Our results show that ‘archeobatrachian’ anurans with the same locomotor mode clump together with very little overlap in more or less distinct domains of the morphospace defined by the first two PCA axes (Fig. 3A), which together represent more than 86% of the total variance (PC1: 63.36%, PC2: 22.76%). This indicates that species with different locomotor modes tend to have different limb proportions. Most jumpers and swimmers show negative values of the PC1 associated with comparatively short forelimbs and long hindlimbs, but most jumpers and swimmers clearly differ in the PC2, which relates to jumpers often having tibiofibulae that are clearly longer than the femora and longer proximal tarsals than swimmers. In contrast, walker/hoppers are restricted to positive values of the PC1, having relatively longer forelimbs and shorter hindlimbs with respect to jumpers and swimmers, and all but burrowing specialists show positive values of PC2. The departure of burrowing species (e.g., Pelobates spp.) from most walkers/hoppers mainly relates to the fact that the femur is clearly longer than the tibiofibula.

The Barremian anurans from Iberia show diverse limb proportions, some of which overlap extant ‘archeobatrachian’ species in the morphospace, whereas others occupy somewhat distinctive areas, although most still within the 95% confidence interval of extant forms in the PC1–PC2 morphospace (Fig. 3A), calling for caution when interpreting inferences for these particular taxa. Most of these fossils are located in the morphospace closer to extant jumpers and swimmers than to walkers/hoppers, some having proportions more similar to those of particular neobatrachians than to any extant ‘archeobatrachian’ species (Supplementary Material 1). For instance, Wealdenbatrachus jucarensis shows extreme negative values of PC1, whereas the pipimorphs Neusibatrachus and Gracilibatrachus show extreme negative values of PC2. Conversely, Eodiscoglossus santonjae and MUPA-LH 11392 plot almost in the center of the PC1–PC2 morphospace, in between extant species with different locomotor modes, which might suggest that their limb proportions are rather generalized.

Application of the pFDA classification rules with the estimated optimal Pagel's lambda (0.4) results in four extinct species classified as jumpers and the remaining three as swimmers, with high posterior probabilities (> 95%) in most cases (Fig. 3B). Moreover, those classifications with high posterior probabilities at optimal lambda do not vary across the entire range of lambda values, including classification of the pipimorphs Neusibatrachus and Gracilibatrachus as swimmers and Wealdenbatrachus jucarensis, MUPA-LH 7990, and MUPA-LH 11392 as jumpers. It has to be noted that, in spite of its seemingly generalized limb proportions, MUPA-LH 11392 is consistently classified as a jumper, which relies on the fact that the distance to the centroid of jumpers is consistently shorter than those of other locomotor groups. In contrast, Eodiscoglossus and Iberobatrachus are classified at optimal lambda with relatively low posterior probabilities (< 70%) as swimmer and jumper, respectively, but their classification vary across the range of lambda, being alternatively classified as swimmers at low values and jumpers at high values. Fossils were never classified as hopper/walkers with the classification rules derived from the pFDA, disregarding the value of lambda (Fig. 3B).

The fossil frogs from the Iberian Plate represent at least five distinct genera, namely the already known Eodiscoglossus, Gracilibatrachus, Iberobatrachus, Neusibatrachus, and Wealdenbatrachus, but MUPA-LH 11392 certainly represents an additional one. In turn, these taxa might represent at least five families (although they are not defined yet) and four major radiations of ‘archeobatrachian’ anurans: stem lalagobatrachians, costatans, stem (or alternatively basal) pipanurans, and basal pipimorphs (Báez, 2013, Báez, 2016, Báez and Gómez, 2016, Báez and Gómez, 2019 and Gómez, 2016; Supplementary Material 1; Table 1; Fig. 2). Conversely, sampled tropical and subtropical assemblages represent between three to seven families and two to five major radiations each, being ‘archeobatrachian’ anurans largely unrepresented (Supplementary Material 1; Fig. 4A). In addition, the inferred phylodiversity encompassed by this Barremian assemblage rivals or even surpasses the current anuran phylodiversity of the Iberian region and of sampled tropical and subtropical assemblages across the world (Supplementary Material 1; Fig. 4A). In ecological terms, the Barremian assemblage is similar to the sampled extant wetland assemblages in that each encompasses swimmers and is rich in jumpers, but differs from the latter in lacking walker/hoppers and in having a larger proportion of swimmers (Fig. 4A). The ecomorphological disparity calculated by SOV on the standardized limb proportions (Fig. 4B) and the standardized ratios plus svl (Fig. 4C) is statistically similar between extinct and most extant assemblages, although the latter is somewhat higher for the Barremian assemblage. The Barremian wetland encompasses taxa not only with diverse limb proportions, but also of different size classes ranging from the miniature adult pipimorphs (20 mm svl) to the medium-size Wealdenbatrachus jucarensis (50 mm svl), which barely resembles in size and proportions some extant species of Discoglossus.

