Trace fossils from the Desejosa Formation ( Schist and Greywacke Complex , Douro Group , NE Portugal ) : new Cambrian age constraints

Trace fossils from a new locality in the Desejosa Formation, Freixo de Espada a Cinta area, northeast Portugal, are described, including Teichichnus rectus and the first Cambrian record from Iberia of the ichnogenus Rosselia , identified as R. cf. socialis . A literary review of the Cambrian record of Rosselia reveals no occurrences older than Cambrian Age 3. The occurrence of Rosselia in the Desejosa Formation therefore adds evidence to that of earlier reports on trilobite remains from the upper part of the Desejosa Formation for a Cambrian age of this unit. Both Rosselia and Teichichnus are zindicative of the Cruziana ichnofacies, which is representative of a shallow-marine depositional environment, consistent with earlier interpretations for the depositional conditions of the upper part of the Desejosa Formation in this sector, and in the equivalent Cambrian units in Spain.

carbonate-bearing Bateiras Formation with the Cloudinarich units (Vidal et al., 1994a) in the Salamanca province (Spain). Trace fossils have been reported as locally common in the Desejosa Formation (e.g. Coke, 2000;Coke et al., 2000) though without published detail to date. At present time, evidence for a Cambrian (or younger) age for the upper part of the Desejosa Formation is based on poorly preserved indeterminable trilobites (Rebelo and Romano, 1986), and its geochemical profile is also consistent with Cambrian age (Dias da Silva et al., 2011;Dias da Silva, 2014).
The first detailed documentation of trace fossils from the Desejosa Formation is here presented on the basis of material collected from a new trace fossil locality (Dias da Silva, 2014). These include Teichichnus rectus and some of the best preserved Cambrian material of Rosselia to date, providing important new age constraints on the Desejosa Formation, and also giving valuable information about the depositional setting of the Desejosa Formation. The regional stratigraphic implications of these new data are also discussed in the CIZ context.

