Dyrosauridae de Stefano 1903

Dyrosauridae de Stefano 1903

Hyposaurinae Nopcsa 1928

Hyposaurinae indet.

Type horizon and age: Recovered in a marine sandstone (bed 11, Fig. 1D View Figure 1 ), provisionally associated with a topmost unit of the Mocuio Formation, Maastrichtian (Upper Cretaceous) ( Strganac et al. 2014).

Type locality: Cemitério de Bentiaba   GoogleMaps , Municipality of Moçâmedes, Namibe Province, Angola ( 14°17′05.5′′S; 12°22′34.8′′E).

Material: Posterior portion of the skull including the braincase, right quadratojugal, part of the left ectopterygoid wing, one tooth and bone fragments accessioned as MGUAN-PA548 GoogleMaps ( Fig. 2 View Figure 2 ).

State of preservation: The specimen comprises an articulated posterior part of the skull, a part of the left ectopterygoid, one tooth, and several small bone fragments ( Fig. 2A View Figure 2 ). The tooth is eroded at its base, but is in good condition, unlike the rest of the material. The exposed and covered bone surface was eroded to a similar degree, precluding the identification of some details of the suture lines. Due to the non-preservation of the entire parietal process, only the base of the process is preserved with a cavity in the centre. In posterior view, the supraoccipital/exoccipital have been severely eroded in their central right side. However, it is possible to observe the posterior part of the parietal, the posterior parts of the squamosal, the supraoccipital, the exoccipitals, part of the basioccipital, the quadrates, and part of the quadratojugals (Supporting Information, Figs S1 View Figure 1 and S 2 View Figure 2 ).

Description

Parietal ( Fig. 2B–D View Figure 2 )

The parietal is preserved only by the posterior portion and a part of the interfenestral bar, observable in dorsal view. The poor preservation of the specimen makes the suture lines with the squamosal hard to discern; however, comparison with data in the literature and extant crocodylians allow us to infer that these suture lines are closer to the interfenestral bar than to the lateral margin of the squamosal ( Fig. 2D View Figure 2 ). Both the posterior and the anterior margins of the parietal are strongly eroded, and the posterior one is concave in dorsal view (ch. 90 [1]) ( Fig. 2D View Figure 2 ). The parietal forms part of the posterior margin of the supratemporal fenestrae and creates a thin interfenestral bar dorsally (12-mm wide) (ch. 19 [1], 20 [0], and 25 [2]). The presence of ornamentation on the margins of the supratemporal fenestrae and on the parietal of MGUAN-PA548 cannot be ascertained due to the state of preservation (ch. 21 [1/2] and ch. 93 [3/4]). The posterior margin forms a gently sinuous suture with the supraoccipital and the parietal has a small occipital portion (ch.92 [1] and 96 [0]).

Squamosal ( Fig.2B–E View Figure 2 )

Dorsally, the squamosal is anteriorly extended to form part of the skull roof, constituting the posterior region of the supratemporal arch (= upper temporal bar or postorbital-squamosal bar) ( Fig. 2D View Figure 2 ). The preserved supratemporal arch is situated more ventrally than the interfenestral bar, resulting in the cranial table being more convex than horizontal/flat in posterior view (ch. 159 [2]). The squamosal forms a part of the lateral and posterior margin of the supratemporal fenestrae. The margin is large and rounded. The posterior margins of the infratemporal fenestrae are unornamented. The parieto-squamosal transition is hard to recognize because of the poor state of preservation of the specimen. What can be discerned, is that these limits are not in the posteriormost margin of the supratemporal fenestra, but they are closer to the median axis of the cranium ( Fig. 2D View Figure 2 ). The posterior surface of the squamosal is concave in the region lateral to the occipital tuberosities, with the squamosal extending posteriorly to a lesser extent than the occipital condyle (ch. 105 [1] and 107 [0]). In occipital view, the squamosal is also ventrally extended and the suture line with the left exoccipital can be observed running ventrolaterally, exhibiting a sharp prong (ch.103 [1], 104 [1], 106 [1], and 108 [1]) ( Fig. 2B View Figure 2 ). In occipital view, above the occipital tuberosities, the squamosal and the parietal appear to have a small posterior lamina (also visible in dorsal view) that can be associated with the top of the post-temporal canal ( Fig. 2B, D View Figure 2 ). This feature corresponds to character 94 of Jouve et al. (2021) described as the ‘posterior extension of the parietal’. However, in MGUAN-PA 548, the suture between the squamosal and parietal is unclear due to erosion. To avoid confusion, we will refer to this feature as an extension of the squamosal–parietal region, acknowledging the possibility that, in the studied specimen, the squamosal may also contribute to the extension.