Limb proportions in sampled ‘archeobatrachians’ correlate well with locomotion, as has previously been observed in diverse anurans (Emerson, 1988, Enriquez-Urzelai et al., 2015, Jorgensen and Reilly, 2013, Lires et al., 2016 and Zug, 1972), which allowed us to infer the locomotor abilities in extinct taxa. The extinct frogs from the Iberian wetland, which show diverse limb proportions but more typical of jumpers or swimmers, were classified accordingly by the pFDA rules. Some classifications do not depart from expectation, as in the cases of the pipimorph taxa or Wealdenbatrachus. The pipimorphs Neusibatrachus and Gracilibatrachus, which are part of the stem of the highly aquatic pipids (Báez, 2013, Báez and Sanchiz, 2007 and Gómez, 2016), are consistently inferred as swimmers, in agreement with the aquatic lifestyle inferred on the basis of several skeletal features other than limb proportions (Báez, 2013 and Gómez and Pérez-Ben, 2019). Also concurring with previous suggestions mainly based on limb proportions and ilial morphology (Báez and Gómez, 2019), specimens of Wealdenbatrachus are invariably classified as jumpers.

On the contrary, the locomotor behaviors of the remaining Barremian frogs from Iberia have not been advanced previously, but some remarks have previously been made on their hindlimb proportions. In this regard, Iberobatrachus has been considered more similar to Wealdenbatrachus than to Eodiscoglossus, although differences in their proximal tarsals have been stressed (Báez, 2013 and Báez and Gómez, 2019). In line with these previous observations, Iberobatrachus is classified as a jumper under most analytical conditions, yet many of these classifications are with posterior probabilities only marginally over 50% and swimming as the primary locomotor mode should not be dismissed. It is also noteworthy that Iberobatrachus clearly differs from Wealdenbatrachus in having proportionally shorter hindlimbs relative to svl (although this feature was not part of our analyses; Fig. 1J, K) and in lacking a well-developed dorsal crest on the ilial shaft, which has been interpreted as indicative of different locomotor abilities (Báez and Gómez, 2019). Crested ilia characterize many, but not all, proficient jumpers among neobatrachian anurans, though they are only present in species of Discoglossus among extant ‘archeobatrachian’ jumpers (Gómez and Turazzini, 2016, Přikryl et al., 2009 and Reilly and Jorgensen, 2011), which are known to exert lower jump forces than neobatrachians (Herrel et al., 2016). Therefore, the sole lack of a dorsal crest on the ilium cannot be taken as evidence against jumping locomotion. In any case, it is possible that the uncertainty in the classification might indicate that both jumping and swimming were important parts of the locomotor repertoire of Iberobatrachus, which would agree with semiaquatic habits, although a more thorough scrutiny of its anatomy and the integration of additional data from extant taxa will be needed to elucidate this issue.

