Dyrosauridae is an extinct family of neosuchian crocodyliforms that survived the Cretaceous-Palaeogene (K-Pg) extinction and thrived in marginal settings of the Tethys Ocean from the Late Cretaceous until their extinction during the early Eocene (Swinton 1950; Halstead 1975; Buffetaut 1976, 1978a, b, 1982; Brochu et al. 2002; Jouve 2005, 2007, 2021; Martin et al. 2019; Jouve et al. 2021). Occurrences of dyrosaurid species are known from North and South America, Africa, and Asia, with the most localities and most complete remains being found in Africa (e.g., Denton et al. 1997; Khosla et al. 2009; Sena et al. 2017; Jouve et al. 2021; Amoudji et al. 2021). The previous record of dyrosaurids in South America is essentially limited to seven regions (Cope 1886; Langston 1953, 1965; Buffetaut 1991; Barbosa et al. 2008; Hastings et al. 2010, 2011, 2015; Jouve et al. 2021). In Colombia (Fig. 1A), the published records of dyrosaurid remains are from the Cerrejon Mine and from Ortega, Tolima. In the former locality three species have been described from the middle Paleocene coal-bearing Cerrejon Formation (Hastings et al. 2010, 2011, 2015). From Ortega, six vertebral centra were described, recovered from red claystone that probably correspond to dyrosaurids of Maastrichtian age (?) (Langston 1953, 1965; Jouve et al. 2021). However, the colour of sediments cast doubt on the age assignment since the outcrops in the locality dominantly correspond to a Paleocene unit, especially the red sediments (Raasveldt 1956; Núñez et al. 1984a, b).

Here, we describe a dyrosaurid vertebra collected by the second author in late 2013, in the Colombian Llanos Foothills (Fig. 1A). The specimen was found in the Piñalerita Creek, north of Sabanalarga town (Casanare), in beds of the Cuervos Formation (Figs 1; 2; 3).

The vertebra was collected on the Piñalerita Creek, 6 km north of Sabanalarga town, Casanare (Fig. 1A, B). It was collected from rocks of the late Paleocene-early Eocene Cuervos Formation (Jaramillo & Dilcher 2001; Jaramillo et al. 2005) (Fig. 2). The Cuervos Formation is sandwiched between two dominant quartz arenite successions with cross-bedded beds, the Barco Formation (below) and the Mirador Formation (above) (Figs 1B; 2). The Cuervos Formation is around 475 m in thickness, it consists dominantly of gray and greenish gray mudstone mottled by rhizoliths), with intercalations of thick beds of slightly carbonaceous shale and greenish gray sub/litharenites with crossbedding (Fig. 2). The vertebra described here was collected 276 meters above the base of the unit. The bed containing the vertebra is greenish gray, massive with rhizoliths, has calcareous concretions and is locally thinly laminated (Fig. 3).

The vertebra is assigned to a possible Thanetian age based on the stratigraphic ranges of the palynologic taxa in the Llanos Basin. Previous palynologic studies of the Cuervos Formation in the Piñalerita section and in the basin showed rich and diverse pollen/spore assemblages, which are considered the most reliable biostratigraphic tool in the Paleogene succession of Colombia (Jaramillo & Dilcher 2001; Jaramillo et al. 2005; Jaramillo et al. 2011). Between the samples UFP 8 to UFP 27 of Jaramillo & Dilcher (2001), there is a last appearance datum (LAD) and first appearance datum (FAD) for several markers, where the model ages by Jaramillo et al. (2011) suggest a possible stratigraphic position below the Thanetian-Ypresian boundary (Fig. 2).

Palynofacies and lithofacies analyses from this locality by Jaramillo & Dilcher (2001), indicate that the Los Cuervos Formation represents coastal plain/fluvial floodplain deposits incised by discrete meander rivers and eventual lacustrine deposits. The palynoflora is dominated by palms with intermittent presence of damp tropical forest. The depositional environment of the vertebra-bearing bed corresponds to coastal plain muddy deposits.

