The Late Devonian and Early Mississippian are probably the less known subsystems of the Late Palaeozoic in Mexico. Therefore, we propose in this article two groups of data: (a) the first concerns the microfauna of these ages in San Salvador Patlanoaya; (b) the other one is a reappraisal of the same period in Mexico. Both sets of data allow a palaeogeographic and geodynamic reconstruction of the Mexican plates at the Devonian/Carboniferous boundary, i.e. during the Strunian, an informal and controversial stage or substage, which corresponds moreover for us to the assembly of a first Pangea [22,25], particularly well illustrated in Mexico with the Acatlán Complex.

San Salvador Patlanoaya (Fig. 1) is an important section for the Palaeozoic of Mexico [9,23,24,26]. The base of this Formation was first dated as Osagean [26], but this age is in fact Kinderhookian, according to the biostratigraphic criteria of North-American brachiopods Rugauris and Rhytiophora. We discovered, under the Early Mississippian base [sensu 26] of the Patlanoaya Formation, a 100 m-thick sequence (Fig. 2), which was consequently attributed by us to a questionably Late Devonian [9,24]. Lithologically, this sequence is dominated by very fine sandstones with calcareous cements. Fossiliferous beds are very rare. Some wackestones, with ferruginous oolites (Fig. 3, 10), contain questionable fossils. Bioclastic marly wackestones/packstones, rich in ostracods, yield sporadically the studied microfauna/microflora (Fig. 3, 1–9): Petschoria (?) aff. antiqua Berchenko [2], Kettnerammina grandis Chuvashov [6], K. woodi (Berchenko) [2], Kamaena aff. delicata Antropov sensu Berchenko [2], Tolypammina (?) bransoni Conkin, Conkin and Canis [7] (Fig. 3, 5 and 8).

This assemblage, which is devoid of dasycladacean algae, corresponds probably to an outer ramp environment (−50 to −200 m deep) similar to that described in the Late Devonian Sonyea Group of New York [18] or in the deep carbonate deposits of Germany [19].

Petschoria (?) aff. antiqua [2] is relatively abundant (Fig. 3, 1, 2, 4–9). Generically related to the atypical Petschoria of the Ukrainian Late Tournaisian [2], this taxon differs very much of the true Petschoria of the Late Carboniferous (as illustrated, for example, by [14]). This species is probably new and has a length and a diameter circa three times more than P. antiqua. Although the internal microstructure looks like a recrystallised Petschoria, these microfossils have also some resemblances to the coeval, large ‘Tolypammina’ cyclops Gutschick and Treckman [11].

The other pseudo-algae Moravamminids seem to be more characteristic of the Strunian, and of its equivalents of Ukraine: Ct1a₁ and Ct1a₂ [2] (Fig. 4): Kettnerammina grandis [6] (Fig. 3, 3), K. woodi [2], Kamaena aff. delicata [2], and are identical to taxa described in Ukraine and Urals [2,6].

A true foraminifer Tolypammina (?) bransoni [7] is characterised by its small size and its attachment type. It is described from the Chouteau Limestone (Late Kinderhookian), but similar forms, attached on conodonts, are known in the Devonian from Europe [19]. The assignment of this foraminifer to the genus Tolypammina remains questionable, due to its wall being apparently silicified and not originally agglutinated.

In Sonora occur several sequences of Late Devonian age, which rest discordantly upon Cambrian or Ordovician deposits. The Murciélagos Formation was already re-dated by us [24] from material of Brunner [3,4], which is of Late Frasnian age, with Proninella ex gr. tamarae Reitlinger, Parathurammina ex gr. paracushmani Reitlinger, Amphipora sp., and of the Famennian with the charophyte Chovanella sp. [3, (pl. 5, figs 1–6)] and the conodont Polygnathus communis Branson and Mehl [4, (p. 84)]. The overlying Represo Formation begins in the Strunian with Laxoendothyra concavocamerata ‘var. 1’ (Conil and Lys) and Laxoendothyra parakosvensis imminuta (Conil and Lys) [24]. Similarly, Mamet [21, (p. 54)] has confirmed in Sonora the Late Frasnian, identified the Famennian with Serrisinella sainsii (Mamet and Roux) and primitive Tournayellidae, and proposed a ‘Late Famennian–Kinderhookian’ age for an assemblage with Paracaligelloides sp., Latiendothyra sp. and Septaglomospiranella sp., which is very probably Strunian in age, because of the abundance of Paracaligelloides in the Strunian parastratotype of the ‘Carrière du Parc’ in France (material B. Mistiaen, unpublished), and in Russia [6].

