Paleotemperature record of the Middle Devonian Kačák Episode

The Middle Devonian Epoch, ~ 393–383 million years ago, is known for a peak in diversity and highest latitudinal distribution of coral and stromatoporoid reefs. About 388 million years ago, during the late Eifelian and earliest Givetian, climax conditions were interrupted by the polyphased Kačák Episode, a short-lived period of marine dys-/anoxia associated with climate warming that lasted less than 500 kyr. Reconstruction of the seawater temperature contributes to a better understanding of the climate conditions marine biota were exposed to during the event interval. To date, conodont apatite-based paleotemperatures across the Eifelian–Givetian boundary interval have been published from Belarus, France, Germany and North America (10–36° S paleolatitude). Here we provide new δ18Oapatite data from the Carnic Alps (Austria, Italy) and the Prague Synform (Czech Republic). For better approximation of the paleotemperature record across the Kačák Episode, a latitude-dependent correction for Middle Devonian seawater δ18O is applied. Because δ18Oapatite data from shallow marine sections are influenced by regional salinity variations, calculated mean sea surface temperatures (SST) are restricted to more open marine settings (22–34° S paleolatitude). Water temperatures reach ~ 34 °C in the Prague Synform and ~ 33 °C in the Carnic Alps and suggest that SSTs of the southern hemisphere low latitudes were ~ 6 °C higher than previously assumed for this time interval.

www.nature.com/scientificreports/ paleotemperatures were published from southern hemisphere low latitudes between 10° and 36° S for localities in France (Pic de Bissous) 25 and Germany (Blankenheim, Hillesheimer Mulde, Schönecken-Dingdorf and Blauer Bruch) 25,26 . Additional data from Nevada, USA (northern Antelope mountains) documented a negative shift in the δ 18 O apatite during the onset of the LEE1 (upper part of the kockelianus Biozone) 27 , and more recently, paleotemperature reconstructions from Belarus were provided for the latest Eifelian interval (ensensis Biozone) 28,29 .
Here we discuss late Eifelian to earliest Givetian paleotemperature estimates based on new δ 18 O apatite data from a shallow to deeper water transect within the Carnic Alps (Austria, Italy) and an offshore section from the Prague Synform (Czech Republic). Together with palaeotemperature estimates from other localities within the Rheic Ocean and the North American shelf, mean sea surface temperatures for each region and biozone are calculated and compared (Fig. 1). Because paleotemperature estimates from shallow marine localities appear to be biased by varying salinities, the temperature record across the Kačák Episode is limited to more open marine deposits of the Carnic Alps.
Keeping in mind that Middle Devonian coral-stromatoporoid reefs showed a remarkable wide latitudinal dispersal of up to 50° on either hemisphere with an acme in diversity just after the Kačák Episode during the early and middle Givetian (~ 387-382 million years ago) [30][31][32] , only a detailed study of the event interval can show how the marine biosphere dealt with environmental changes at that time. Estimation of paleotemperatures across the event interval contributes to uncover critical temperatures and to identify ecological limits of climate sensitive low latitude communities.

