An endemic flora of dispersed spores from the Middle Devonian of Iberia

Diverse assemblages of dispersed spores have been recovered from Middle Devonian rocks in northern Spain, revealing a significant endemism in the flora. Middle Devonian Iberia was part of a relatively isolated island complex (Armorican Terrane Assemblage), separated by considerable tracts of ocean from Laurussia to the north‐west and Gondwana to the south‐east. The Middle Devonian deposits of the Cantabrian Zone of northern Spain are entirely marine and comprise a thick clastic unit sandwiched between extensive carbonate units. The clastic unit, the laterally equivalent Naranco, Huergas and Gustalapiedra formations of Asturias, León and Palencia provinces, represents a nearshore‐offshore transect across a marine shelf. This unit is also believed to encompass the Kačák Event, an important global extinction event. The recovered palynomorphs include marine (phytoplankton, chitinozoans, scolecodonts) and terrestrial (spores) assemblages. These are abundant and well preserved, although of variable thermal maturity. Here, we describe the dispersed spores and consider their significance as regards biostratigraphy, palaeophytogeography and Kačák Event interpretation. The dispersed spores represent a single assemblage assignable to the lemurata–langii Assemblage Zone (lemurata Subzone) indicating a probable early (but not earliest) Givetian age. Signs of endemism include various taxa known only from this region, some taxa appearing to have discordant ranges compared with elsewhere, and the absence from Iberia of certain prominent taxa characteristic of coeval assemblages elsewhere, such as those with grapnel‐tipped processes. The abrupt interruption of carbonate deposition, with a change to rapid deposition of thick clastic deposits, provides support for a monsoonal cause of the Kačák Event.

D I S P E R S E D spore assemblages from Middle Devonian deposits of northern Spain have previously been relatively neglected and briefly considered in only three publications (Cramer 1966(Cramer , 1969Fombella Blanco 1988). This is unfortunate as these deposits are otherwise well studied and exhibit a number of extremely interesting features. They represent an entirely marine onshore-offshore transect, represented by a clastic unit sandwiched between thick carbonate sequences, securely age-constrained by conodonts and sedimentologically/palaeontologically well characterized. The sequence is of particular interest because: (1) it accumulated around a small group of islands in a palaeogeographically isolated location situated between the large continents of Laurussia and Gondwana (Torsvik & Cocks 2013); and (2) the clastic unit is believed to incorporate the Ka c ak Event (Garc ıa-Alcalde 1998). This extinction event has been identified globally (House 2002) although its characteristics and causes remain the subject of much debate (Marshall et al. 2007;Becker et al. 2016). The aim of this study is to describe the composition and distribution of dispersed spore assemblages from throughout this sequence and use them to test and refine previously erected spore biostratigraphical schemes, identify palaeophytogeographical differentiation on this isolated island group, and shed light on the nature of the Ka c ak Event.

GEOLOGICAL SETTING
Devonian sediments crop out in the Cantabrian Zone in northern Spain in a large partial arc from the northern coast near Gij on, south through Asturias and in an eastwest orientation across Le on and Palencia provinces (Garc ıa- Alcalde et al. 2002). They are interpreted as having been deposited in a transect across the shelf from nearshore (Asturias) to offshore (Palencia) (Garc ıa- Alcalde et al. 2002). Details of the stratigraphic sequence are given in Figure 1. The sedimentary sequence broadly alternates between calcareous and clastic in character. This paper concerns the Middle Devonian clastic unit classified as the laterally equivalent Naranco, Huergas and Gustalapiedra formations located in Asturias, Le on and Palencia, respectively. They are bounded by thick limestone sequences below (the laterally equivalent Moniello, Santa Luc ıa and Polentinos formations) and above (the laterally equivalent Cand as, Portilla and Cardaño formations).
The age of the Naranco, Huergas and Gustalapiedra formations is Eifelian-Givetian, supported directly by biostratigraphical analysis based on various marine macrofaunal groups (Garc ıa-L opez et al. 2002). Conodonts from the bounding limestone formations provide a refined age constraint. Those from the underlying Santa Luc ıa Formation (and possibly lowermost Huergas and Naranco formations) belong to the Polygnathus costatus costatus The sediments of the Naranco, Huergas and Gustalapiedra formations are in excess of 500 m in thickness in Asturias but reduced to only 50 m in thickness in Palencia. The deposits are characterized by a lower and an upper unit. The lower unit is coarser in character, dominated by thick sandstone layers, and represents a more abundant terrigenous clastic supply. The upper unit has a more mixed character, with rapid alternations between sandstones and siltstones (and occasional limestones) and represents a combination of course terrigenous and finer basinal deposits. In Le on, it includes euxinic nodular black shales that are pelagic-like in character. Garc ıa-Ramos (1978) details the wide range of facies present in the Naranco Formation and its equivalents that represent varied marine and transitional depositional environments associated with large deltas (Garc ıa- Alcalde et al. 2002).
For this project, 30 localities were examined and samples collected for palynological analysis. The location of each site is indicated in Figure 2. Details of the sites' location and precise geological setting, as well as samples taken, are provided in Askew & Wellman (2018).

PREVIOUS PALYNOLOGICAL STUDIES
Previous palynological studies of the Naranco and Huergas formations are limited to three papers. Cramer (1966) described an assemblage from one sampled site of the Naranco Formation, believed to be the '120' site of Cramer (1969) and this study. Cramer (1969) expanded the earlier work with four additional sites from the same general area. These studies described a spore assemblage containing many new species, with others mostly showing an affinity with Russian and European floras known at the time, relatively early in the history of Devonian spore studies. Cramer interpreted the spore assemblage as supporting the Eifelian-Givetian age the Naranco Formation had previously been assigned based on invertebrate macrofossil evidence. Fombella Blanco (1988) described an assemblage of spores (including one newly described species) and phytoplankton from a single site in the Huergas Formation, relocated during the present study and described in Askew & Wellman (2018), but made no comment on the formation's age. Table 1 lists the species named in these studies, along with a correlation with the species identified and taxonomy utilized in the present study.

