A synthetic overview of Mesozoic magmatism in Uruguay is given by Muzio [34]. The first studies of subvolcanic rocks in the Uruguayan Paraná basin only consider the post-Permian doleritic dykes from the area of Melo [46,47]. At first, the sills were considered as “intrusive basalts” [17]. Then they were assimilated to lava flows in the lateral continuity of those of the Paraná [9] before their intrusive nature was confirmed near the village of Cuaró [36], and later, in other areas within the basin (i.e. Los Novillos, Cerro Pereira and Ramón Trigo; Fig. 1) [7,8]. All the magmatic intrusions (dykes and sills) were formally defined as the Cuaró Formation and considered Triassic in age by Preciozzi et al. [36].

The sills are almost horizontal with tabular shapes but locally top contacts present significant dips. Their thickness is variable but can be estimated up to some tens of meters. The most extended Los Novillos sill has an exposure of ca. 1200 km², followed by the Cuaró sill, with an exposure of ca. 1000 km². Low-grade contact metamorphism has been recognized in the Permian sedimentary host rocks, as well as roof – pendants included in the doleritic mass.

The dykes are not homogeneously distributed. They can be grouped into two main swarms: the central (Melo – Caraguatá) and southwestern swarms (San Gregorio). They are mostly subvertical and show widths ranging between 1 and 20 m. Close to sills, they can exceed 50 m. They cross-cut Permian sediments as well as Cretaceous basaltic lava flows. On aerial photographs, dykes trend from N80° to N150°, but generally close to N120°; there are also N20° trending dykes. A contact metamorphism is described within the walls of some dykes, where silicified sandstones and conglomerates are present. Chilled margins and hydrothermal veining are also present. Most of the dolerites in both dykes and sills present similar microgranular to granular porphyritic textures and the following mineral assemblage: Plagioclase (An_35-50) + Clinopyroxene (Augite) + Olivine ± opaque minerals, confirming historical observations [8,46].

We did not observe direct evidence of multiple injections within the intrusions (no internal chilled margins and/or significant grain-size variations across the width of the intrusions) but a single site 40A gave evidence for composite flow involving coeval magmas with complex mingling structures (rounded enclaves; “cauliflower-shaped contacts”). At this site, the main magma is represented by the classical dolerite and the other occurs as a finer grained dolerite. In addition, none of the sills provides evidence for phenocryst accumulation nor cumulate textures.

Despite their regional structural trends and their spatial and temporal relationship with tholeiitic basalts of the PLIP, few authors accepted to consider them as the feeders of the Paraná basalts. In contrast, they have been considered as independent “intra-cratonic” magmatism and feeder channels of eroded older basaltic lava flows [7]. Féraud et al. [18] were the first to relay the sills and the dykes with the PLIP.

Doleritic bodies are here considered as the magmatic feeders through which basaltic magmas gave rise to the huge amount of the PLIP lava flows. Although some intrusions are possibly older, it is commonly accepted that the bulk magmatism was emplaced in a short time span at ca. 132 Ma at the beginning of the opening of the South Atlantic Ocean [14,19,39]. Yet, there is not enough evidence that dolerites intruded in a single magmatic event. Only two ⁴⁰Ar/³⁹Ar ages are available at 131.6 ± 1.3 Ma on a dyke and 132.1 ± 1.06 Ma on a sill [45]. Most available K/Ar ages are younger and range from 121 ± 6 to 131.8 ± 6.4 Ma [34]. If we interpret them as cooling ages, they become compatible with a main intrusive event at ca.132 Ma followed by a significant period of hydrothermal activity.

Our fieldwork was focused on the western Melo region, where the number of dykes is larger and better exposed than elsewhere. The most conspicuous sills were also studied (i.e. Los Novillos, Cerro Pereira and Cuaró). Geological maps at 1:50,000 are still not available in the study area. The tectonic evolution of the Paraná basin, Uruguay is not well constrained yet.

The dyke swarm structural analysis was made using photo-interpretation analysis. Some dykes are segmented (Fig. 2a). Branching-dykes are common (Fig. 2a–e) as well as angular jogs in relay zones (Fig. 2a,d). Local “en-echelon” dyke distribution was recognized, indicating that sinistral (Fig. 2b) and dextral (Fig. 2d) strike-slip movements, depending on the dyke swarm orientation, occurred during dyking. Others show kinks and local bending (Fig. 2c,e). The dykes are closely linked to sills (Fig. 2b). Although we did not observe directly a dyke transforming into a sill, such feature has been previously described [8]. Most of these structures and the general trend of the dyke arrays suggest a north–south to NNE–SSW finite extension direction, synchronous with dyking.