Supraoccipital/exoccipitals ( Fig.2B–D View Figure 2 )

The supraoccipital is the most eroded element of the bones preserved in this specimen, and as such, in occipital view, informative morphological details are very difficult to discern. Although, the suture with the exoccipital is eroded, the parietal suture is visible, slightly concave, and gently sinuous ventrally ( Fig. 2B View Figure 2 ). The supraoccipital presents a large hole (due to the preservation), measuring 60 mm wide and 20 mm high. The occipital tuberosities, even though not entirely preserved, are well-developed, widely separated and overhanging by an extension from the squamosalparietal region (ch. 94 [1] and 120 [2]). However, due to the state of preservation their shape is not clear. The right occipital tuberosity, more developed, is diagonally flattened due to the aforementioned hole on the occipital side. The left tuberosity has a rounded base, but the rest of it is not preserved. It is also unclear whether the occipital tuberosity and the post-temporal fenestra participate in the supraoccipital or not. The paraoccipital processes of the exoccipitals are well preserved, especially the left one. In occipital view ( Fig. 2B View Figure 2 ), the large lateral suture of the left contact with the squamosal can be observed well, while the right one can only be inferred due to erosion (ch. 122 [1]). The paroccipital processes of the exoccipitals are ventrally directed along the quadrate, and they end with rounded and well-developed processes at the same level as the occipital condyle (ch. 119 [1]). The suture between the exoccipital and basioccipital is partially discernible, allowing for an approximate placement on the occipital condyle but the exoccipital largely participates (ch. 118 [0]). In ventral view, the exoccipitals do not bear any crest or ornamentation (ch. 123 [0] and 124 [0]). In occipital view, the foramen magnum is symmetrically flanked by two rounded structures positioned laterally and slightly dorsally ( Fig. 2B View Figure 2 ). These bony projections, referred to here as bosses, measure approximately 15 mm in diameter and exhibit an occipital projection of just under 5 mm.

Basioccipital ( Fig.2B, D View Figure 2 )

The basioccipital is 26 mm long anteroposteriorly, 37 mm wide lateromedially and shows a depression on the ventral surface (ch. 129 [1]). In occipital view, MGUAN-PA548 presents a ‘V’-shaped morphology (tipping ventrally), and the occipital condyle is twice as wide as it is high (ch. 2 [2]). The posterior surface of the main body of the basioccipital below the condyle is not vertical; it is inclined, faces posteroventrally, and is visible in posterior view (ch. 132 [1]).

Basisphenoid ( Fig.2C View Figure 2 )

Only the posteriormost part of the basisphenoid is preserved. The medial 7 ustachian foramen is present, well visible in ventral view but not in occipital view (ch. 127[0]). Only the contact with the basioccipital has been preserved. A small fragment of the basisphenoid is also preserved and is placed to the right of the central block of the basisphenoid.

Quadrate ( Fig.2B–E View Figure 2 )

The quadrate is very long, almost flattened, directed posteroventrally (very verticalized), and sutured completely laterally with the squamosal (ch. 135 [2], 136 [0], 140 [1], and 141 [2]). Both the right and left quadrates are well-preserved, their suture lines with the exoccipitals are straight, and their condyles extend further ventrally than the occipital condyle (ch. 139 [1]). Contact with the squamosal is made along the posterior wall of the supratemporal fenestra, ventrally and laterally to the presumed location of the temporal canal. The quadrate has no ornamentation, and its entire surface is smooth. Only a small portion of the left quadratojugal is preserved and follows the same orientation as the quadrate. Lastly, the posterior margin of the infratemporal fenestra is posterior to the supratemporal fenestra margin (ch. 29 [1]).

Foramen magnum ( Fig.2D View Figure 2 )