SAMPLING AND METHODS
A total of 130 samples were collected for palynological analysis from across the outcrop area of the Naranco, Huergas and Gustalapiedra formations during two fieldtrips in January and August-September 2015. Sampling focused on beds of dark, fine-grained material, predominantly mudstones and siltstones, interspersed within the mostly sandstone formations (Fig. 3). As outlined in Askew & Wellman (2018), three relatively long sections, two incomplete and one nearly complete, were sampled with the remaining localities representing either spot samples or relatively short sections. It is important to note that during the extensive folding and thrusting of the Spanish sequence the clastic sediments of the Naranco, Huergas and Gustalapiedra formations often acted as a slip plane between the more rigid limestone formations above and below. Furthermore, compared to the bounding limestones they are often deeply weathered, with surface exposures vegetated. As a consequence: (1) long sections are rare (see above); (2) good upper and lower boundary sections are rare and difficult to identify; and (3) short exposed sections are difficult to place in stratigraphical context. Rock samples were processed at the University of Sheffield using standard HCl-HF-HCl acid maceration, followed by sieving using a 15 lm mesh to remove the finest mineral matter. Remaining mineral residue was removed using heavy liquid centrifugation with ZnCl 2 . The organic residue was then assessed, with three barren samples and one with very poor preservation not being processed further. The remaining kerogen samples were evaluated by strew mounting on microscope slides and viewing using a light microscope.
The palynomorph assemblages are mostly well preserved but of variable thermal maturity, between 2À and 4 on the TAI scale (Traverse 2008). All samples were oxidized in Schulze's solution for between 5 and 60 min, except for four extremely mature samples that were oxidized for around 19 h. At this stage, the samples were spiked with Lycopodium tablets at a ratio of one tablet per millilitre of solid yield. Two to four strew mounts were made of these final preparations. From these, 113 samples were counted to 200 identifiable palynomorph specimens or the end of the slide in the occasional samples with less than 200 palynomorphs per slide. A note was kept of the number of modern Lycopodium spores encountered to enable quantitative assessment of the fossil material using an equation derived from one given by Stockmarr (1971). Slides were examined using standard light microscopy techniques and photographed using a Meiji Techno Infinity 1-5C camera mounted on a Meiji Techno MT5300H transmitted light microscope.

SYSTEMATIC PALAEONTOLOGY
Taxa are organized alphabetically by genera and species. Species with an uncertain identification are signified cf. or ?, after Matthews (1973). A limited synonymy is provided for species previously recorded from the Naranco and Huergas formations. Dimensions are given for each species as diameters and where three numbers are given these correspond to the minimum value (arithmetic mean) and maximum values. Species occurrences are given as locality code characters for brevity (for an explanation see Askew & Wellman 2018). Previous records are taken from consultation of the John Williams Index of Palaeopalynology (for details see Riding et al. 2012). Materials (rock samples, residues and slides) are housed in the collections of the Centre for Palynology, Department of Animal and Plant Sciences, University of Sheffield, UK. Figured specimens are located using an England Finder. Genus ACINOSPORITES Richardson, 1965 Type species. Acinosporites acanthomammillatus Richardson, 1965. Acinosporites acanthomammillatus? Richardson, 1965 Figure 4A, L Description. Amb shape uncertain. Laesurae not observed. Exine appears quite thick, ornamented with biform processes. Sculptural elements are rounded, up to 6 lm wide, often joined to form ridges and surmounted by apical spines up to 4 lm high.
Remarks. Identification is highly questionable, as it is based entirely on the characteristic ornament of rounded ridges topped by spines, in a low number of incomplete and very poorly preserved specimens. Previously reported species are reproduced as cited and followed in parenthesis by the reference in which they were reported. Previously reported species synonymized with species found in this investigation are listed against them.
Occurrence. Sites 3, 11, G. Description. Amb circular to subcircular. Laesurae straight, extending one-half to the whole amb radius. Equatorial thickening of 2-4 lm observed, with some minor separation of the exoexine from a possible inner body. Proximal surface laevigate. Distal surface bears an ornament of microgranulae, large conate elements up to 3 lm high and wide as well as more irregular, parallel sided, sometimes biform elements up to 5 lm high.
Remarks. This genus is distinguished by bearing its ornament atop anastomosing ridges. These ridges are not visible in these specimens except for some small joining of ornament bases. However, this species has been repeatedly recorded as having an extremely variable morphology, including a lack of obvious ridges (McGregor & Camfield 1976;Richardson et al. 1993 Dimensions. 21 (35) 54 lm (43 specimens measured).
Previous records. Reported from Upper Ordovician to Eifelian strata and with an almost worldwide distribution (although absent from Australia).

Figure 4D
Description. Amb circular to subcircular. Laesurae straight, extending from one-third to five-sixths of the amb radius. Tips of the laesurae may be slightly spatulate. Wall 1-2 lm thick. Surface laevigate.
Remarks. This taxon is distinguished from A. dilutus by its shorter laesurae.

Figure 4H
Description. Amb circular to subcircular. Laesurae straight, accompanied by narrow labra along their length, extending from one-half to two-thirds of the amb radius. Wall c. 1 lm thick. Surface laevigate.
Remarks. This taxon is distinguished from A. sp. A by its possession of labra.