Mesoscale structural analyses show that the dykes as well as their host rocks are affected by a brittle extensional deformation. Normal fault-planes contain slickensides showing a dip-slip sense of shear (Fig. 3a). They are concentrated in two main families (Fig. 3b). Their directions were observed to be mostly subparallel to the dyke margins. Slickenline contour plot shows mainly two maxima distributions plunging in opposite directions, to the north and to the south (Fig. 3c). The brittle fault was recorded after the cooling of the magma but fracturing was observed to be synchronous with a hydrothermal activity, which is recorded by the presence of hematite, calcite or silica veinlets within both the dykes and their host rocks. Dip-slip slickensides on hydrothermal veins show the synchronicity between hydrothermal fluid circulations and faulting. Hydrothermal activity may locally be recorded by a strong alteration within the doleritic mass, especially along some margins. Along some normal faults, the development of doleritic breccias shows that host rock sediments were fluidized during deformation and re-injected into the solidified magma.

Most slickensides show a quasi north–south finite extension direction during the hydrothermal event. The mean main trend for dykes gives almost the same finite extensional direction, suggesting that the hydrothermalism directly relates to the magmatic event. We conclude that dyking and subsequent hydrothermalism occurred in a bulk north–south finite extensional direction.

Among 55 doleritic outcrops visited, most were not suitable for magnetic studies because of alteration and bad outcrop conditions. However, 13 sampling sites (9 sills and 4 dykes) were selected and 452 specimens were collected (Fig. 1). From these specimens, only 439 specimens were used for the statistics; some of the remaining specimens were used for thin sections and rock magnetic analyses. Sampling was made as close as possible to the margins of the intrusions but most of the cores were taken at a distance of several metres from the contacts (Table 1 gives estimates of the distances of the groups of samples from the contacts). Samples were cored parallel and/or perpendicular to the contacts.

Bulk magnetic susceptibility was measured using a KLY-3 kappabridge (AGICO, Brno) and gave values ranging between 0.38 × 10⁻⁴ and 1.142 × 10⁻¹ SI, with a corrected degree of anisotropy Ṕ indicating an anisotropy between 0.1 and 12% except for two specimens which reach 25% (in Table 1, the mean sampling site values never exceed 10%; the anisotropy percentage is calculated using 100(Ṕ-1)). In both sills and dykes, different fabrics were evidenced by selecting samples showing coherent clustering of the AMS tensor's main axes calculated from bootstrap statistics [44]. We verified afterward if the fabrics do or do not depend on the spatial distribution of the samples in each site.

The sills gave relatively fresh dolerite samples. The shape parameter (T) indicates mostly oblate ellipsoids, the bulk magnetic susceptibility ranges between 3.8 × 10⁻⁴ and 3.6 × 10⁻² SI and the anisotropy is low, between 1.1 and 2.2% (Table 1). AMS tensor's minimum axes (K₃) are mainly subvertical, perpendicular to the intrusive contacts. Except for sites 53 and 54 (Fig. 4a) where a single fabric is recorded, in the seven other sills, one to two additional fabric(s) can be evidenced (Table 1). For some of these additional fabrics, K₃ remains subvertical. In the concerned sites, a switch between subhorizontal K₁ and K_2, most often independent on the spatial distribution of the samples, can be observed (fabrics 2-1 compared to 2-2,17-1 to 17-2, 32-1 to 32-3, 45-1 to 45-2 in Table 1). In the other cases, K₃ is no longer subvertical (fabrics 10-2 compared to 10-1, 32-2 to 32-1, 36-2 to 36-1, 37-2 to 37-1, 45-3 to 45-1 in Table 1 and Fig. 4b,c). These additional fabrics are most often dependent on the spatial distribution of the samples relative to the contacts.