GeoloGICAl SeTTInG
The Iberian branch of the European Variscan Massif has been traditionally divided into several geological zones (Lotze, 1945;Julivert et al., 1972). On historical grounds these divisions were based mainly on stratigraphic features, but tectonic, magmatic and metamorphic characteristics were also considered. The CIZ, in its current definition (Julivert et al., 1972), is the most extensive zone in the Iberian Massif (Fig. 1).
It has been subdivided in several domains in Spain (Pérez-Estaún et al., 2004) and Portugal (Dias et al., 2006). One of its most outstanding stratigraphic features is the presence of a thick (>5km) and widespread basal metasedimentary unit known as the Schist and Greywacke Complex. It generally displays a (very) low metamorphic grade and little deformation, which allows the identification of the original lithologies and primary sedimentary structures thus suggesting a turbiditic nature for much of this sequence. The scarcity of lithological marker beds and fossil content has resulted in a poorly constrained age and an uncertain regional correlation. Several regional lithostratigraphic units have been established being recognized a general upward-shallowing environment from basinal and slope to platform deposits (e.g. Díez Balda, 1986;San José et al., 1990;Vidal et al., 1994a, b;Valladares et al., 2002). The age of the oldest part of the Schist and Graywacke Complex remains poorly known, nevertheless the earlier suggestions of a "Riphean" age (>650Ma) were based on microfossils without age significance, and the entire succession is now thought to be Ediacaran and younger (e.g. Pereira et al., 2012a,b).
The studied region is located next to the Douro River canyon which is the north-eastern border between Spain and Portugal for more than 100km, providing an impressive natural cross-section of the northern CIZ. Recent geological survey in this area has resulted in new stratigraphic and ichnologic data (Fig. 2;Dias da Silva, 2014). In the northern and central part of the canyon, sedimentary features of the older rocks are largely obliterated due to pervasive Variscan deformation which resulted in a high-temperature and low-pressure metamorphism (Escuder Viruete et al., 1994, 2000Dias da Silva, 2014). The southern sector, near Freixo de Espada à Cinta (Portugal), shows more favourable conditions with well-preserved sedimentary structures and trace fossils. The geological structure of this area is mainly composed of cartographic-scale upright folds belonging to the late Variscan regional event (Fig. 2). The Desejosa Formation is extensively exposed in the region and has been affected by only very low metamorphism and gentle deformation during the Variscan Orogeny, which has made it possible to build up a synthetic stratigraphic column of the infra-Ordovician rocks, unveil the tectonic structure, and position the trace fossils described here inside the sedimentary sequence (Fig. 2). The Desejosa Formation was first described by Sousa (1982Sousa ( , 1983b. It typically consists of millimetre-to centimetre-scale alternations of generally parallel-laminated lutites and fine-grained sandstone, resulting in a striped appearance. The most common sedimentary structures are graded bedding and load-structures. More rarely there are greywacke beds up to half a metre thickness. In the study area, the Desejosa Formation consists of a terrigenous sequence more than 300m thick, which displays thin beds of black slates and sandstones, the latter frequently showing bioturbation and sedimentary structures. Some scarce and thin discontinuous beds of calc-silicate rocks and carbonates are referred in Ferreira da Silva and Ribeiro (1994). The uppermost beds (40m) present slump structures (Fig. 2, synthetic column), broken beds and microconglomeratic levels. The trace fossils described here were recently identified in the Freixo de Espada à Cinta area (Fig. 2) in a location that could be estimated at about 100m below the top of Desejosa Formation. The intensive bioturbation suggests a shallow depositional environment as was proposed for the upper part of the Douro group by Sousa (1982).
Concordantly overlying the Desejosa Formation, a new informal unit has been defined and mapped. It was first named Mazouco Formation (Dias da Silva et al., 2011) after a small village which however is located out of the exposure. For this reason we propose to change the name to Montes Ermos unit, after a small hill located well inside the exposure of this lithostratigraphic unit (Dias da Silva, 2014). The Montes Ermos unit, with a thickness greater than 150m, is roofed by the Ordovician basal unconformity. It occupies the nucleus of a late Variscan first order synform also folding the Ordovician and Silurian age units (Moncorvo Synform, Fig. 2). In a small area (around Montes Ermos hill) it contains small cordierite blasts related to Variscan granitic stocks. This unit is made up of a monotonous sequence of grey fine-grained sandstone and siltstone beds ranging from 60 to 120cm. The boundaries between beds are marked by dark millimetre-thick layers displaying iron oxide when weathered. No fossils have been found in this unit.
The above described are the only lithostratigraphic units of the Schist and Graywacke Complex in the eastern portion of the Moncorvo Synform. The São Domingos, Pinhão and Rio Pinhão Formations as defined by Sousa (1984) were not identified here. The upper limit of the Douro Group is the regional angular unconformity known as Toledanic (e.g. Gutiérrez Marco et al., 1990), underlying the Lower Ordovician stratigraphic units of the Vale de Bojas and Marão Formations (Sá et al., 2005;Dias da Silva, 2014).

deSejoSA FoRmATIon TRACe FoSSIlS
A new ichnofossil locality in the Desejosa Formation was found about 3.5km NE of Freixo de Espada à Cinta village (Fig. 2). The trace fossils are observed on weathered bedding-planes (Figs. 3A; 5D-E) on both sides of a small valley with a moderately good geological exposure.

Trace fossil description
In order to obtain 3D views of the ichnofossil, vertically oriented parallel sections were produced with a rock-saw to show the disruption of primary sedimentary bedding by burrowing animals, including oblique sand-filled plugs with evidence for lateral displacement or short plug-shaped burrows. Only two ichnogenera were noteworthy of a detailed description in the sectioned material, namely Rosselia Dahmer, 1937, andTeichichnus Seilacher, 1955 (illustrated trace fossil material is housed in the collections of Area de Paleontología, Universidad de Extremadura, Badajoz).