Figure 4G
Description. Amb circular to subcircular. Laesurae straight, accompanied by narrow labra along their length, extending from three-quarters to the whole amb radius. Wall c. 1 lm thick. Surface laevigate to occasionally scabrate.
Remarks. This taxon is distinguished from A. sp. B by its longer laesurae and from A. dilutus by its distal ornament.
Remarks. Some of the specimens have shorter trilete rays than specified in the original description, though this was considered to be insufficient to justify an alternative identification.

Figure 4I
Description. Amb circular to subcircular. Laesurae straight, extending from two-fifths to the whole amb radius. Equatorial thickening of 1-3 lm observed. Proximal surface laevigate. Distal surface bears an ornament of microgranulae.
Remarks. Distinguished from A. abrepunius by its denser ornament. Specimens seen here may have shorter laesurae than originally described.
Remarks. The species as originally described has a less variable trilete size and slightly larger ornament than the specimens found here. Owens ( Remarks. The species as originally described is markedly larger than the specimens described here, while the ornament seen here occasionally exceeds 1 lm in height, hence the uncertain identification. Occurrence. Sites 2, 3, 4, 5, 7, 10, 11, 13, 14, 20, 22, 598, G. Previous records. Reported from Pragian-Givetian strata and with a worldwide distribution (Breuer & Steemans 2013).
Previous records. Reported from Lower-Middle Devonian strata with a worldwide distribution (Breuer & Steemans 2013).
Apiculiretusispora? sp. A Figure 6G-H Description. Amb circular to subcircular. Laesurae straight, extending to the equator. Possible 2-3 lm wide equatorial thickening observed. Proximal face bears subcircular thickened region extending from two-thirds to three-quarters of the distance to the equator. Distal surface bears an ornament of microgranulae.
Remarks. These specimens are doubtfully assigned to this genus owing to their lack of obvious curvaturae, although these could be incorporated in the thickened wall. Some previously figured specimens of Apiculiretusispora do not show obvious curvaturae, such as Breuer & Steemans (2013, fig. 10F). This species' interradial thickenings are reminiscent of the structures seen in A. arabensis Al-Ghazi, 2009 although they are not of such a clear rounded shape.
Occurrence. Site G.

Figure 6A
Description. Amb circular to subcircular. Laesurae straight, extending almost to the equator. Laesurae connected by curvaturae perfectae delimiting obvious proximal contact areas, up to 15 lm away from the equator. Distal surface bears an ornament of microgranulae and granulae together with a prominent, F I G . 6 . Each figured specimen is identified by slide including the sample code, followed by its England Finder reference. A, cf. Dimensions. 147 (198) 268 lm (2 specimens measured).
Remarks. This taxon could be accommodated by the genus Apiculiretusispora except that it is outside the normal size range for the genus, with the larger specimen being of megaspore size.
Occurrence. Site G.
Remarks. The identification of this species is doubtful for similar reasons cited in McGregor (1973); the original description and figures are inadequate for confident identification of modern material. In addition, the specimens seen here have slightly larger luminae than the specimens in McGregor (1973 Dimensions. 24 (32) 41 lm (5 specimens measured).
Occurrence. Sites 7, G. Description. Amb subtriangular with straight to slightly convex interradial margins. Laesurae straight, simple or accompanied by narrow labra along their length, extending from one-half to twothirds of the amb radius. Proximal surface bears thickened kyrtomes in the interradial areas, reaching the tips of the laesurae. Surface otherwise laevigate.
Remarks. This species meets the generic criteria for Concavisporites, although this genus has not previously been used in the Palaeozoic.