Other probable Late Devonian Formations summarised in various works [10,12,17] exist in Sonora: Martin Formation and Caliza Cristalina Inferior Formation, and in Chihuahua [17]: lower Monillas, Canutillos and La Percha. All these units must be accurately re-studied, especially the La Percha Formation, apparently very similar in lithology, to the Latest Devonian/Earliest Mississippian member of the San Salvador Patlanoaya Formation (Fig. 2). The deposits attributed to Devonian cherts in Tamaulipas are in fact Pennsylvanian in age and rhyolitic [20].

At the end of the Devonian, many plates collide: Ligerian and Brittany phases in Europe, re-assembly of the central Iapetus [15], the Antler event in western USA [5], and in Mexico there is the assembly of Acatlán Complex [16]. Consequently, a first Pangea may have occurred [22,25]. Nevertheless, an epicontinental seaway remains open in Sonora, especially in the Bisani area (Murciélagos Formation and equivalents [10,12,17]). Concomitantly, the Antler Ocean is probably closing and Rheic Ocean is probably opening.

The Latest Devonian/Earliest Mississippian (i.e., the ‘pre-7 zone’ of [1]) is the beginning of the subsidence of the Pedregosa Basin (New Mexico) [1]. This rifting is also responsible for the opening of the San Salvador Patlanoaya basin. The three basins, Pedregosa, Bisani and Patlanoaya, are possibly linked, due to the occurrence of deposits of the Devonian–Mississippian boundary, absent of the rest of Mexico. These three basins may correspond to a unique rift (Fig. 5). The latter is probably separated of the yet sutured Antler Ocean, but begins its connection with the Rheic. The faunal communications in this case are to the east with the Rheic, extending up to Ukraine, and the Antler does not extend through Alaska and Arctic Canada with the Urals (both ways of communication are theoretically possible). The connection with the eastern Rheic (i.e. western Europa and northern Africa cratons) is evident in the Osagean (late Early Mississippian), as indicated by the identification of Waulsortian mud mounds in Oaxaca [8,24]. The radiolaria-bearing series of San Salvador Patlanoaya (number 5 in Fig. 1) [9,23,24] correspond probably to another phase of the spreading of the Rheic, and to the acquirement of true oceanic environments.

A Latest Devonian/Earliest Mississippian microfauna discovered in San Salvador Patlanoaya (Puebla, Mexico) contains some microproblematical carbonate micro-organisms and rare encrusting foraminifers: Petschoria (?) aff. antiqua, Kettnerammina grandis, K. woodi, Kamaena aff. delicata and Tolypammina (?) bransoni. This poor assemblage is questionably controlled by two factors: (a) an outer ramp environment in which the foraminifers are rare; (b) the restriction of the seas existing in North America during the Latest Devonian/Earliest Mississippian. The migration of fauna indicates probably a junction between the rift of North America and the opening Rheic Ocean. Many Tournayellids, Endothyrids and Issinellidss migrate during this period: Serrisinella [21], Laxoendothyra [24], Paracaligelloides, Latiendothyra, Septaglomospiranella [21]. No Quasiendothyrids are cited, although they occur in Alaska [13]. This indicates that the migrations along the shelf of the Antler Ocean are impossible during this period.

The Pedregosa Basin, the Bisani Basin (Sonora) and San Salvador Patlanoaya have many stratigraphic similarities. Consequently, the San Salvador Patlanoaya was probably located more to the northwest at this period (Fig. 5).

During the Strunian, the rift Pedregosa/Sonora/San Salvador Patlanoaya constitutes an extremity of the Rheic Ocean. This is probably true until the Late Tournaisian (Osagean), because Waulsortian mud-mounds are well developed in Oaxaca [8,24].