The anisotropy in dykes (up to 10%) is larger than in sills, and T exhibits both prolate and oblate signatures, often predominantly prolate. Three fabric types are evidenced in Table 1: a- K₃ almost perpendicular to the dyke walls and subhorizontal K₁ or K₂ (normal fabrics 33-1, 33-2, 40A-1 and 44; Fig. 4e,f); b- K₁ almost perpendicular to the dyke walls and subvertical or steep K₃ (inverse fabrics 22-1, 40A-3 and 40B-3); c- K₂ almost perpendicular to the dyke walls (intermediate fabrics 22-2, 40A-2, 40A-4, 40B-1 and 40B-2). In site 22, fabrics 22-1 and 22-2 do not depend on the spatial distribution of the samples from the dyke margins. This is also the case in sites 40A and 40B. In addition, in 40A where two different dolerites mingle, the fabrics are observed to be independent from the rock type. Fabrics 33-1 and 33-2 are partly dependent on the spatial distribution of the samples. In the site 33, coring was made in a rectangular zone 2 m wide (3 to 5 m from the northern contact, Table 1) and 6 m long parallel to the contact. Most of the samples of the eastern part of the sampling zone define the fabric 33-1 while most of the samples of the western part define the fabric 33-2 but sometimes, as in other sites, the same core exhibits different fabrics in different samples.

On the bases of available radiometric data [45], previous descriptions [8] and from our own structural (Fig. 2a) and petrological observations, we are confident that most dykes and sills relate to a single main magmatic event. The structural data (e.g. dyke arrays, dyke contact orientations, normal faults and slickensides) show a predominantly NW–SE trend for the dykes (Fig. 1) and a dominant east–west trend for the normal fault planes (Fig. 3b). Normal faults are subparallel to the dyke segments, which are not systematically parallel to the general dyke trend. These structural features and geological observations document a general north–south extension synchronous with dyking and subsequent hydrothermal activity. The presence of micro-cracks and hydrothermal minerals in the dykes indicates that magnetic fabric could secondarily have been influenced by the deformation associated with hydrothermalism. Additional late (post-cooling) faulting may also affect some intrusions.

In sills, magnetic fabrics are mostly oblate with K₃ perpendicular to the margins. This suggests that most fabrics probably relate to their primary (magmatic) petrofabric. These primary fabrics, acquired above solidus temperatures, may directly reflect magma flow directions during magma emplacement [41]. In such a case, subhorizontal K₁ or K₂ are compatible with magma flow (Fig. 4a,b). During the crystallization of the magma, they may also result from crystal settling [35] in a stagnant magma (weak P’, K₁∼K₂) but we have no petrological evidence of cumulative textures, or result from vertical compaction of a magmatic mush under the weight of the sediments above the sills. In addition, such fabrics can also be secondary, that is acquired at sub-solidus temperatures: a vertical K₃ is also consistent with the vertical shortening expected in an extensional setting. In such case, sub-solidus re-equilibrations of some of the magnetic minerals should have occurred within the sills. Objectively, our results on sills are not helpful to discriminate between these different primary and secondary processes (flow, compaction and sub-solidus recrystallization under stress). Nevertheless, thermomagnetic curves within sill 37 (Fig. 6a) show three different magnetic carriers: one is likely magnetite, with a Curie temperature at ∼570 °C, the others, with lower Curie temperatures at ∼520 °C and ∼200 °C, are probably maghemite and undetermined impure phases (goethite? maghemite? magnetite?), respectively. Fabric 37-1 is carried by magnetite and a small amount of maghemite (sample 37-12) while fabric 37-2 (sample 37-4) is carried by magnetite (in lower amount than in sample 37-12), maghemite and a small amount of impure phases (sample 37-4, Fig. 6a). Despite uncertainties on the nature of some of the magnetic phases, the magnetic mineral assemblages are clearly different from one fabric to the other. In the same rocks, these different assemblages are unlikely to have been produced in the same temperature range during cooling. Fabric 37-1, consistent with flow, is likely primary. This leads us to propose a secondary origin for fabric 37-2 (Fig. 4c) but it could also result from late stage crystallization of the magma [43]. Vertical K₁ in fabric 37-2 is inconsistent with the vertical shortening excepted during extensional deformation which unlikely explains the fabric. For this reason, we suspect that the superimposed fabric 37-2 could result from sub-solidus fluid circulations which partly transformed high temperature magnetic phases (magnetite) into lower temperature phases (maghemite).