Figure 7A
Description. Amb subcircular to vaguely subtriangular. Laesurae not easily discerned, believed to extend to the equator. Proximal surface presumed to be laevigate. Distal surface bears an ornament of densely packed, tapering spines up to 19 lm high. Spines often have wide bases with marked shoulders at around one-third to one-half of their height, before a thinner tapering portion begins. Large, rounded granulae are also seen, though these are probably broken spine bases.
Remarks. The specimens seen here are sometimes smaller than those of Turnau (1996) and with less variation in the length of the spines. Aside from this we draw similar conclusions: the species resembles Acinosporites macrospinosus Richardson, 1965 except for having more densely distributed spines not clearly superimposed on convolute ridges.
Occurrence. Site G.
Previous records. The study of Turnau (1996) was carried out on Middle Devonian strata from Poland. A. macrospinosus, which the species has been compared to, is reported from upper Pragian to Lower Carboniferous strata and has a worldwide distribution.
Remarks. The specimens described here do not show obvious curvaturae as seen in the original figured specimens. These specimens resemble Latosporites sp. B Owens, 1971, though Latosporites is a laevigate genus.
Occurrence. Sites 2, 3, 4, 5, 7, 10, 13, 14, G. Description. Amb subtriangular with straight to slightly convex interradial margins. Laesurae straight, extending from two-thirds to the whole amb radius. Proximal surface laevigate. Distal surface bears an ornament of microgranulae, with a prominent row of spines around 1 lm high in the interradial regions. The areas opposite the laesurae are laevigate to very finely scabrate.
Dimensions. 26 (29) 30 lm (3 specimens measured). Remarks. The specimens seen here have shorter interradial spines than the species as originally described, possibly attributable to damage.
Occurrence. Sites 3, 7, G. Description. Amb circular to subcircular. Laesurae straight, extending from three-quarters to the whole amb radius. An equatorial thickening up to 4 lm is sometimes observed. Curvaturae perfectae joining the tips of the laesurae are sometimes observed, often coincident with equator. Proximal surface laevigate. Distal surface bears an ornament of rounded granules and pointed spines. Well-preserved specimens may show slightly bulbous tips on the spines. These ornamental units are 2-3 lm high and separated by at least 1-2 lm.
Remarks. This species is distinguished from D. eifeliensis by its original authors by having a coarser ornament. The stated ornament dimensions in the original descriptions of the two species are almost identical, however, and we consider this species' much greater wall thickness to be a more useful diagnostic feature.
Occurrence. Sites 2, 3, 7, 11, 20, 120, 598. Description. Amb circular to subcircular. Laesurae not observed. Equatorial thickening of 5 lm observed. Surface (probably distal) bears a dense ornament of clavate elements, with narrow bases c. 1 lm wide and widened heads up to 3 lm wide. Elements have rounded tips and are up to 10 lm high.
Remarks. The sculptural elements on this species closely resemble those of D. pilatus Breuer et al., 2007, except for being slightly narrower, often taller and distributed much more densely, in contrast to the sparse, irregular ornament of D. pilatus. The specimen also resembles a small example of the megaspore Jhariatriletes emsiensis Moreau-Benoit, 1979. Only a single specimen was found.
Remarks. Distinguished from D. cf. hemeri in this study by its rather more substantial looking muri and more irregularly shaped luminae. The figured specimen could be interpreted as having an interradially thickened wall, in the fashion of Camarozonotriletes, but we interpret this specimen as having a widely opened trilete mark, leaving the proximal exine gathered around the margins. Though known primarily from earlier sediments this species is known from the Givetian, through a taxon known from the Givetian of Spitsbergen and placed in open nomenclature by Allen (1965) Description. Amb circular to subcircular or ovoid. Laesurae straight, sometimes accompanied by narrow labra along their length, extending from one-half to the whole amb radius. Proximal surface laevigate. Distal surface bears a reticulate ornament with muri, c. 1-2 lm high and wide, enclosing mostly polygonal luminae 2-6 lm wide. Small projections may occur at muri junctions.
Remarks. The species as originally described is rather larger and with larger sculptural elements than the specimens described here. In addition, the biform elements at muri junction are not obvious in these specimens, though this is also the case on some of the original figured specimens.
Previous records. Reported from the Eifelian-Givetian of Spain (Cramer 1969)  Description. Amb circular to subcircular. Monolete mark observed, extending to the equator. Equatorial thickening of 3 lm observed. Proximal face bears interradial muri, six in each half. Distal surface bears an annulus, otherwise laevigate.

Figure 8A
Description. Amb circular to subcircular. Laesurae straight, accompanied by narrow labra along their length, extending from one-half to the whole amb radius. Equatorial thickening of 1-2 lm observed. Proximal face bears interradial muri, 2-3 per sector. Distal surface bears an ornament of microgranulae.
Remarks. The species as originally described is larger than the specimens seen here, with a wider cingulum.
Remarks. The species as originally described is larger than the specimens reported here and these specimens do not show fading muri towards the equator.
Occurrence. Sites 3, 7. Description. Amb circular to subcircular. Laesurae straight, may be accompanied by narrow labra, extending from one-third to the whole amb radius. Equatorial thickening of 1-2 lm observed. Proximal face bears interradial muri, 4-12 per sector and arranged parallel to one another, producing a distinct herringbone pattern. Distal surface laevigate.
Dimensions. 23 (36)  Description. Amb circular to subcircular. Laesurae straight, may be accompanied by narrow labra, extending from one-half to the whole amb radius. Proximal face bears interradial muri, 4-8 per sector. Distal surface bears an ornament of microgranulae.
Remarks. These specimens possess more interradial muri than the species as originally described. Description. Amb circular to subcircular. Laesurae straight, may be accompanied by narrow labra along their length, extending from one-third to three-quarters of the amb radius. Equatorial thickening of 1-5 lm observed. Proximal face bears interradial muri, 3-10 per sector. Distal surface laevigate.
Remarks. The original description for this species does not specify how many interradial muri it should possess and does not mention labra, though they may be visible on one of the figured specimens (pl. 4, fig. 9).