The dyrosaurids are distributed worldwide, except in Europe and ancient Eurasia (India not included), with a temporal range extending from the late Cretaceous to the middle or late Eocene (e.g. Buffetaut 1982; Hastings et al. 2015; Jouve 2021) (Appendix 4). They began their apparent diversification during the Maastrichtian, reaching their maximum diversification during the Paleocene, and declined from the Ypresian (Fig. 8A). Only one species is known during each stage from the Lutetian and later, the taxic dyrosaurid diversity seems to have been dwindling from the Lutetian to their extinction during the Bartonian or Priabonian. On the contrary, the Paleocene seems to be the dyrosaurid golden age. Phylogenetic correction of the taxic diversity does not consistently change the global patterns of the curves, increasing the diversity in older time bins, because the phylogenetic estimate only constructs backward-extending ghost lineages (Tennant 2016). Post Selandian South American dyrosaurid diversity is almost unknown while, as demonstrated by the remains described herein, it existed. The intervals with the worst known dyrosaurid diversity are mainly those from the Eocene, those that remain uncorrected by the phylogenetic corrections. For these reasons, the phylogenetically corrected diversity could be less relevant and reliable than the uncorrected one.

Crocodylians show a similar evolution to their diversity through time, but an opposite geographical distribution. Whereas the Late Cretaceous to Eocene crocodylian sampling comes essentially from Europe, North America, and Asia, where dyrosaurids are absent or very poorly diversified (De Celis et al. 2020; Jouve 2021), the evolution of their diversity follows those observed for dyrosaurids. Crocodylian diversity rose during the Maastrichtian-Paleocene, reached its maximal speciation during the Thanetian (Solórzano et al. 2019), as did the African dyrosaurids, and was followed by an Ypresian-Lutetian drop. A strong correlation is observed between dyrosaurid and crocodylian diversity (Table 3), with high Pearson coefficients of correlation reaching 0.98 (p: 0.00028). Spearman’s correlation is only high and relevant when taxic dyrosaurid diversity is compared to sub-sampled crocodylian diversity (Appendix 9). This suggests that dyrosaurid and crocodylian evolution follows a globally similar pattern from the Late Cretaceous to the Eocene and could have been impacted by the same biotic and abiotic factors, even if crocodylians are mainly found in the northern hemisphere, and dyrosaurids in former Gondwana.

Comparison of dyrosaurid diversity with several abiotic factors is interesting and has previously been attempted. Martin et al. (2014) did not find a clear correlation between dyrosaurid extinction and sea surface temperature evolution, but the inconsistent sampling for the sea surface temperatures for the key period in dyrosaurid evolution, Paleocene-Eocene, is strongly biased and questions the reliability of the results (Jouve et al. 2017). Forêt et al. (2024) did not find clear correlation between temperature and tethysuchian evolution, like Mannion et al. (2015) for marine crocodyliforms. Jouve & Jalil (2020) did not find a global correlation between temperature proxies (δO18 from Prokoph et al. 2008) and tethysuchian diversity, but a positive correlation for the Late Jurassic-earliest Cretaceous (Oxfordian-Cenomanian) interval, and on the contrary, an inverse correlation with the same proxy for the Turonian-Thanetian and a strong positive correlation with sea level. So, the signal provided by dyrosaurids or tethysuchians is not clear and correlation of their evolution with biotic and abiotic factors is also unclear.

Mean temperatures (Grossman & Joachimski 2022) and temperature proxy (δO18; Cramer et al. 2009) from two different datasets used herein show a clear rise in the global temperature during the Ypresian, corresponding to the Early Eocene climatic optimum (EECO) (Fig. 8B, C). The mean δO18 (Cramer et al. 2009), also shows three other temperature rises, one during the Danian, at the boundary between Thanetian and Ypresian (corresponding to the Paleocene-Eocene Thermal Maximum, PETM), another during the early Bartonian (Middle Eocene climatic optimum, MECO) (Fig. 8B). The Maastrichtian seems to correspond to a period of a strong drop in temperature (Fig. 8C). δO18 proxy for the sea surface temperatures also shows a strong rise during the Thanetian-Ypresian (Judd et al. 2022; Fig. 8D). Both global and sea surface temperatures show a similar highest level during the Ypresian, but while the lowest temperature values were obtained for the latest Cretaceous period, the lowest sea surface temperatures seem to be middle Paleocene (Fig. 8). Neither of these highest or lowest temperature values correspond to highest or lowest dyrosaurid diversity values. Sea level from Miller et al. (2005) data follows more or less the sea surface temperatures with a maximal rise during the Ypresian (Fig. 8E). Jouve & Jalil (2020) found a strong correlation between both proxies for the Turonian-Thanetian time interval.

