Coincidence of photic zone euxinia and impoverishment of arthropods in the aftermath of the Frasnian-Famennian biotic crisis

The lowermost Famennian deposits of the Kowala quarry (Holy Cross Mountains, Poland) are becoming famous for their rich fossil content such as their abundant phosphatized arthropod remains (mostly thylacocephalans). Here, for the first time, palaeontological and geochemical data were integrated to document abundance and diversity patterns in the context of palaeoenvironmental changes. During deposition, the generally oxic to suboxic conditions were interrupted at least twice by the onset of photic zone euxinia (PZE). Previously, PZE was considered as essential in preserving phosphatised fossils from, e.g., the famous Gogo Formation, Australia. Here, we show, however, that during PZE, the abundance of arthropods drastically dropped. The phosphorous content during PZE was also very low in comparison to that from oxic-suboxic intervals where arthropods are the most abundant. As phosphorous is essential for phosphatisation but also tends to flux off the sediment during bottom water anoxia, we propose that the PZE in such a case does not promote the fossilisation of the arthropods but instead leads to their impoverishment and non-preservation. Thus, the PZE conditions with anoxic bottom waters cannot be presumed as universal for exceptional fossil preservation by phosphatisation, and caution must be paid when interpreting the fossil abundance on the background of redox conditions.


Geological Background
The active Kowala quarry is located in the southern limb of the Gałęzice-Bolechowice syncline in the southern part of the Kielce region of the Holy Cross Mountains (Fig. 1). During the Devonian, the Holy Cross Mountains area was part of a carbonate shelf that extended along the southern margin of the continent of Laurussia near the equator [25][26][27][28][29] . The Famennian sediments of the Kowala quarry were deposited in the intrashelf Chęciny-Zbrza basin 28,30,31 . The lower Famennian deposits investigated in this paper are 21 m thick and crop out in a trench located in the north-central part of the quarry (N50°47′43,476′′, E20°33′53,568′′, Fig. 1). The section comprises monotonous deposits of thin-bedded, dark, carbonaceous shales and thin-bedded, grey, micritic limestones. The investigated interval is confined to lithologic unit H-4 of Racki & Szulczewski 32 , which stratigraphically encompasses the Late triangularis through Early marginifera conodont zones 32 . Earlier, the lower Famennian interval investigated in this paper was assumed as being confined to the crepida conodont Zone 9,15 , on the basis of lithological similarities and its position relative to the neighbouring trench section investigated by Marynowski et al. 33 . However, the new conodont dating showed that our section is slightly older and represents the Palmatolepis minuta minuta Zone (according to the latest conodont zonation of Spalletta et al. 34 ), which corresponds to the previously established Late triangularis conodont Zone of Ziegler & Sandberg 35 . The conodont assemblage includes: Palmatolepis protorhomboidea, Pa. delicatula delicatula, Pa. superlobata, Polygnathus brevilaminus, Po. procerus, Icriodus alternatus alternatus, and Ic. deformatus deformatus.
The depositional environment is generally interpreted as deep shelf, below storm wave-base, but at least episodically within the limits of the photic zone 33,36-38 . Material and Methods collection of fossils. We collected 2029 specimens in total directly from the investigated section. At least 32 specimens were collected from each of the 34 investigated shale or marly shale beds, regardless of the preservation state or systematic position of the specimens. The specimens were not retrieved from limestones as they are much harder to work with because they are very likely to be damaged while splitting the rock.
For the purposes of this study, fossilised regurgitates, coprolites, and trace fossils were excluded, as they represent the signs of animal activity, and not the animal itself. In some cases, the total abundance of some body fossils within each bed was difficult to establish because their skeletons tend to disarticulate post-mortem (crinoids, coelacanth fish) or were extremely rare (angustidontid arthropods); these fossils were marked on diagrams but not evaluated quantitatively. Finally, we included 1840 of the 2029 specimens collected in situ in our quantitative palaeoecological analyses.
Since during the fieldwork, obtaining the same number of specimens for each bed was not always achievable and some of the collected specimens were excluded from the study, the general number of each fossil type in each bed was converted to a percentage contribution. Such a calculation allowed for comparisons with all the needed proportions being saved, despite the differences in total number of specimens between the studied beds. total organic carbon (toc) and total sulphur (tS) content. For total carbon (TC), total inorganic carbon (TIC) and total sulphur (TS) measurements, an ELTRA CS-500 IR-analyser was used (at Faculty of Earth Sciences, University of Silesia in Katowice, Poland). TOC was calculated as the difference between TC and TIC. Gas chromatography coupled with mass spectrometry (Gc-MS). GC-MS analyses were carried out with an Agilent Technologies 7890 A gas chromatograph and Agilent 5975 C Network mass spectrometer with a Triple-Axis Detector (MSD). Separation was obtained on a fused silica capillary column (J&W HP5-MS, 60 m x 0.25 mm i.d., 0.25 µm film thickness) coated with a chemically bonded phase (5% phenyl, 95% methylsiloxane). The GC oven temperature was programmed from 45 °C (1 min) to 100 °C at 20 °C/min, and then to 300 °C at 3 °C/ min (hold 80 min), with a solvent delay of 10 min. Helium was used as a carrier gas at a constant flow of 2.6 ml/ min. Analyses were performed at the Faculty of Earth Sciences,, University of Silesia in Katowice, Poland. For more details see 41 . inorganic geochemistry. The 34 pulp samples were analysed at Bureau Veritas Acme Labs Canada Ltd.
Major, minor, and trace elements were analysed using inductively coupled plasma optical emission spectrometry (ICP-OES) and inductively coupled plasma mass spectrometry (ICP-MS) (details described in 42 ). The precision and accuracy of the results were better than ±0.05% (mostly ± 0.01%) for the major elements and generally better than ±1 ppm for the trace elements.

