux et al. (2013) globally. Differing taxonomic opinions also confuse the issue, with species concepts used differently by different authors. However, numerous papers, such as those cited in Table 1, can be used for comparative purposes, allowing comparisons to be made between this assemblage in northern Spain and contemporary assemblages in Laurussia, Gondwana and China (considered separately here).
Three measures of similarity are used here. (1) Percentage similarity (sensu Wicander andPlayford (2017a, 2017b)) is simply the percentage of co-occurring species between this assemblage and another. (2) The coefficient of similarity (CS) (sensu Clark and Hatleberg (1983)) can be expressed as CS = 2|x|/(|a + b|) where a and b are the total number of species in assemblages a and b, respectively, and x is the number of species they share. Results below 0.2 are considered to indicate low similarity with results between 0.2 and 0.55 considered moderate to high (Clark and Hatleberg, 1983). (3) The Jaccard Index (JI) can be expressed as JI = |x|/(|a| + |b| -|x|) where a and b are the total number of species in assemblages a and b, respectively, and x is the number of species they share. This gives a value between 0, for completely dissimilar assemblages, and 1 for identical assemblages. These latter two measures have been used previously in Palaeozoic palynological applications (e.g. Le Hérissé et al., 1997;Wellman, 2018;Wellman et al., 2013). In all cases, taxa assigned only to genera and those designated "?" have been excluded from the calculation, while those designated "cf." have been treated as valid identifications where appropriate.
Values for these metrics applied to 18 Middle Devonian microphytoplankton assemblages are given in Table 1. The locations of the assemblages are shown on a palaeogeographic map in figure 6. Values for percentage similarity do not seem to scale well with the other metrics; the present author prefers to make the below conclusions based on CS and JI. The present assemblage is seen to not be particularly similar to any other Middle Devonian assemblage with good stratigraphic and taxonomic control. It is most similar to the assemblages of Playford

A C C E P T E D M A N U S C R I P T
(1977) (CS:0.375; JI:0.231) and Wicander and Wood (1981) (CS:0.347; JI:0.210) from Canada and the USA, but even this is no more than a moderate degree of similarity. A comparable degree of similarity is seen with the Argentinian assemblage of Barreda (1986)  On a global scale, Middle Devonian microphytoplankton assemblages are broadly cosmopolitan in mid to high palaeolatitudes, with many species in common between the two main supercontinents of Laurussia and Gondwana (Molyneux et al., 2013). Western China also shows a marked similarity to these supercontinental assemblages (Zhu et al., 2008). Some endemic species do, however, exist within this general cosmopolitanism, primarily in the high palaeolatitudes of Gondwana (Molyneux et al., 2013). Phytoplankton provincialism in the Late Devonian has been suggested to be a result of temperature differences and ocean currents (Shen et al., 2018), and there is no reason to believe similar factors are not at play in the Middle Devonian, restricting some species to high-latitude Gondwana.

A C C E P T E D M A N U S C R I P T
The Spanish assemblage analysed here is shown to be endemic to a relatively large degree, being dissimilar to some extent to all the studies analysed. It is important to note this is not due to new, endemic species found in Spain. Any potential new species in this study were not found in great enough numbers to justify erecting new species; they were left in open nomenclature and, hence, were excluded from the similarity analysis. Instead, it is the particular combination of species seen here, including both predominantly Laurussian and predominantly Gondwanan species while also lacking forms common elsewhere, that renders this assemblage distinct.
This assemblage contributes to our knowledge of Middle Devonian phytoplankton biogeography by describing an endemic assemblage outside the high latitudes of Gondwana, characterised by a specific combination of co-occurring species. Spain was an isolated island during the Middle Devonian (see section 1 and figure 6), therefore its endemism could be due to this geographic isolation, possibly aided by ocean currents.

Biostratigraphical implications
Biostratigraphic schemes for Middle Devonian phytoplankton are lacking owing to a paucity of comprehensive study (Le Hérissé et al., 2000). Simple range charts are more common, however they are not particularly useful here. For example, the only Spanish species that are used in the chart of Wicander and Wood (1997) range throughout the Middle Devonian. The same is almost true of the chart given by Le Hérissé et al. (2000); all but two species found here are shown to be long-ranging. The two exceptions are Palacanthus tripus and Villosacapsula globosa that are both indicated as Frasnian-Famennian species.

A C C E P T E D M A N U S C R I P T
As mentioned in section 2, the formation in which this assemblage was found is well ageconstrained by conodonts and dispersed spores (Middle Devonian: early Givetian). In terms of acritarchs, the zonation scheme of Jardiné et al. (1974) has the Middle Devonian corresponding to zones L1-L3. Zones L1 and L2 are not differentiated, being characterised by (using modern taxonomy) Duvernaysphaera angelae, D. kraeuselii, Eisenackidium valentina, Multiplicisphaeridium exasperatum, Navifusa bacilla, Polyedryxium decorum, P. fragosulum, P.
decorum Deunff, 1955', 'P. cf. talus Deunff, 1966 and Veryhachium stelligerum. The underlying K and overlying L4 zones do not include any species found in the present assemblage. Interestingly, this scheme depicts V. valiente as going extinct in the Ludlow, in stark contrast to other reports of its existence in the Devonian, many of them cited in Table 1.
Also found here are two representatives of the later L3 zone. P. "talum" has a complicated taxonomic history, ultimately being invalid as it was named in an unpublished thesis (Fensome et al., 1990). As the description for 'P. cf. talus Deunff, 1966', cited by Jardiné et al. (1974, is in that thesis and unavailable to the present author, no stratigraphic conclusions have been drawn from it. This leaves only V. stelligerum representing the L3 zone, found here only in various samples taken from the section at the Crémenes-Las Salas site and potentially indicating a slightly later date. This would fit with the site's south eastern location, interpreted as further offshore; the clastic sediment may have taken longer to arrive here and produced a degree of diachronism in the Naranco, Huergas and Gustalapiedra formations. It should be noted, however,

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T that V. stelligerum has been reported elsewhere across a much wider time range (see supplementary text). The lack of precise ages for these biozones is unfortunate; they simply indicate an Eifelian-Givetian age, agreeing with previous interpretations.
Certain taxa reported here have not been previously recorded in the Middle Devonian. The potentially discordant ranges of these taxa could be explained by reworking, though this is considered unlikely. There is no suggestion of reworking in the dispersed spore assemblage and the clear majority of the microphytoplankton community is typically Middle Devonian. The uncertain identification and rarity of most of the apparently temporally aberrant taxa mean they are probably not of great biostratigraphic importance.

A C C E P T E D M A N U S C R I P T
 A rich assemblage of generally well-preserved microphytoplankton has been recovered from Middle Devonian deposits in northern Spain.
 The assemblage shows a marked endemism, having a low similarity to assemblages from both Laurussia and Gondwana, reflecting the isolated position of Iberia during the Middle Devonian. Certain taxa also have potentially discordant ranges.
 The microphytoplankton assemblage supports the Middle Devonian age of the formation, agreeing with conodont and dispersed spore data.
 This new assemblage adds to our knowledge of endemic assemblages existing within the broadly cosmopolitan microphytoplankton assemblages of the Middle Devonian.