As discussed in previous papers (Hastings et al. 2015; Barbosa et al. 2008; Jouve & Jalil 2020; Jouve et al. 2008, 2021; Jouve 2021), dyrosaurids diversified at the end of the Late Cretaceous and the beginning of the Paleocene, a time interval of strong drop in temperatures. To the contrary, in the Ypresian the Early Eocene Climatic Optimum (EECO) corresponds to an apparent drop in dyrosaurid diversity (Fig. 8; Table 2). A similar drop is observed in the Northern hemisphere for crocodylian diversity (De Celis et al. 2020), with a high extinction rate during the Lutetian-Bartonian (Solórzano et al. 2019). Also, could the Paleocene-Eocene Thermal Maximum (PETM) have had an impact on the dyrosaurid and crocodylian diversities as it did for mammalian faunas, explaining the lower Ypresian diversity? These drops in diversity are contrary to what we would expect from taxa assumed to be tropical. Even if the low diversity sets limits on the reliability of any statistical analysis, none of the tested proxies were found to correlate with dyrosaurid diversity evolution (Table 4; Appendix 9).

In well documented localities with continuous Late Cretaceous-Early Eocene fossil records such as in the Oulad Abdoun Basin in Morocco, the number of Paleocene dyrosaurid species is particularly high during the Paleocene (6 species), but the diversity is only 2 species during the Ypresian (Appendix 4; Bardet et al. 2014, 2017). The Ypresian dyrosaurids are particularly well preserved and known by numerous complete skeletons (Jouve et al. 2006). It is interesting to note that Paleocene diversity is particularly high, with a different fauna in each stage, with two species from the Danian, five during the Selandian-Thanetian, and only two during the Ypresian, while both Paleocene and Ypresian have a comparable temporal length (respectively 10 and 8.2 my). So, this African Paleocene high diversity and Ypresian decline is coherent with local observations in well sampled sites, and could thus, be a good image of the diversity, and not strongly impacted by collecting and preservation biases, at least in Africa. The Paleocene could be an epoch of high post crisis diversification, related to intense colonisation of new environments, liberated by the extinction of large marine Mesozoic reptiles (Jouve 2021), but likely with a quick rate of turnover. Differences in dyrosaurid environmental preferences, fresh water during the Cretaceous versus marine during the Paleogene (Jouve 2021), are congruent with this hypothesis. The Ypresian could have been a more stable period for dyrosaurid diversity, with a low rate of diversification, and not a real decline. In the Oulad Abdoun Basin the same three species are present during all the Ypresian (two dyrosaurids and a crocodylian species). In this basin, the same phenomenon is observed with the chelonians, with at least nine Paleocene and only three Ypresian species (Bardet et al. 2014, 2017). This postcrisis restoration, then stabilisation, is followed by a strong climatic cooling reaching its acme during the Priabonian. The dyrosaurids did not survive to this period. The role that each successive and sometimes overlapping event played on dyrosaurid diversity during this relatively short interval of geologic time (latest Cretaceous-middle Eocene) cannot be clearly distinguished at present. Similar evolution is observed for crocodylians in the Northern Hemisphere, in particular in North America: a strong Maastrichtian-Paleocene rise, Selandian high turnover (high speciation and extinction rates), positive Thanetian diversification with high speciation and low extinction rates, and an Ypresian-Lutetian drop (Solórzano et al. 2019; De Celis et al. 2020). The Ypresian crocodylian diversity compared to the Paleocene shows a much lower speciation rate, a low extinction rate, and a null rate of diversification. The diversity is stable (Solórzano et al. 2019). This pattern could correspond to what is observed in dyrosaurids, explaining the possible correlation observed between the evolution of both taxa (Table 3).

Compared to South America and Africa, the dyrosaurid history seems to be different in North America and India, where the diversity is low. They are known from fragmentary remains on the Indian continent from Maastrichtian to Bartonian (Buffetaut 1978b; Khosla et al. 2009), and from Maastrichtian to Thanetian, maybe Priabonian, in North America (Ehret & Hastings 2013) (Appendix 4). These land masses may have different reasons for their similarly low dyrosaurid diversities. While the Maastrichtian-Eocene crocodyliform fauna of India is poorly known, several species are known from abundant material in North America, where at least four longirostrine crocodylians and one dyrosaurid have been recovered, spanning the Cenomanian to Eocene (Brochu 2004, 2006; Jouve 2021). In North America, two to three contemporaneous marine gavialoid species are found in each stage from Campanian to Thanetian, when they are scarce and probably restricted to fresh-water environments in Africa. Due to the occurrence of multiple crocodylian species during this time, the relative lack of dyrosaurid species in North America likely indicates that this is not a sampling bias and that actual dyrosaurid diversity was indeed low. The lower dyrosaurid diversity might be due to the presence of several crocodylian species and/or a combination of a cooler climate and the presence of the paleo Gulf Stream (Jouve 2021). To the contrary, the reliability of the weak Indian diversity is doubtful. The complete absence of dyrosaurids in Eurasia could be due to cooler climatic conditions and the presence of colder oceanic currents, less favourable to them than to crocodylians (Jouve 2021).