Remarks.
Distinguished from E. annulatus in this study by its shorter trilete rays.
Emphanisporites sp. A Figure 8D Description. Amb circular to subcircular. Laesurae straight, extending to the equator. Equatorial thickening of 2-3 lm observed. Proximal face bears interradial muri, 5-7 per sector and arranged parallel to one another, producing a distinct herringbone pattern. Distal surface bears an ornament of microgranulae.
Remarks. This species has the herringbone muri pattern of E. mcgregorii but possesses a distal ornament.
Genus GEMINOSPORA Balme, 1962 Type species. Geminospora lemurata Balme emend. Playford, 1983.  Remarks. The specimens described here differ from those assigned to G. lemurata primarily in having a larger ornament and sometimes a somewhat thinner exoexine, though still thicker than in Grandispora. The specimens described here are rather smaller than those originally described for the species.
Previous records. Reported from Pragian to upper Famennian strata with an almost worldwide distribution (although absent from Australia). 1953 Hymenozonotriletes argutus Naum. Sp. n.; Naumova, pl. 9, fig. 9 Remarks. This taxon has a complicated taxonomic history. Naumova (1953) described and illustrated two new species, which are very different in structure and morphology, but called both of them Hymenozonotriletes argutus sp. n. These were described as spore number 73 (p. 41) and spore number 169 (p. 67) illustrated as line figures in her plate 4, figure 10 and plate 9, figure 9, respectively. Cramer (1966) (1969) because we are not in a position to propose a formal comb. nov. This is because: (1) comparison with Naumova's material is tentative because she provided only line drawings of spores (as also noted by Cramer 1969); and (2) Naumova's (1953) first described species has priority but we are concerned with the second described species of the same name.
Comparison. The characteristic ornament of this species is similar to that of G. douglastownensis and Ancyrospora nettersheimensis Riegel, 1973 but is far more pronounced and also lacks the bifurcate ornament of Ancyrospora.
Occurrence. Sites 2, 3, 5, 7, 10, 11, 13, 20, 120, 598, G.  Remarks. The species as originally described has folds that accompany the laesurae and may extend to the spore equator, features not seen here. As the few specimens found of this species are poorly preserved this is attributed to a preservational deficiency.
Occurrence. Sites 3, G. Remarks. This species has a very broad original description. As a result a wide variety of forms, with a relatively small, simple ornament but widely ranging exoexine:intexine ratios, can be placed in this species. The specimens in this study have a larger size range than the species as originally described but show a continuous spread.
Remarks. This species is distinguished from G. inculta in this study by its slightly subtriangular shape. A pronounced, though often discontinuous, ring is observed in the middle of the exoexine offlap, potentially representing curvaturae. A similar structure is visible in Breuer & Steemans (2013, fig. 33B).
Occurrence. Sites 2, G. Remarks. This species is distinguished from the other Grandispora species in this study by its characteristic ornament.
Occurrence. Sites 7, 11, 20. Remarks. This species is distinguished from the other Grandispora species in this study by its much larger exoexine:intexine ratio. We use this to refer the specimens provisionally to G. stolidota, though no intexine measurement is given in the original description and the complex, biform sculptural elements could not be seen in these specimens.
Occurrence. Sites 7, G. Remarks. This species is distinguished from the other Grandispora species in this study by its subtriangular shape, large exoexine:intexine ratio and spinose ornament.
Remarks. The specimens seen here meet the specific characteristics in all respects.
Remarks. These specimens are doubtfully assigned to G. muninensis as the presence of labra could not be determined with certainty.
Previous records. Reported from Upper Silurian to middle Tournaisian strata with a worldwide distribution. Remarks. The specimens found here have a greater size range than originally described for the species.
Remarks. The specimens found here show a greater range of variation in the length of the trilete rays than the species as originally described, though no meaningful division of the population could be made. This species may form one end member of a morphological series with A. elegans at the other extreme (Owens 1971); see remarks above pertaining to that species for a fuller explanation.
Remarks. The species as originally described is somewhat larger than the specimens found here.
Occurrence. Sites 3, 7, 120, 598. Description. Amb circular to subcircular. Laesurae straight, may be accompanied by narrow labra along their length, extending from onehalf to the whole amb radius. Curvaturae perfectae sometimes observed joining the laesurae tips. Proximal surface bears an apical, subcircular to subtriangular thickened area extending from c. one-third to one-half of the distance to the equator. Distal surface laevigate.
Remarks. Labra accompanying the laesurae are not always mentioned in descriptions of this species, though they have been figured (Breuer 2008;Breuer & Steemans 2013). These specimens bear a resemblance to Concentricosporites sagittarius (Rodr ıguez) Rodr ıguez, 1983, though they have no obvious crassitude and a diffuse outer margin to their thickened area, as opposed to the hard edge seen in C. sagittarius.
Remarks. The specimens found here meet the specific description for R. rotundus in all respects. The species is sometimes considered as part of a continuous morphological series with R. goensis and R. triangulatus (McGregor 1973).
Remarks. These specimens meet the specific description for R. semizonalis in all respects.
Previous records. An often-reported taxon reported from upper Pragian to lower Famennian strata from North America, China and Europe.  Remarks. The specimens found here meet the specific description for R. triangulatus in all respects.
Previous records. Reported from Devonian, especially Lower to Middle Devonian, strata and with a worldwide distribution (Breuer & Steemans 2013).

