The fish-to-tetrapod transition—followed later by terrestrialization—represented a major step in vertebrate evolution that gave rise to a successful clade that today contains more than 30,000 tetrapod species. The early tetrapod Ichthyostega was discovered in 1929 in the Devonian Old Red Sandstone sediments of East Greenland (dated to approximately 365 million years ago). Since then, our understanding of the fish-to-tetrapod transition has increased considerably, owing to the discovery of additional Devonian taxa that represent early tetrapods or groups evolutionarily close to them. However, the aquatic environment of early tetrapods and the vertebrate fauna associated with them has remained elusive and highly debated. Here we use a multi-stable isotope approach (δ13C, δ18O and δ34S) to show that some Devonian vertebrates, including early tetrapods, were euryhaline and inhabited transitional aquatic environments subject to high-magnitude, rapid changes in salinity, such as estuaries or deltas. Euryhalinity may have predisposed the early tetrapod clade to be able to survive Late Devonian biotic crises and then successfully colonize terrestrial environments.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank N. Scharff and B. Lindow (Natural History Museum of Denmark) for authorizing the sampling of Devonian material; M. Creuzé des Châtelliers for access to the zoology collections of the Centre de Ressources pour les Sciences de l’Evolution (CERESE, FED 4271, Université de Lyon, Université Claude Bernard Lyon 1); O. de Lataillade (Ferme du Ciron), P. François (Pierrelatte), E. Liatout (Maison Liatout) and the fishery (Maison Pupier) for providing present-day material; and L. De Brito. M.Z. was supported by Natural Science Foundation of China (41530102) and Key Research Program of Frontier Sciences of CAS (QYZDJ-SSW-DQC002). This study was supported by the CNRS INSU program InterrVie, and the Institut Universitaire de France (C.L.).Reviewer information
Nature thanks M. Böttcher, O. Nehlich and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Extended data figures and tables
a, Subcomplete skull of the placoderm Remigolepis acuta (4-117; numbers in parentheses refer to sample numbers, reported in Supplementary Table 2). b, Subcomplete skull of the sarcopterygian Eusthenodon sp. (27-1151) in dorsal view. c, Dermal scales of the sarcopterygians Holoptychius sp. (30-1186). d, Partial hemimandible of an indeterminate member of Tetrapoda (42-1375). Scale bars, 10 cm.
a, δ34Sbone of Devonian tetrapod apatites plotted as a function of sulfur concentration. Samples for which sulfur concentrations are higher than 0.6% (outlined in red) are considered to have been potentially affected by diagenetic alteration. b, Covariation of δ18Op and δ18Oc as a function of carbonate concentration. Samples with both δ18Op − δ18Oc differences higher than 14.7‰ and carbonate content higher than 13.4% (outlined in green) are considered to have been potentially affected by diagenetic alteration. (See also Supplementary Table 2). In a, b, each dot represents an independent fossil sample (n = 51) and corresponds to the average value of three repeated measurements. Each error bar corresponds to 1 s.d. (Supplementary Table 2).
a–f, Sulfur content (mol% × 1,000) of apatites from Devonian tetrapod is graphically represented against various elements to test possible associations as secondary precipitated mineral phases in fossil bioapatites. The elements tested were magnesium (Mg; a), manganese (Mn; b), iron (Fe; c), copper (Cu; d), strontium (Sr; e) and barium (Ba; f). In a–f, each dot represents an independent Devonian apatite sample (light blue, n = 42; Supplementary Table 3) or associated matrix sample (light green, n = 7; Supplementary Table 3). A 0.6 wt% of sulfur is equal to 0.019 mol% × 1,000. Red straight lines correspond to calculated stoichiometric lines between the two elements. Black straight lines have been calculated using Pearson’s correlation test between the two elements for Devonian apatite samples and associated matrices. N.S., no significant correlation.
Spectra of three representative fossilized biogenic apatites of Devonian vertebrates with various sulfur contents (wt% as measured by the VarioPyrocube elemental analyser). The [PO4] and [SO4] absorption bands used to calculate the SO42−/PO43− ratios are indicated in black and red dashed lines, respectively. Each spectrum corresponds to a single measurement (Supplementary Table 4).
SO42−/PO43− ratios of apatites from Devonian tetrapods calculated from the infrared spectra are graphically represented against the S/P ratios calculated from elemental analysis. The two calculated regression lines (dashed) are both highly significant (Pearson’s correlation test; n = 42 and n = 36, respectively) and show a slope close to 1, which indicates that the majority of the sulfur present in apatite is in the form of sulfate that is structurally substituted for the phosphate in the crystal lattice of the apatite. Each dot represents an independent Devonian apatite sample (n = 42; Supplementary Tables 3, 5).
The apatite samples of Devonian tetrapods with a sulfur content over 0.019 mol% × 1,000 (Supplementary Table 3) have been outlined with a red circle. These samples display a significant negative correlation for their calcium and barium elements (Pearson’s correlation test; n = 21). This observation suggests that calcium is substituted with barium during the recrystallization process. In the same manner, some sulfates may also substitute in for the phosphates within the apatite lattice. Each dot represents an independent Devonian apatite sample (light blue, n = 42; Supplementary Table 3) or matrix sample (light green, n = 7; Supplementary Table 3).
a, b, Sulfur content (a) and isotope composition (b) of apatites from Devonian tetrapods are graphically represented against the cystallinity index (CI). None of them displays a significant correlation (Pearson’s correlation test; n = 47) with the crystallinity index, which indicates that recrystallization processes were not systematically associated with sulfate incorporation by elemental substitution and alteration of the sulfur isotope compositions. In a, b, each dot represents an independent Devonian apatite sample (n = 42; Supplementary Tables 2, 5).
a–c, The oxygen isotope composition of apatites from Devonian tetrapods is graphically represented against the crystallinity index (CI; a), the CO32−/PO43− ratio (b) and the SO42−/PO43− ratio (c). None of these indicators displays a significant correlation (Pearson’s correlation test; n = 47) with the oxygen isotope composition, thus arguing in favour of at least a partial preservation of the pristine oxygen isotope composition. In a–c, each dot represents an independent Devonian apatite sample (n = 47; Supplementary Tables 2, 5).
The phosphorus (P) content of apatites from Devonian tetrapods is graphically represented against the calcium (Ca) content. On the whole, the Ca and P contents of Devonian samples are significantly correlated (Pearson’s correlation test; n = 42) with a slope of 1.19, close to that defined by the NIST 1400 and NIST 1486, both of which are modern bones (red dots; Ca/P = 1.49). These results indicate that the loss of phosphate during recrystallization was almost stoichiometric for all samples. Each dot represents an independent Devonian apatite sample (light blue, n = 42; Supplementary Table 3), a matrix sample (light green, n = 7; Supplementary Table 3) or an international standard sample (red, n = 2; Supplementary Table 3).
Sulfur isotope composition of present-day vertebrate bones and muscles and environmental waters and food. n = 24 biologically independent animals
Carbon, oxygen and sulfur isotope composition of Devonian East Greenland and Chinese vertebrates. n = 51 independent Devonian apatite samples
Elemental concentration of Devonian apatite and some associated matrix (m) samples along with the two elemental standards NIST 1400 and NIST 1486
Fourier Transformed Infrared (FTIR) spectroscopy data
Crystallinity Index (CI), CO32−/PO43− and SO42−/PO43− ratios calculated from FTIR spectroscopy analyses