There is a strong correlation between North American crocodylian diversity and African dyrosaurid diversity. It seems to be stronger than that obtained between global crocodylian diversity and African dyrosaurid diversity, as the Spearman’s and Pearson’s correlations for all datasets are significant, corrected or uncorrected (Table 3; Appendix 9). This suggests that both groups in both continents have a similar evolution to their diversity, and are probably impacted by the same factors. On the contrary, the evolution of crocodylians in Europe through the K-Pg crisis strongly differs from what is observed in North America, with a strong drop in diversity (Puértolas-Pascual et al. 2015; De Celis et al. 2020). So, regional particularities must also be considered when evaluating crocodyliform diversity.

The new dyrosaurid from Colombia suggests that Late Paleocene-middle Eocene South American dyrosaurid diversity is particularly poorly known, and this prevents meaningful comparison between its evolution and that of North America and Africa.

Dyrosaurids are widely distributed, but they are better known in Africa. The present specimen originates from the Thanetian and is thus the youngest known South American dyrosaurid. It suggests the presence of dyrosaurids later than previously supposed, to at least the end of the Paleocene. This new record provides a different image of the evolution of South American dyrosaurids, with a history that may be more like that of Africa.

Evaluating dyrosaurid diversity is particularly difficult, because its low diversity precludes the use of reliable statistical methods for comparison to changes in biotic and abiotic factors as well as their evolution being tied to multiple global events: the K-Pg extinction, colonisation of new ecological spaces, the Paleocene-Eocene Thermal Maximum, and the Early Eocene Climatic Optimum. All these events confuse the interpretation and identification of which and when each factor is at play. The same is true for the evolution of the crocodylians in the northern hemisphere. The reasons why the dyrosaurids declined and became extinct during the Eocene are obscure. Competition with crocodylians and climatic changes are the most frequently mentioned reasons, but the Paleocene high diversification and turnover mask the clear identification of the stage of their real decline. As suggested here, our knowledge of the African fauna suggests that the Ypresian was probably a stage of stabilisation in dyrosaurid diversity, and not a period of a real decline, congruent with what is observed for crocodylians in the northern hemisphere. The Lutetian could be the real beginning of their decline in diversity, corresponding to a strong drop in temperature and strong climatic cooling, which is more congruent with the climatic sensitivity usually hypothesised for crocodyliforms, and in particular dyrosaurids. A similar Lutetian decline is observed in crocodylian diversity. Contrary to the distribution of crocodylians across the warm temperate climatic belt, dyrosaurids were less diversified and restricted to tropical climates, which are likely to be more impacted by cooling, thereby reducing their chances for survival and potentially causing extinction.

A clear faunal segregation exists between crocodylians from the northern hemisphere and dyrosaurids in the southern hemisphere, yet we found strong similarities between their evolution. The African and South American dyrosaurid diversity from the Late Cretaceous to middle Paleocene has a quite similar evolution to those of crocodylians, particularly in North America. To the contrary, late Paleocene and Eocene diversity is particularly low in South America compared to what is observed for African dyrosaurids and northern hemisphere crocodylians. No South American dyrosaurids were known from this period of time previous to the present discovery, and this strongly suggests that the Late Paleocene-Early Eocene South American dyrosaurid diversity is probably poorly known and underexplored. So, we recommend focusing prospecting work in South America on the poorly known Thanetian to Late Eocene. New material could provide information on the evolution of South American dyrosaurids and help evaluate whether their history on this continent is close to that observed in Africa (and also North American crocodylians) or represents a regional particularity (like the European post K-Pg crisis). This could provide new tools to study the impact of climatic variations on crocodyliform evolution and the influence of regional factors.

The authors are grateful to Dr Alex Hastings, an anonymous reviewer, and editor E. Côtez for their constructive comments.