Magnetic fabrics in dykes are more difficult to interpret than in sills. There is still a debate on the relationships between magnetic fabrics given by AMS and petrofabrics resulting from magma flow [13,43]. Normal or inverse fabrics can result from magma flow [42]. Examples of such fabrics are observed in all the dyke sites but site 33 exhibits two different normal fabrics (Fig. 4e,f) with predominantly prolate shapes (Fig. 5). Fabric 33-1 with a subhorizontal K₁, parallel to the dyke margins, suggests horizontal or vertical magma flow (Fig. 4e). It is possible that fabric 33-2 results from the same flow as examplified in a similar context [10,11] or from theoretical considerations [15,20,25,28]. However, fabric 33-2 with a subvertical K₁, is also consistent with horizontal or vertical flow. From thermodynamics, it is reasonable to assume that for a same magma composition, different magnetic mineral assemblages, carrying different fabrics, result from different events or temperature ranges during cooling. Thus, we used thermomagnetic curves (samples from both fabrics) to examine which of these two superimposed fabrics is more likely related to magma flow. The results from samples 33-16 (fabric 33-1) and 33-2 (fabric 33-2) are presented in Fig. 6b. They show a single magnetic carrier, likely magnetite, evidenced by a Curie temperature at ∼580 °C in fabric 33-1, and two main carriers, maghemite and magnetite in fabric 33-2 evidenced by a susceptibility drop (possibly due to a phase transformation) at ∼450 °C and a Curie temperature at ∼580 °C, respectively. Such a difference in the mineralogical assemblage suggests that only one of these two fabrics is reliable to magma flow. Even though both magnetic assemblages can be magmatic in origin, maghemite can also be of secondary origin and result from recrystallization of primary phases [43]. For this reason, we interpret fabric 33-1 as primary, resulting from magma flow (Fig. 4e). We propose that fabric 33-2 is secondary (or primary but unrelated to magma flow) and superimposes on fabric 33-1. Because maghemite can crystallize in reduced environment, it is likely that subvertical “per ascensum” fluid circulations [1] parallel to K₁ (Fig. 4f) have induced partial and/or total recrystallization. Such a hypothesis is also supported by direct field evidences of hydrothermalism. Fabric 44 compares with fabric 33-2 (Fig. 4d,f) though the outcrops are separated by more than 150 km (Fig. 1). In addition, site 44 is located at about 50 m from the northern contact of a particularly thick dyke. This distance makes it difficult to offer any interpretation of the fabric as a result of magma flow. The exceptional thickness of this dyke can be explained by the proximity (less than 200 m) of a neighbouring sill to which the dyke is probably connected.

Based on previous discussion, we attribute a primary origin to the fabrics in sills which are identified by the sub-number “-1” in Fi (Table 1), though the fabrics which are identified by the sub-number “-2” and which show a permutation between subhorizontal K₁ and K₂ may also be primary. All the other fabrics in sills are likely secondary. In dykes, primary fabrics reliable to magma flow concern fabrics 2-1, 33-1 and 40A-1. If some of the other fabrics possibly relate to magma flow, most of the ones which are superimposed likely result from late stage crystallization of the magma or secondary events (fluid circulations, tectonic deformation). Silva et al. [43] describe exsolution processes enhanced by slow cooling rates. These processes occur in magnetic minerals above and below solidus temperatures of doleritic magmas and affect the magnetic fabrics. They may also be active in our samples, especially because we have sampled thick intrusions and, most often situated several metres from the contacts.

Superimposed fabrics in sills show, from one fabric to another, permutations K₁-K₂ (e.g. site 2 or 17, Table 1) and K₁-K₃ (e.g. site 37, Fig. 4b,c). In dykes, the three permutations K₁-K₂ (e.g. site 33, Fig. 4e,f), K₁-K₃ (e.g. site 40A, fabrics 40A-1/40A-3) and K₂-K₃ (e.g. site 40A, fabrics 40A-1/40A-2) occur. If some of these permutations are explainable when the intensities of the concerned axes of the ellipsoid are approximately equal (and/or when P’ is weak), they become more difficult to understand when the intensities are different in ellipsoids significantly anisotropic. In the latter case, we suspect that secondary processes (recrystallization associated or not with fluid circulations or tectonic deformation) can have been significant.

Among secondary processes involving fluids, it is difficult to discriminate between fluids associated to magmatic events or later circulations. For example, maghemite has been related to meteoric fluid circulations associated with climatic changes [32]. However, maghemite has also been associated to changes in local stress fields [1]. Superimposed fabrics are common in many dyke swarms where interpretation of fabrics is still a matter of debate [1,13,38,43]. Further systematic sampling and analyses should help to interpret these fabrics especially in the Cuaró intrusions where they seem common. Detailed characterizations of the magnetic mineralogy, analysis of the relationships between petrological and magnetic fabrics, anisotropy of anhysteretic remanent magnetization and hysteresis loops taking into account the effects of the grain-size, are required to further interpret the origins of the superimposed fabrics that we have evidenced here in the Cuaró intrusions.