Geological settings
The Carnic Alps, part of the Proto-Alps 34 , were situated on the northern peri-Gondwana shelf located around 34° S during the Devonian. Today the area forms a more than 2000 m high mountain chain along the Austro-Italian border with Middle Devonian rocks well-exposed between the Biegengebirge in the west and Mount Osternig in the east ( Fig. 2A,B) 35 . Coeval sections from different bathymetric settings were investigated (Fig. 2B). The section studied in the Val di Collina quarry (N46° 35′ 49.07″; E12° 55′ 29.22″) consists of back reef and reef environments belonging to the Spinotti and Kellergrat formations, respectively. The Zuc di Malaseit Basso (ZMB) section (N46° 33′ 19.06″; E13° 11′ 10.6″) was deposited within a distal slope setting of the Hoher Trieb Formation, while the Wolayer "Glacier" section (N46° 36′ 46.56″; E12° 52′ 33.66″) is represented by condensed pelagic deposits of the Valentin Formation. These sections were correlated using high-resolution conodont biostratigraphy 11 .
The Middle Devonian Srbsko Formation in the Prague Synform (part of the Bohemian Massif) is generally developed in siliciclastic facies 36 and was located around 28° S. The studied Jirásek quarry section I (N49° 54′ 50.2″; E14° 04′ 34.2″) represents a unique section where the stratigraphic equivalent of the Kačák Member of the Srbsko Formation is developed in carbonate facies. The so-called Upper Dark Interval (UDI), exposes a dark gray thin-bedded to nodular limestone interval representing the uppermost portion of the Acanthopyge Limestone (Choteč Formation, Fig. 2A,C) 37 . Upper Eifelian deposits are unconformably overlain by a limestone breccia, wherefore the middle/upper? part of the ensensis Biozone within the UDI was probably reworked 18 .  www.nature.com/scientificreports/   Fig. 3; Supplementary Fig. 1). The condensed pelagic sequence of the Wolayer "Glacier" section yielded abundant conodonts, which were extracted from cm-thick limestone levels separated along slightly uneven stylolite layers. δ 18 O apatite values measured across the Eifelian-Givetian boundary vary between 17.7 and 18.0‰ ( Fig. 3; Supplementary Fig. 1).
In the Prague Synform, the δ 18 O apatite record at the Jirásek quarry section I includes 28 measurements covering the upper part of the kockelianus Biozone and most of the ensensis Biozone. Values range between 17.6 to 18.7‰ (VSMOW). The highest value of 18.7‰ is observed during the late kockelianus Biozone while the lowest value is observed before the entry of the ensensis Biozone ( Fig. 3; Supplementary Fig. 1). Variations of about 1‰ are recognized within the sampled interval.

Discussion
For a better understanding of the paleotemperature record across the E/G boundary, results from this study are compared with published δ 18 O apatite values [25][26][27][28] (Fig. 3). In total, 67 measurements are documented from the late Eifelian (kockelianus and ensensis biozones) and 21 values from the earliest Givetian (hemiansatus Biozone). The highest resolution is documented for the late Eifelian with 28 δ 18 O apatite values reported from the Prague Synform (28° S). All data included in this study derive from localities allocated in southern hemisphere latitudes between ~ 10-36° S. In addition to the common approach used for paleotemperature calculation from δ 18 O apatite , we applied a latitude-dependent correction for δ 18 26 and Narkiewicz et al. 28 (Fig. 3, Supplementary Fig. 2). This could be explained by regional salinity variations as suggested for localities between 10 and 14° S 28,29 or upwelling of colder intermediate waters in the Rheic Ocean [45][46][47][48] . SSTs modelled for the Middle Devonian 44 give significantly lower SSTs along the southern hemisphere Euramerican shelf compared to temperature estimates for open ocean settings of the same latitude. According to the Holocene SST record, a highest temperature of 32 °C (central equatorial Indian Ocean) and a lowest temperature of 17 °C (eastern equatorial Atlantic) between 0 and 5° S are documented 49 . As indicated by modern sea surface temperatures 50 , SST of low latitude waters does not always show a gradational temperature decrease towards higher latitudes. Hence, temperature variations within the low latitude belt like those documented in δ 18 O apatite from Euramerican shallow shelf localities between 10 and 24° S likely reflect, similar to present day SSTs, regional factors as for example evaporation, runoff via large river delta systems, submarine groundwater discharge, karst water flow in coastal areas, ocean circulation, or upwelling zones along continental shelfs.
Mean SST record across the Eifelian/Givetian boundary. The interpretation of the paleotemperature trend across the Eifelian/Givetian boundary is difficult, because of the low number of isotope values available. Data of only two localities, at 34° S (ZMB section, Italy) and 36° S (Pic de Bissous sections, France), span the entire interval from the kockelianus until the hemiansatus Biozone. Additionally, δ 18 O apatite values from the epeiric Belarusian Basin (10-14° S) 28,29 , as well as those derived from conodonts of shallow marine deposits at 24° S 25,26 , 34°S (this study, Supplementary Fig. 2)  Additionally, more open marine sections at 25-34° S show slightly decreasing mean SST from north to south during each biozone-interval ( Fig. 4A-C vs. D-F, Table 2). A mean latitudinal temperature gradient of ~ 5 °C between 25 and 34° S is observed for the pre-/onset LEE1 interval (kockelianus Biozone) which is somewhat higher compared to the present day mean temperature gradient of ~ 4.5 °C for the same latitudinal range 43 . Calculating a mean latitudinal temperature gradient between 28 and 34° S, a decrease from ~ 3 °C during the pre-/onset LEE1 (kockelianus Biozone) to ~ 1 °C during the LEE1-LEE2 (ensensis Biozone) is observed (compare Fig. 4D-E). The present-day mean latitudinal temperature gradient between 28 and 34° S is ~ 3.3 °C 43 . Thus, we argue that global warming resulted in a minor latitudinal SST gradient in the subtropics and probably slightly higher latitudes that may have had an important effect for marine biota. Specifically, climate sensitive organisms like corals and stromatoporoids could spread more easily towards higher latitudes along broad shelfs. Hence, we suggest that the Kačák Episode could have triggered the early-middle Givetian acme of coral distribution and diversity.