Figure 11G
Description. Amb circular to subcircular. Laesurae straight, extending to the equator. Curvaturae observed at the tips of the laesurae. An area of thinner exine surrounds the laesurae with concave sides and rounded ends, extending three-quarters of the distance to the equator. Proximal surface laevigate. Distal surface bears an ornament of microgranulae.
Remarks. This simple species of Retusotriletes lacks diagnostic characters that enable identification with previously described species of this genus. No established species could be found with a similar thinned exinal structure.
Retusotriletes spp. Figure 11H-I Description. Various forms referable to Retusotriletes but lacking diagnostic characters at the species level.
Remarks. Some specimens found here may have a rather coarser ornament than previously described for the species but no meaningful distinction or alternative identification was forthcoming. The species can bear a resemblance to Apiculiretusispora brandtii, though it has a completely detached exoexine. The taxonomic history and validity of this species is complex and disputed, indeed Troth et al. (2011) described it as invalid, choosing to synonymize it with G. lemurata. Despite this, the species is widely used in biostratigraphy, including by Breuer & Steemans (2013) who reported it as occurring across a wide stratigraphical range encompassing that of the sediments studied here. We have chosen to retain this species in light of this wide usage.
Occurrence. Sites 2, 3,7,10,11,13,19,20,22,120 Description. Amb circular to subcircular. Laesurae often obscured, may extend to the equator. Equatorial thickening of 3-5 lm observed. Proximal surface laevigate. Distal surface bears a pattern of convolute ridges around 2-3 lm wide and high, separated by 1 lm at most. Ridges occasionally topped by small spines around 1 lm in size. A separated exoexinal layer may sometimes be seen, though poor preservation precludes a more confident identification of this structure.
Remarks. The assignation of these specimens to S. praetervisus is difficult due to their poor preservation and rare preservation of exine remnants. The distal ornament and size range is the basis of their identification.
Description. Amb circular to subcircular. Laesurae straight, may be accompanied by narrow labra along their length, extending from two-thirds to the whole amb radius. Equatorial thickening of 1-2 lm observed. Proximal surface laevigate. Distal surface bears an ornament of microgranulae and larger granulae up to 7 lm high, usually narrower than they are high, with rounded or flat tops. Some of these flattened tips develop into a short bifurcation.
Remarks. The species as originally described often has somewhat wider ornament than the specimens described here, usually being as wide as it is high, though with considerable variation in size.
Remarks. The exoexine of the specimens seen here is sometimes detached, though this is interpreted as due to damage. A comparison can be drawn between these specimens and Acinosporites tristratus Breuer & Steemans 2013, but these specimens have a primarily granule-based ornament rather than consisting primarily of ridges as in Acinosporites. It should be noted that biform ornament is not mentioned in Clayton & Graham's original description of the species, though it does appear in the specimens assigned to this species by McGregor & Camfield (1982) and in the present study. Description. Amb subcircular to subtriangular. Laesurae straight, extending from one-quarter to the whole amb radius. Equatorial thickening of 3-9 lm observed. Proximal surface laevigate. Distal surface bears a 4-7 lm wide subcircular to subtriangular annulus, positioned around one-half to ⅗ of the distance to the equator, otherwise laevigate. A thin flange is present at the equatorial margin, extending by one-quarter to one-half the radius of the main body of the spore interradially but narrowing opposite the trilete rays so as to disappear entirely.
Remarks. The equatorial thickening of the specimens seen here can be somewhat larger than originally described. Some specimens seem to possess a very fine, irregular distal ornament, though this is interpreted as a preservational effect.
Occurrence. Sites 3, 7, 13, 120, 598, G. Description. Amb subtriangular. Laesurae straight, extending to the equator. Equatorial thickening of 5-10 lm observed. Surface laevigate. A thin flange is present at the equatorial margin, extending by one-quarter to one-third the radius of the main body of the spore interradially but narrowing opposite the trilete rays.
Remarks. This species lacks the annulus of Z. armillatus and its equatorial flange does not disappear entirely opposite the trilete rays. Only a single specimen was found.

OTHER FORMS
Spore type A Figure 13D-E, G-H Description. Amb ovoid. Laesurae observed, in an arrangement reminiscent of a monolete mark with ends that bifurcate around one-third of the distance to the equator. These secondary branches terminate a little over halfway between their point of branching and the equator. The entire laesural structure is surrounded by narrow labra. A structure at the perimeter of the spore may be an equatorial thickening of 2-6 lm thickness or a separated exoexine. Surface laevigate.
Remarks. This taxon bears some resemblance to Gneudnaspora divellomedia (Chibrikova) Balme, 1988. However, the specimens seen here show a pronounced regularity in their monolete-like/trilete-like haptotypic structure, unlike the inherently irregular scars seen in cryptospores like G. divellomedia.

DESCRIPTION OF THE SPORE ASSEMBLAGE
The spore assemblages comprise 68 spore taxa (55 identified species belonging to 27 genera) of which 10 taxa are known only from this assemblage. Count data is presented in Askew & Wellman (2018). The spore assemblages recovered from the various sites do not differ significantly in their taxonomic composition and can be considered as essentially representing a single spore assemblage. The ranges of some important taxa within the Playa del Tranqueru section are displayed in Figure 14 (the remaining short sections are not sufficiently stratigraphically controlled to be correlated with this long section). The spore assemblage is not particularly variable, as is evident from analysis of the three sections of significant length measured through the formations. These sections include almost all taxa found in this study, with only Dibolisporites sp. A, Retusotriletes cf. microgranulatus? and Zonotriletes simplicissimus being absent. None of these taxa are stratigraphically important, as they are either not reported from elsewhere or are known to occur across the stratigraphical range of the formations (Breuer & Steemans 2013). The three long sections exhibit few taxon inceptions, with the vast majority occurring at the base of the sections. Those taxa that do appear higher up the sections rarely become significant components of the assemblage, with the exceptions of Devonomonoletes cf. sp. 1 and three species of Emphanisporites in the San Pedro de Nora section, and none are considered to be of stratigraphical importance. The spore assemblage is dominated by laevigate crassitate spores that can be accommodated in the genus Ambitisporites spp. (20% of the assemblage), similar simple trilete spores with a distal ornament (31%), retusoid forms including Apiculiretusispora (17%) and Emphanisporites (15%). The taxon Grandispora cf. inculta is also important, as it constitutes a significant component of the spore assemblage (8%).
It seems very unlikely that reworking or sorting of spores has seriously skewed spore assemblage composition. The underlying Upper Silurian to Lower Devonian San Pedro Formation and La Vid Group spore assemblages (Cramer 1966) contain many distinctive taxa that are not found in the Middle Devonian spore assemblage, suggesting there was no large scale reworking of taxa. The fact that the range of spore sizes in the Middle Devonian spore assemblage is large, including very small simple spores as well as very large, complex spores such as Grandispora spp., suggests that size sorting is limited.