The osteology of Eodiscoglossus santonjae has recently been revised in detail; its generalized limb proportions have been regarded as part of its overall plesiomorphic morphology (Báez and Gómez, 2016). This appears to be confirmed by our limb data, as Eodiscoglossus plot almost at the centroid of the ‘archeobatrachian’ morphological diversity between extant species of the three different locomotor modes considered herein (Fig. 3A). However, this taxon is classified as a swimmer under most analytical conditions, yet with relatively low posterior probabilities, being otherwise classified as a jumper. This is interesting, since generalized limb proportions might imply an all-purpose morphology capable to cope with different demands imposed by diverse microhabitats (e.g., Soliz et al., 2017 and Tulli et al., 2016). Among the locomotor and other substrate-related behaviors of anurans, however, Eodiscoglossus does not seem to be able to jump long distances, walk fast or burrow effectively. Most walker/hoppers that also burrow exhibit stout, proportionally short hindlimbs and feet with a more or less hypertrophied prehallux (Emerson, 1976 and Fabrezi et al., 2017), whereas Eodiscoglossus shows only a tiny ossified prehallux, despite having a hypertrophied prepollex (Báez and Gómez, 2016). Also, it has not the comparatively long forelimbs and short hindlimbs of efficient fast-walkers relative to other anurans (Reynaga et al., 2018). Therefore, in this case the ambiguity in the classification might indicate that Eodiscoglossus was a semiaquatic unspecialized swimmer with some jumping capability, although this remains to be tested with additional data.

Taken together, our results suggest that among these Early Cretaceous frogs, there were some jumpers and swimmers having more extreme limb proportions than extant ‘archeobatrachian’ species, but also species having more generalized limb proportions with varying degrees of jumping and swimming abilities, typical of aquatic and semiaquatic frogs (Emerson, 1978, Emerson, 1988 and Soliz et al., 2017). Interestingly, this ecomorphological structure with good representation of aquatic or likely semiaquatic jumping and swimming species matches the expectation given that the depositional environments might have been part of a large subtropical wetland system extending across northeastern Iberia (Buscalioni and Poyato-Ariza, 2016 and Buscalioni et al., 2018). It is most similar in ecomorphological structure to that from Pagdanan Range on the Philippine archipelago among the extant wetland assemblages surveyed (Fig. 4) and contrasts with that of the frog assemblage currently inhabiting northeastern Iberia, which is dominated by terrestrial jumpers and walker/hoppers, dwelling borders of swamps or temporary ponds developed under more temperate and drier climatic and environmental conditions (AmphibiaWeb, 2018 and Trochet et al., 2014). Additional comparisons with the extant assemblages further stress the remarkable taxonomic and ecomorphological diversities represented among Barremian fossils, which, despite the intrinsic biases of the fossil record, equal or even surpass present-day values of taxonomic, phylogenetic, and ecomorphological diversity (Supplementary Material 1; Fig. 4). The high ecomorphological disparity depicted by fossils, encompassing jumpers and dedicated swimmers plus jumper/swimmer putatively semiaquatic generalists of different size classes, certainly agrees with previous observations on these anurans (Báez and Gómez, 2019 and Buscalioni and Poyato-Ariza, 2016) as well as with the paleoenvironmental reconstruction of northeastern Iberia as a complex subtropical wetland during the Barremian (Buscalioni et al., 2018 and Fregenal-Martínez and Buscalioni, 2009). This large wetland system constituted a heterogeneous environment with diversity of intermingled aquatic, semiaquatic, and terrestrial microhabitats (Buscalioni and Poyato-Ariza, 2016, Buscalioni et al., 2018 and Gomez et al., 2015) and might support a high diversity of anurans and other components of the biota from the Iberian Plate by providing a mosaic of available niches (Provete et al., 2014 and Tews et al., 2004).

It is also noteworthy that the Barremian assemblage is exclusively made up of ‘archeobatrachian’ taxa, whereas extant assemblages worldwide are largely dominated by neobatrachians and only a few ‘archeobatrachian’ groups may be part of subtropical or tropical wetlands, namely a few bombinatorid costatans, pipid xenoanurans, and megophryid anomocoelans (Supplementary Material 1; Fig. 4A). The current scenario is not surprising, since ‘archeobatrachians’ represent only 4% of all extant anuran species (Frost, 2019). This, coupled with the observed pattern in the morphospace of limb proportions showing some Barremian fossils falling outside the hulls defined by extant ‘archeobatrachians’ (Fig. 3A), indicates that the extant taxonomic and ecomorphological ‘archeobatrachian’ diversity has become impoverished. Early Cretaceous assemblages from Laurasian continents other than that from Iberia also lack neobatrachians, whose diversification was well underway then as it is evidenced by the anurofauna from the Aptian Crato Formation of Brazil (Báez et al., 2009) and as it is inferred based on molecular time estimates (e.g., Pyron, 2014). Despite Neobatrachians reached a widespread distribution across most Gondwanan continents by the end-Cretaceous, they remained absent from Iberia and most other Laurasian terrains (Báez et al., 2012, Gao and Chen, 2017, Ikeda et al., 2016 and Xing et al., 2018; this study), which reinforces the hypothesis of an origin and early diversification of these anurans in Gondwana (Báez et al., 2009 and Báez et al., 2012) and suggests some geographical isolation from Laurasian landmasses during most of the Cretaceous.