Ichnogenus Rosselia dahmer, 1937
Rosselia cf. socialis Dahmer, 1937  alternations of dark (muddy) and light (micaceous siltstone) concentric laminae (Fig. 3D, G), or with pronounced siltstone laminae (Fig. 4D). Vertical sections show elongated cone-shaped burrows, some being approximately cylindrical, 10 to 20mm wide and up to 30mm long (representing incomplete specimens). The horizontal section presents ovate or circular shapes (Fig.  5E, F) with a narrow mud or sand-filled central shaft (Fig.  5F). In some cases it was possible to see signs of vertical repetition of the burrowing (Fig. 4G).
Discussion. The assignation of this material to Rosselia is based in the general conical-shaped burrows with concentric laminae that surround a central shaft (sandy or muddy). The relationship between Rosselia and Cylindrichnus has been discussed by numerous authors (e.g. Frey and Howard, 1985;Nara, 1995;Uchman and Krenmayr, 1995), leading to the differentiation of both genera with Rosselia presenting conical to irregularly bulbous concentric lamellae and Cylindrichnus showing vertical to horizontal cylindrical or sub-cylindrical burrows (Frey and Howard, 1985). However, recent studies suggest that the type material of Cylindrichnus concentricus consist of U-shaped or bow-shaped burrows (see Belaústegui and Gibert 2013, for dicussion), calling into question the identification of earlier reports of this ichnogenus. Vertical repetition in Rosselia probably represents examples of equilibrichnia (cf. Nara, 1997). Cretaceous and younger Rosselia typically have a spindle-shaped burrow-form, with conical forms as the result of erosional truncation (Nara, 1995). This is also seen in the type material of Rosselia socialis from the Devonian of Germany (Dahmer, 1937;Schlirf et al., 2002). In both cases the central portion of Rosselia socialis often is sand-filled. Another related form is Asterosoma von Otto, 1854, with concentrically layered horizontal or inclined burrows that radiate from a central point. Asterosoma has been described from the middle Cambrian of Jordan (Hofmann et al., 2012), but no evidence for a radiate arrangement has been observed in the Desejosa Formation material.   d i a s d a S i l v a e t a l .   G e o l o g i c a A c t a , 1 2 ( 2 ) , 1 0 9 -1 2 0 ( 2 0 1 4  The concentric lamellae observed in the Desejosa Formation materials are closely compared with the reports of Rosselia socialis, differing only in the predominantly mud-filled nature of the central shaft in this locality (only one specimen is sand filled; Fig. 5F). We consider that this difference is a reflection of the depositional conditions which created thin intercalations of sand, silt and mud beds, and therefore it does not present ichnotaxonomic significance. Well-developed specimens of Rosselia socialis are spindleshaped, with conical forms being the result of erosional truncation. Because of the lack of evidence for spindleshaped forms in the Desejosa Formation, this material is assigned to Rosselia cf. socialis. The Rosselia in Figure 4 D, F and G shows some similarity with burrows that Mata et al. (2012) interpreted as equilibrichnia behavior of sea anemones. Because the studied material always presents multiple concentric laminations with a mud or sandy central shaft and not a simple cylindrical burrow with a central shaft, the Desejosa material is better assigned to Rosselia than to plug-shaped burrows such as Bergaueria, Conichnus, or Dolopichnus.

The exact process by which the concentric lamination in
Rosselia is formed is not well understood, although Cenozoic specimens, in particular, have been related to the feeding activity of terebellid polychaetes (i.e. Nara, 1995). However, the producers of Cambrian Rosselia are unknown as there exist no fossil evidence for Cambrian terebellids. In a model presented by Nara (1995) the organism was a surface detritus feeder that constructed the burrow walls from surface materials. On the other hand, Goldring (1996) distinguished between two mechanisms for the formation of concentrically laminated burrows, one made in response to the infilling sediment pushed aside by the producer, and a second produced by the addition of successive wall-layers by the living organism. With regard to the Desejosa material we consider it more likely that the burrow formed in response to sediment influx rather than the result of sediment manipulation from feeding activity.
Description. Retrusive spreiten-burrows with horizontal long axis. A specimen examined in a serially sectioned slab (Fig. 5A, B) has a spreite lamina that is 15cm long and 3.5cm high. The base of the lamina is gently inclined with a difference of close to 2cm between the preserved extremities. The spreiten are crudely developed but show a retrusive pattern (Figs. 4C; 5A, B), and spreite lamellae indicate a successive displacement along the horizontal axis (see sketch in Fig. 4).
Discussion. These trace fossils can be assigned to Teichichnus in being wall-like vertically oriented spreiten.
The lamellae are somewhat irregular but the burrows can be compared with material that Pacześna (1996) attributed to Teichichnus rectus from the early and middle Cambrian of Poland. Axial displacement of the spreiten is also present in the type material of Teichichnus rectus material from the late early Cambrian Kussak Formation of Pakistan (Seilacher, 1955, plate 24, fig. 1).