COMPARISON WITH PREVIOUSLY DESCRIBED SPORE ASSEMBLAGES: BIOSTRATIGRAPHICAL AND PALAEOPHYTOGEOGRAPHICAL IMPLICATIONS
The distinctive biostratigraphical marker species Geminospora lemurata occurs in the lowest sample of the two incomplete sections and in the third-lowest sample of the nearly complete Playa del Tranqueru section, although the two lower samples are very poor, with very few preserved palynomorphs of any kind (see Fig. 14 and Askew & Wellman (2018, charts 1-3) for details). Based on the presence of G. lemurata, and in the absence of younger species, the assemblage described here can be correlated with the following spore zones erected for the Old Red Sandstone continent (Euramerica): the lemurata-magnificus Spore Assemblage Biozone of Richardson & McGregor (1986) and the AD Oppel Zone (Lem Interval Zone) of Streel et al. (1987). Both of these have been attributed an early to ?middle Givetian age. Beyond Euramerica there are few spore zonation schemes, although Breuer & Steemans (2013) developed a scheme for northern Gondwana based on material from Arabia and North Africa. In their scheme, G. lemurata appears at the base of the lemuratalangii Assemblage Zone (Geminospora lemurata Interval Zone) which is considered to be of early (but not earliest) Givetian age (ranging from c. 385. 5-387.7 Ma;Loboziak & Melo 2002;Marshall et al. 2007;Turnau & Narkiewicz 2011;Becker et al. 2012).
Previous work on conodonts recovered from the limestones that sandwich these formations provide a tight age bracket for their deposition: middle Eifelian to early Givetian. This represents a period of between 386-387 and 391-392 Ma, representing between 4 and 6 million years (based on Becker et al. 2012), depending on which part of each conodont zone is represented. Correlation with the lemurata-langii Assemblage Zone (Geminospora lemurata Interval Zone) would suggest that the strata yielding the spore assemblages were deposited in the upper part of this range in a period of only a little over 1.5 million years between 386 and 387.7 Ma, represented diagrammatically in Figure 15.
The scheme of Richardson & McGregor (1986) reports taxa other than the nominal ones that are considered characteristic of the lemurata-magnificus Zone, many of which are also present in the Spanish assemblage, namely Acinosporites acanthomammillatus, Apiculiretusispora brandtii, Emphanisporites annulatus, Grandispora protea, G. velata, G. inculta and Verrucosisporites scurrus. However, the Spanish assemblage also includes other taxa, such as Ambitisporites avitus, A. plicata, Dictyotriletes gorgoneus, E. micrornatus, and G. douglastownensis, which are considered to have gone extinct much earlier in Euramerica according to Richardson & McGregor (1986). The absence of Contagisporites optivus and Samarisporites triangulatus, the nominal species of the overlying optivus-triangulatus Spore Assemblage Biozone of Richardson & McGregor (1986), would seem to preclude the Spanish assemblage from belonging to this zone.
The AD Oppel Zone of Streel et al. (1987) is based on the work of Riegel (1975) from Eifel, summarized by Riegel (1982). This summary reports A. acanthomammillatus, A. lindlarensis, A. brandtii, E. annulatus, E. rotatus and G. velata as occurring on the Eifelian-Givetian boundary. However, Brochotriletes cf. foveolatus, also present in the Spanish assemblage, is reported as becoming extinct earlier, in the upper Emsian. The occurrence in the Spanish assemblage of A. acanthomammillatus, E. annulatus, G. protea, G. velata and Rhabdosporites minutus support this assemblage's placement in the AD Oppel Zone (Lem Interval Zone) of Streel et al. (1987). The absence of the distinctive and widespread spore Samarisporites triangulatus, the nominal species of the overlying TA Oppel Zone of Streel et al. (1987), would seem to preclude the Spanish assemblage from belonging to this zone. It is clear from the above observations that the Spanish assemblage contains a number of taxa that appear either earlier or later than expected based on their occurrence on the continents of Euramerica or Gondwana. Furthermore, it is evident that certain taxa that often dominate coeval spore assemblages from Euramerica and/or Gondwana are absent from Iberia. The absence of spores with grapnel-tipped processes (Ancyrospora spp.) and Rhabdosporites langii is particularly noteworthy.
Numerous species of Ancyrospora are present and often dominate Middle Devonian spore assemblages from Euramerica (Richardson & McGregor 1986). However, it is interesting that Breuer & Steemans (2013) do not report Ancyrospora spp. from the lemurata-langii Assemblage Zone of northern Gondwana. Cramer (1966) reports Ancyrospora spp. as only appearing in the Spanish sequences after the 'Emphanisporites annulatus suite' he described from the Naranco Formation. Ancyrospora is considered to be produced by lycopsid plants based on analysis of spore wall ultrastructure (Wellman 2002), but it is not entirely clear if the plants were homosporous and/or heterosporous or whether they were trees, shrubs or herbs.
Likewise, Rhabdosporites langii often dominates Middle Devonian spore assemblages from Euramerica (Richardson & McGregor 1986) but this taxon is also common in spore assemblages from the lemurata-langii Assemblage Zone of northern Gondwana (Breuer & Steemans 2013). Its presence both north and south of Iberia makes its absence all the more puzzling. R. langii was produced by aneurophytalean progymnosperms (reviewed in Wellman 2009). Interestingly, Marshall (1996) proposed an evolutionary hypothesis whereby the microspore (Geminospora lemurata) and megaspore (Contagisporites optivus) of heterosporous archaeopteridalean progymnosperms evolve from Rhabdosporites spores, indicating a close phylogenetic relationship between these taxa.
The endemic nature of the Spanish assemblage makes biostratigraphical age assessment based on spore zonations/ranges erected elsewhere problematic. Taxa may not be appearing at the same time as elsewhere, preventing accurate correlation with other assemblages. However, a number of factors lend credence to the age determination suggested herein. Implying that G. lemurata appears earlier in Spain is contrary to its very well defined inception in the lowermost Givetian around the world (Loboziak et al. 1991). In addition, the conodont data indicate deposition of the formations had ceased before the inception of S. triangulatus (Fig. 15), precluding the possibility of the Spanish spore assemblage being much younger than is stated here.
In conclusion, the flora of the isolated islands of Iberia has various interesting features. There are clear differences with the spore floras from both the large continent to the north-west (Euramerica) and the large continent to the south-east (Gondwana) with: (1) some taxa endemic to Iberia; (2) taxa that are present in Euramerica and Gondwana (including some extremely common forms) entirely absent; (3) some taxa persisting for longer in Iberia than elsewhere; and (4) some taxa possibly appearing earlier in Iberia than elsewhere.