Conclusion
The main results of this study can be summarized as follows: • New conodont δ 18 O apatite data around the Kačák Episode (late Eifelian to earliest Givetian) from the Carnic Alps (34° S) and the Prague Synform (28° S) indicate mean SSTs between 31 and 34 °C. • SST of late Eifelian southern hemisphere low latitudes was at least 6 °C higher than previously assumed.  www.nature.com/scientificreports/ • Specifically, climate sensitive organisms could spread more easily towards higher latitudes along broad shelfs. Hence, we suggest that the Kačák Episode could have triggered the early-middle Givetian acme of coral and stromatoporoid distribution and diversity.

Materials and methods
Conodont extraction. Conodont elements were extracted from limestones using 5% formic acid (Carnic Alps 2013, 2015; Prague Synform 2013) and 6% acetic acid (Prague Synform 2020) diluted in 5 L buckets filled with tap water. In order to prevent dissolution of conodont elements, the acid was exchanged twice a day until the carbonates were completely dissolved (generally after 7-10 days). In case of large amounts of insoluble residues, heavy liquid separation was applied at room temperature (sodium polytungstate; density: 2.79 g/cm 3 ). The colour alteration index (CAI) of the conodont elements from the Carnic Alps is 4-5 and 5 ( Supplementary  Fig. 3) which agrees well with earlier observations of Brime et al. 53 , who published a comprehensive conodont CAI study of the Carnic Alps. Conodonts from the Jirásek quarry (Prague Synform) have a CAI of 3 (Supplementary Fig. 3). , CO 3 2− and OH − ) of single conodont elements 54 . Oxygen isotopes of bioapatite (conodont crown tissue without basal tissue preserved) within this study were analysed by TC-EA IRMS using a ThermoFinnigan Delta V Plus mass spectrometer at the Geozentrum Nordbayern (FAU Erlangen-Nürnberg, Germany). Depending on the amount of trisilverphosphate precipitated after dissolving conodont apatite in HNO 3 , triplicate measurements were conducted whenever possible. Values are reported in ‰ relative to VSMOW. The standard deviation of replicate sample analyses was ± 0.02 to ± 0.34 (1σ; see Supplementary  26 are correlated within biozonal boundary ranges via high-resolution magnetic susceptibility and geochemical data. Published data of other sections from Germany and from France 25 are plotted according to the absolute age 25 and correlated via conodont biozone boundaries [55][56][57][58] . Data from North America 27 are correlated with the ZMB section of the Carnic Alps 11 via the australis/kockelianus biozone boundary and the upper limit of the T-R cycle Id (calibrated to 393.37 Ma) 11 . Data published from Belarus 28 are biostratigraphically constrained by conodonts to the ensensis Biozone and were already combined within the standard-corrected data of Joachimski et al. 25 .

Repository.
Calculation of latitude dependent δ 18 O of seawater. Seawater δ 18 O is dependent on the ratio evaporation vs. precipitation which is higher in subtropical latitudes relative to equatorial and higher latitudes. The calculation of spatial gradients in δ 18 O seawater follows Roberts et al. 38 , which prerequisites exact paleolatitude calculation for each locality. Since the climate conception of an ice-free Middle Devonian 40,42 compares better with early Paleogene climate conditions rather than with preindustrial climates, we adopted latitude-dependent seawater δ 18 O estimates for the early Paleogene 38 .
Paleolatitude calculation. The calculation of paleolatitudinal positions for the various areas follows Scotese and Wright 59 (Supplementary Fig. 2). Evaluation of the tectonic setting around Val di Collina quarry 53 results in a slightly more southern paleolatitudinal position than suggested by Scotese and Wright 59 . In general, the paleo-position is calculated to a latitudinal precision of 1°, based on the equation used.

Data availability
All data of this study are provided in the paper and Supplementary Information.