dISCuSSIon
The ichnogenus Rosselia has been typically reported from shallow-marine deposits (Uchman and Krenmayr, 1995) and its presence with Teichichnus is indicative of the Cruziana Ichnofacies, which is representative of a shallowmarine depositional environment between fair-weather and storm wave base, consistent with earlier interpretations of the depositional conditions of the upper part of the Desejosa Formation in this sector (Sousa, 1983a, b).
The age of the Douro Group has been problematic, being referred to as ante-Ordovician and probably late Precambrian to Cambrian in previous studies (Teixeira, 1955;Teixeira et al., 1964;Teixeira, 1981;Sousa, 1982Sousa, , 1983aSousa, , b, 1984Rebelo and Romano, 1986;Ferreira da Silva and Ribeiro, 1994;Coke et al., 1995;Coke and Gutiérrez-Marco, 2001;Pereira et al., 2006). Brachiopods (including Lingulella major) from strata in the Serra do Marão area (Portugal) that were originally attributed to the Desejosa Formation by Teixeira et al. (1964), were considered the oldest skeletal fossils of Portugal (Teixeira, 1981). A more recent study was made by Coke and Gutiérrez-Marco (2001), who failed to recover additional brachiopods from the Desejosa Formation in the Marão area, and remarked on inconsistencies in the locality description provided in the reports of these specimens. Sá et al. (2002) reported on the recovery of some of the type material of L. major, until that time thought to be lost. They concluded that the rock matrix is identical to that of the Ordovician "Quartzitos sem Ferro" formation (Marão Formation in Sá et al., 2003Sá et al., , 2005 cropping out in the same area. The poorly preserved trilobites described by Rebelo and Romano (1986) are then the only fossil evidence for Phanerozoic age for the upper part of Desejosa Formation. This is now corroborated by the new discovery of trace fossils described here, as discussed below.
Trace fossils can be used to differentiate Phanerozoic and pre-Phanerozoic rocks, with some ichnogenera offering an additional age precision. The appearance of trace fossil genera through the Ediacaran-Cambrian boundary and into the Cambrian, led to an ichnofossil-based zonation (e.g. MacNaughton, 2007;Buatois and Mángano, 2011). The ichnotaxa described from the Desejosa Formation can be considered within this zonation. The ichnogenus Teichichnus is an indicator for Cambrian (or later) age (e.g. Crimes, 1987;McIlroy et al., 1998). In Newfoundland the first appearance of Teichichnus belongs to the second Cambrian trace fossil-based zone -Rusophycus avalonensis Zone -and closely follows the appearance of Rusophycus. In the East European Platform, Teichichnus is common in the Lontova horizon, whereas there are no clear reports from the basal Cambrian Rovno horizon. The presence of Teichichnus point to an age of ca. 530Ma, or younger, with no additional precision.
The ichnogenus Rosselia is known from the early Cambrian to the Holocene (e.g. Nara and Haga, 2007), but early Palaeozoic examples are scarce, with the first appearance of this ichnogenus being poorly constrained. In order to assess the potential ichnostratigraphic implications of Rosselia we have undertaken a literature review of Cambrian occurrences of Rosselia and Cylindrichnus (Table I, Electronic Appendix available at www.geologica-acta.com). Direct comparison of the Rosselia found in the Desejosa Formation with the previously reported Cambrian localities is complicated by the general lack of cut material in those studies. Examples of early Cambrian Rosselia are frequently cited specimens from the Kussak Formation in Pakistan (Seilacher, 1955), being assigned to the Cambrian Stage 4. However, this author noted that the conical structure seemed to be connected to a vertical burrow and therefore this material could only be cautiously compared to Rosselia. Although Seilacher (1955) described Rosselia as abundant, we are not aware of any additional documentation of this material. Examples of Rosselia with sand-rich laminae were reported by Desjardins et al. (2010a, b) from the Lake Louise and St. Iran formations of the Gog Group, Alberta, Canada. These affected planar and hummocky cross-stratified sandstones with an upwards-flaring laminae and a cylindrical central shaft. The age of this group is constrained by acritarchs from the lower part of the St. Piran Formation that Downie (1982) compared to those of the Lükati horizon of the East European Platform (Cambrian Stage 3). Belaústegui and Gibert (2009) suggested that Cylindrichnus operosus (Orłowski, 1990) from the lower Cambrian of Poland is better assigned to Rosselia. Also, the Cylindrichnus concentricus reported from the middle Cambrian of Poland by Pacześna (1996) is comparable to the Rosselia described here. There exist no previous reports of Rosselia from the Cambrian of Iberia, although there are several reports of Cylindrichnus especially in the middle Cambrian. The previous, earliest descriptions of Iberian occurrences of Rosselia were made in the upper and middle part of the Early Ordovician (Floian) in the Serra do Brejo Formation in Figueiró dos Vinhos, Portugal (Cooper and Romano, 1982). This material is in a facies sandier than that of Desejosa Formation but it can be compared to the material described here. The Rosselia of Desejosa Formation represents some of the finest examples of Cambrian Rosselia ichnogenus with clearly defined concentric muddy laminae.
The presence of Rosselia cf. socialis in the upper part of the Desejosa Formation combined with the earlier findings of indeterminable trilobites in a similar stratigraphic position, means that this part of the succession cannot be older than Cambrian Age 3 (ca. 522Ma). However, the stratigraphic relationships in the Desejosa Formation between the Rosselia and the trilobite localities are not clear, due to the strongly erosive Toledanic Unconformity which is topping Desejosa Formation in Açoreira village (at 20 km to the west of this ichnofossil site), were Rebelo and Romano (1986) found the trilobite remains. At present, the upper age limit of the Douro Group has to be based on lithostratigraphic correlation to units within the Spanish part of the CIZ. Sousa (1983aSousa ( , b, 1984 suggested that much of the Douro Group can be correlated with the Aldeatejada Formation (Díez Balda, 1986) in the Salamanca area and the Pusa Formation (San José, 1983) in the Toledo Mountains. Similarly, Pereira et al. (2006) approximated the top of the Douro Group to the upper part of the Pusa Formation and parts of the Azorejo Formation (San José et al., 1974), which they positioned close to the transition of the Cordubian-Ovetian regional stages. A somewhat younger age for the upper part of the Pusa Formation was suggested by Jensen et al. (2010), placing it firmly within the Ovetian (Cambrian Stage 3). A circumstantial argument against a higher position of Desejosa Formation is the absence of prominent carbonate beds in the upper part of the Douro Group.