GEOLOGICAL AND PALAEOENVIRONMENTAL INTERPRETATION
The new biostratigraphical information provided by the spore assemblages has implications for the timing of deposition of the Naranco, Huergas and Gustalapiedra formations. These formations are essentially divided into a lower part, dominated by thick sandstone units, and an upper part, dominated by siltstone/sandstone intercalations. Unfortunately, good palynomorph assemblages were not recovered from the lower sandy part. They are, however, abundant in siltstones throughout the upper part. Interestingly, all of the upper part belongs to a single spore biozone of early Givetian age. In conjunction with the conodont evidence, this indicates the upper part of the formation probably represents a period estimated to be a little over 1.5 million years (c. 386-387.7 Ma) with the lower, sandier part accumulated within a time frame of no more than 3-4.5 million years (Fig. 15).
The deposits of the Naranco and Huergas formations are interpreted as representing large deltas prograding from the land across the marine shelf. It seems likely that this onset of clastic sedimentation terminated the previous carbonate deposition, which only resumed later when the clastic source ceased. The thick sandstone units of the lower part probably represent periods of rapid deposition with numerous disconformities in a nearshore shallow environment. However, at no time was there a long enough break in clastic deposition for significant carbonate deposition to resume. The mixed sandstone/siltstone of the upper part probably represents further offshore, deeper facies with a combination of terrigenous sands and more oceanic siltstone deposits, including euxinic black shales. The Palencia sequences consist largely of calcareous siltstones and are likely to represent offshore environments far from land, although there is an interesting interruption where a thin (<10 m) sandstone known as the Man Member is deposited (see below).
Detailed results of palynofacies analysis will be reported in a subsequent publication. However, a cursory scan of the results of these analyses indicates that the proportion of land-derived spores within palynomorph assemblages clearly decreases offshore across the shelf as terrigenous input diminishes.

THE KA C AK EVENT
The nature of the Ka c ak Event is hotly debated with very little consensus in terms of its timing, duration, causes and effects (Garc ıa-Alcalde 1998; Becker et al. 2016). It is generally considered to have occurred at or near to the Eifelian-Givetian boundary. Some authors suggest a relatively short time interval with a single marine anoxic event, whilst others prefer a more extended time interval possibly with numerous anoxic events. Recently Becker et al. (2016) suggested that the Ka c ak Event was a thirdorder global event with globally elevated extinction rates (at the lower taxonomic level of species and genera) within many taxonomic clades and several ecosystems. They illustrated the event as being more-or-less confined to the lower ensensis conodont zone and pulsed with a lower and upper event.
The Ka c ak Event has previously been identified in northern Spain (House 1996;Garc ıa-Alcalde 1998;Garc ıa-Alcalde et al. 2002). It has been reported to be represented by a thin (1 m) black siltstone unit in the Naranco Formation of Asturias (House 1996), a much thicker unit of euxinic black siltstones in the Huergas According to Marshall et al. (2007), the series of events comprising the Ka c ak took place from the uppermost Tortodus kockelianus conodont zone to the very lowermost Polygnathus hemiansatus zone (latest Eifelian to earliest Givetian). Interestingly this is considered to be just before the inception of Geminospora lemurata (Troth et al. 2011;Marshall 2016). In this study, spore assemblages were recovered from strata above this level. These belong with the lemurata-langii Assemblage Zone (lemurata Interval Zone), indicating an age of early (but not earliest) Givetian. This places the spore-bearing strata above the Eifelian-Givetian boundary and therefore somewhat later than the Ka c ak Event if it is considered to be of only very short duration. Instead the lower, palynologically barren deposits must span the event. Of course, these observations may simply reflect incorrect application or inaccuracies/coarseness of the spore biozonation scheme or misplacement of the Ka c ak Event in the Spanish sequences. Alternatively, it may be that the Ka c ak Event was a drawn-out affair rather than a discrete event of short duration.
Considered as a whole, it is clear that the deposits of the Naranco, Huergas and Gustalapiedra formations represent a relatively long period of clastic sedimentation that interrupted background carbonate sedimentation. Presumably, this was the result of increased terrigenous (predominantly sandy) input from the land. Either the Ka c ak Event is a rapid, discrete event that lies somewhere within this sequence, or it can be considered to be a more prolonged, possibly episodic, event representing a period of environmental perturbation that is manifested in increased terrigenous input from the land. A possible cause for the Ka c ak Event, outlined by Marshall et al. (2007), based on work in the Orcadian Basin of Scotland, was a switch to an increasingly monsoonal climate due to increased insolation. This would have led to increased freshwater runoff from the land, which not only increased terrigenous input, but may also have resulted in ocean stratification and the spread of hypoxia. The Spanish sequences clearly attest to a significant period of increased terrigenous runoff that may indeed have resulted from a switch to monsoonal climates. Perhaps the Ka c ak Event as a whole represents a prolonged climatic event with the black shales reported elsewhere simply representing discrete horizons/locations when oceanic anoxia occurred.
Acknowledgements. AJA was funded by a NERC studentship through the ACCE (Adapting to the Challenges of a Changing Environment) Doctoral Training Partnership (Grant no. X/ 008677-12-1). We thank Drs P. Breuer and P. Steemans and two anonymous reviewers for reviewing earlier versions of this manuscript.