The frogs from the Barremian Iberian wetland are amongst the earliest well-known representatives of some major crown-group radiations, providing valuable data on the early ecomorphological diversification of anurans. The origin of the anuran Bauplan has traditionally been linked with saltatory locomotion (e.g., Gans and Parsons, 1966 and Jenkins and Shubin, 1998), although this view has recently been challenged on different grounds (Ascarrunz et al., 2016, Herrel et al., 2016, Lires et al., 2016 and Reilly and Jorgensen, 2011). Alternatively, a walker/hopper locomotion has been proposed as the primitive anuran condition in some recent studies (e.g., Ascarrunz et al., 2016, Jorgensen and Reilly, 2013 and Reilly and Jorgensen, 2011), although the latter did not consider swimming as a discrete locomotion category in their analyses. It is surprising, though, that among the Barremian frogs from Iberia none was classified here as walker/hopper, whereas jumpers are well represented. However, these jumpers either appear to be generalists or similar in size and proportions to extant Discoglossus, which produced low jump forces most often associated with swimming than jumping in most studied anurans (Herrel et al., 2016). Moreover, dedicated swimmers are also well represented in this Early Cretaceous assemblage, and swimming abilities are also expected for the remaining taxa, including those consistently classified as jumpers, since almost every extant anuran shows some swimming capability (Emerson and De Jongh, 1980, Peters et al., 1996 and Soliz et al., 2017), though species distantly related or with different ecologies might swim rather differently (Nauwelaerts and Aerts, 2002 and Robovska-Havelková et al., 2014). In this regard, general similarities in limb proportions and kinematics between swimming and jumping frogs have previously been acknowledged (Emerson and De Jongh, 1980, Lires et al., 2016, Nauwelaerts et al., 2007, Peters et al., 1996 and Soliz et al., 2017; but see Robovska-Havelková et al., 2014), and there are many wetland-dweller neobatrachians that excel at both types of locomotion (Nauwelaerts et al., 2007 and Soliz et al., 2017). However, recent work on jumping performance across the anuran tree demonstrated that ‘archeobatrachian’ jumpers do not really excel in jumping locomotion as they exert only low jump forces in comparison to jumping neobatrachians (Herrel et al., 2016). Frogs from the Barremian of Iberia exclusively represent miniature to medium-size ‘archeobatrachians’ with jumping and/or swimming abilities in a large and complex wetland, depicting a scenario compatible with the hypothesis regarding early frogs as small poor jumpers or swimmers that made their peculiar Bauplan in aquatic environments evolve (Herrel et al., 2016).

The Early Cretaceous frogs from the Iberian wetland exhibit a remarkable taxonomic, phylogenetic, and ecomorphological diversity, encompassing miniature to medium-size ‘archeobatrachians’ with varying degrees of jumping and swimming abilities, indicating a Konservat-Lagertätte preservation of this wetland anuran assemblage (Buscalioni and Poyato-Ariza, 2016). In addition, this assemblage rivals modern tropical and subtropical wetlands here surveyed, which have relatively high species richness representing up to five major anuran radiations, whereas most temperate frog communities encompasses less than a handful of syntopic species (Sinsch et al., 2012 and Wells, 2007). This further highlights the subtropical nature of this Iberian wetland and suggests that it could be an ancient anuran biodiversity hotspot. It would be of interest also to investigate other less spectacular Early Cretaceous frog assemblages to grasp a better panorama of the early phylogenetic and ecomorphological diversification of crown-group anurans and assess if frogs were also part of the Cretaceous Terrestrial Revolution (Lloyd et al., 2008).