ConCluSIonS
We present the first detailed description of discrete trace fossils from the Desejosa Formation on the basis of material from a new locality in the Freixo de Espada à Cinta area, Portugal. The most reliably identified forms are Teichichnus rectus and Rosselia cf. socialis, being the first report of the ichnogenus Rosselia from the Cambrian of Iberia, and some of the most conspicuous Cambrian examples of this ichnogenus to date.
A review of Cambrian occurrences of Rosselia and other vertical concentrically laminated burrows, show no record older than Cambrian Age 3. Together with the additional evidence of the previously described scarce remnants of unidentified trilobites, this demonstrates that the upper part of the Desejosa Formation is no older than Cambrian Age 3. Also the ichnogenera association (Cruziana Ichnofacies) described in this paper gives evidences of a shallow water depositional environment between fair-weather and storm wave base.
The Desejosa Formation is conformably overlain by a recently proposed lithostratigraphic unit, called Montes Ermos (Fig. 2). This is younger than the Cambrian Age 3 with no further precision. Nevertheless the absence of middle and late Cambrian rocks in the CIZ points to an age towards the later part of the early Cambrian. An upper age constraint for both formations is provided by the unconformably overlying Early Ordovician rocks.

ACKnoWledGmenTS
ÍDS acknowledges the Instituto Geológico y Minero de España (IGME) for a PhD scholarship "Formación de Personal Investigador" and the institutional support provided to the project CGL2010-18905/BTE. SJ acknowledges funding from the Spanish Ministry of Science and Innovation through grants CGL-2008-0473 and CGL-2012-37237 (both co-financed by Fondo Europeo de Desarrollo Regional [FEDER]). The manuscript benefited from the reviews of Masakazu Nara and an anonymous reviewer. This work is a contribution to the IGCP project nº 597 Amalgamation and breakup of Pangaea.