Although its limits are severely eroded, the foramen magnum is located above the occipital condyle with a similar diameter as the latter in occipital view. In anterior view, the diameter of the foramen is reduced by half, forming a conical hole. In ventral view, it is possible to distinguish different foramina, here identified as: the foramen of the anterior bundle of the hypoglossal nerve ( Fig. 2D View Figure 2 —XII), the foramen of the posterior bundle of the hypoglossal nerve ( Fig. 2D View Figure 2 —XII), the cranioquadrate canal foramen ( Fig. 2D View Figure 2 —cqc), the foramen of the glossopharyngeal nerve ( Fig.2D View Figure 2 — IX), the foramen of the vagus nerve ( Fig. 2D—X View Figure 2 ), the foramen of the accessory nerve ( Fig. 2D View Figure 2 —XI) and the median 8 ustachian foramen ( Fig. 2D View Figure 2 —mef). All these foramina have been named and identified following the nomenclature of Erb and Turner (2021). The first four foramina border the basisphenoid laterally (ch. 133 [0]). In ventral view, the medial 8 ustachian foramen is situated anterior to the posterior margin of the supratemporal fenestra (ch. 134 [1]). In the same view, it is possible to identify the cranioquadrate canals, which are hidden in occipital view by the exoccipital (ch. 131 [1]). These canals are positioned on the ventral side of the quadrate, just at the beginning of the exoccipital tuberosities. The temporal canals are filled with sediment and are indistinguishable in anterior view; however, an iron spot on the left side may indicate the position of the left temporal canal.

Isolated material ( Fig.2F, G View Figure 2 )

Some elements have been excavated from the same block of the skull and are interpreted as part of the same specimen.A fragment of the left ectopterygoid has been recognized ( Fig. 2E View Figure 2 ). The fragment is straight anteroposteriorly and slightly concave dorsoventrally. The fragment thickens posteriorly and forms a process that separates it into two distinct parts. The dorsal part bears a 5-mm diameter slit on the left lateral side that gradually fades from the posterior half of the fragment. In addition, the dorsal side has a ridge parallel to the previous slit starting at the posterior half of the fossil. One isolated tooth was collected ( Fig. 2F View Figure 2 ), measuring 42 mm in crown height, 10 mm in crown base mesiodistal length, and 5 mm in crown base labiolingual width. It exhibits an elliptical cross-section, featuring straight carinae devoid of serrations, with the enamel displaying a dozen pronounced ridges, in particular on the lingual side of the tooth (ch. 196 [1], 197 [0], and 198 [0]).

Phylogenetic analysis

Maximum parsimony analysis

The consensus was obtained from 144 trees proposing a tree length of 821.338 steps with a consistency index of 0.385 and a retention index of 0.743 ( Fig. 3 View Figure 3 ). The topology of the cladogram obtained is identical to the one of Jouve et al. (2021) with the well-supported node Tethysuchia (absolute bootstrap frequency of 87) split in two lineages: Pholidosauridae and Dyrosauroidae. This node is supported by 16 unambiguous synapomorphies (ch: 4 [1], 6 [2], 40 [1/2], 51 [0], 59 [1], 62 [1], 70 [1], 104 [1], 125 [1], 149 [1], 163 [2], 164 [2], 166 [1], 168 [1], 179 [1], and 192 [1]).The clade Dyrosauridae highlighted an absolute bootstrap of 56 with 40 unambiguous synapomorphies (ch: 2, 3, 8 [1], 11 [2], 14 [1], 17 [3], 18 [0], 29 [1], 39 [1], 43 [1], 44 [0], 45 [1], 54 [1], 55 [1], 56 [0], 59 [0], 63 [0], 67 [1], 70 [0], 71 [1], 81 [1], 85 [0], 90 [2], 102 [2], 114 [0], 115 [1], 118 [1], 119 [1], 120 [1/2], 122 [1], 124 [0], 139 [1], 140 [1], 142 [1], 148 [1], 159 [1], 162 [0], 169 [1], 186 [1], 193 [1], and 196 [1]). As for Hyposaurinae, five unambiguous synapomorphies and an absolute bootstrap of 21 (ch: 4 [2], 36 [1], 42 [1], 67 [0], and 194 [1]). The strict consensus places the specimen MGUAN-PA 548 in a polytomy with Luciasuchus ( Jouve et al. 2021) , Dyrosaurus spp. and a large group of Hyposaurinae. The polytomy is supported with only one unambiguous synapomorphy (ch. 198 [0]). MGUAN-PA 548 is supported by a unique combination of characters: the occipital condyle much wider than high (ch. 2 [2]) and the parietal without broad occipital portion (ch. 96 [0]). Within the Hyposaurinae lineage, only the genus Dyrosaurus is supported with a bootstrap higher than 71. Although well resolved, the tree does not appear to be stable (low bootstrap frequencies), as can be seen for the whole tree outside the well-defined clades described above.

Bayesian inference

The Bayesian analysis reached convergence for each parameter (ESS> 200), and the good quality of the analysis is supported by PSRF = 1.000 and ASDSF = 0.005345. In the maximum compatibility tree (MCT), the major clades are supported with high posterior probabilities, i.e. Tethysuchia (PP = 0.89), Dyrosauridae with MHNM-kh01 (PP = 0.74), and Hyposaurinae (PP = 0.80) ( Fig. 4 View Figure 4 ). However, within these taxa, the viability of the results decreases with posterior probabilities close to or below 0.5. This is particularly the case in the complex in which specimen MGUANPA-548 is placed. The complex, which includes species from the genera Rhabdognathus , Congosaurus ( Dollo 1914) , Atlantosuchus ( Buffetaut and Wouters 1979) , and Acherontisuchus ( Hastings et al. 2011) , along with the specimen analysed in this study, trends toward a posterior probability of 0.26. However, as the phylogenetic relationships among these taxa become more clearly defined, this posterior probability decreases further. Only the two species of the genus Rhabdognathus ( Swinton 1930) have higher posterior probabilities (PP = 0.56). Identical observations are recognized with the maximum parsimony analysis, major clades exhibit stronger support with bootstrap values exceeding 50 [i.e. Tethysuchia (BS = 87), Dyrosauroidea (BS = 56), and Dyrosauridae (BS = 56) ( Fig. 3 View Figure 3 )]. However, the MCT topology presents some differences with the strict consensus tree of maximum parsimony analysis. First, the tree is divided into two major clades:Tethysuchia and Goniopholididae . Second, Dyrosauroidea is included within Pholidosauridae and not two different branches like the maximum parsimony topology. Within the dyrosaurids, the tree topology is quite similar:the branch of Hyposaurus ( Owen 1849) and Dorbignysuchus ( Jouve et al. 2021) is placed as the sister-lineage to Rhabdognathus ( Fig. 4 View Figure 4 ). The polytomy with MGUANPA-548 is resolved and the hyposaurine are divided in two branches with the first one including Dyrosaurus species, Luciasuchus , and Arambourgisuchus ( Jouve et al. 2005b) .

However, the change of placements of Brachiosuchus kababishensis (as one of the earliest Dyrosauridae ) and Chenanisuchus lateroculi Jouve et al., 2005a (as sister-taxa of Cerrejonisuchus improcerus Hastings et al., 2010 and Anthracosuchus balgorus Hastings et al., 2015 with a posterior probability of 0.99) are notable. Nonetheless, as for clock rates, the whole tree shows rather low rates (clock rates <1), but it is possible to distinguish two events of strong morphological change (clock rates> 2). The first occurs at the base of the Tethysuchia during the Jurassic, associated with the first Pholidosauridae , and the second from the Aptian to the Campanian, covering the entire origin of the Dyrosauridae lineage. There is also a high rate of occurrence at the origin of the taxon Atlantosuchus coupatezi Buffetaut and Wouters, 1979 .

Diversification rates

Our data do not follow the theoretical normal distribution for each time bin and parameter (see Fig. 5 View Figure 5 ; Supporting Information, Table S2; Shapiro–Wilk test, all P -values <0.05). Because the data do not follow a normal distribution, the Student t -test will not be used to analyse the result. The speciation and extinction parameters of the two time bins (before and after 66.0 Mya) remain different and are supported by P -values below 0.05, as determined by Bartlett's and Mann–Whitney U-test ( Fig. 6 View Figure 6 ; Supporting Information, Table S2). For the speciation and extinction parameters, the Fligner–Killeen and Bartlett test does support homoscedasticity with a P -value below 0.05. Despite this, the relative fossilization results from the tests (excluding Shapiro–Wilk) are associated with P -values of 0 ( Fig. 6 View Figure 6 ; Supporting Information, Table S2). Despite the questionable zero value, there seems to be a clear fossilization bias between the two time bins (before and after the K–Pg event, an aspect discussed later).

Estimated size

The absence of a complete cranium makes it difficult to reconstruct the overall size of the specimen. Nevertheless, based on the results of phylogenetic analyses, MGUAN-PA548 is close to the genera Luciasuchus , Acherontisuchus , Atlantosuchus , Rhabdognathus , and Congosaurus , and as such is considered to be an elongated and slender-snouted crocodylomorph ( Jouve and Schwarz 2004, Jouve 2007, Jouve et al. 2008). In 2022, Salih et al. estimated the size of Rhabdognathus keiniensis Jouve, 2007 as between 4.72 and 5.44 m. The posterior width of the holotype of the species Rhabdognathus keiniensis (MNHN TGE 4031; Jouve 2007; Fig.8 View Figure 8 ) exhibits a similar size range to MGUAN-PA548 (MNHN TGE 4031 ≈ 213 mm, based on the figure 9.2 in Jouve 2007, MGUAN-PA548 = 219 mm). Thus, we estimate the size of the Angolan individual also close to 5 m.