Results
Abundance and composition of fossil assemblages. In the studied assemblages (Figs 2 and 3), the most common fossils consisted of phosphatic shells of linguloid (Orbiculoidea sp.; 36.3%) and calcitic shells of rhynchonellid (25.4%) brachiopods. Interestingly, when Orbiculoidea brachiopods dominate, the relative abundance of rhynchonellids lowers and vice versa (Fig. 4). The next most abundant assemblages are orthoconic nautiloids (17.6%) preserved in the form of carbonaceous imprints of their periostracum or poorly preserved internal moulds. In general, they are more abundant in younger beds (Fig. 4). The first small peak in their percentage contribution occurs in beds Kow75 and Kow77 (12.5% and 13.33% respectively). The next peak in relative abundance Arthropods preserved as phosphatised carapaces are the third most abundant group. Concavicarid thylacocephalans (12.1%) and unidentified (due to fragmentation and/or preservation state) arthropod remains (most probably Concavicarididae as well; 2.8%) are the most common arthropods. The rarest arthropods of these units are phyllocarids with only 13 fossils (0.7%) found in situ and one single Angustidontidae (maxillipedes).
Non-calcifying algae (1.8%), preserved as carbonaceous compressions, the bivalve Guerichia sp. (1.7%), crinoid ossicles (1.0%), and coelacanth fish bone clusters or isolated elements (0.5%), are all very rare (Figs 2 and 4). Thus, they have not been included in quantitative analyses. changes in arthropod abundance within the assemblages. In general, the percentage contribution of arthropod fossils decreases towards the younger beds (Fig. 4). The most abundant arthropods are representatives of the Concavicarididae (Thylacocephala), whose abundance is characterised by several rapid changes with maximally 8%. Unidentified arthropod remains (the majority may represent Thylacocephala, as well) follow the same pattern as Thylacocephala fossils, with their maxima in the same beds ( Fig. 4 and Table 3). In the case of www.nature.com/scientificreports www.nature.com/scientificreports/ Phyllocarida, their relative abundance rises only in four intervals. In the intervals between these beds, no phyllocarid remains were found ( Fig. 4 and Table 1).
These changes are also reflected in the total number of the collected arthropod specimens (Fig. 5). The number of specimens found in situ ranges from 0 (beds Kow 64, Kow 140) to 24 (bed Kow 72) and, in general, it gradually lowers towards the younger beds with tendency to differ between neighboring beds (Fig. 5). However, in some cases, when total number of acquired specimens lowers in the same beds, the percentage contribution of arthropods rises (Fig. 5). Specimen numbers per sample range between 9 (bed Kow 128) and 123 (bed Kow 188). Generally, the fossil abundance rises towards the younger beds. However, three peaks of increased abundance (compared to the overall trend) were found: beds Kow 66-77, Kow 128-152 and Kow 172-190 b. In the lower part of the section (Kow 64-77), the changes in arthropod and total specimen numbers are simultaneous and similar in their amplitudes (the relative abundance does not change much). This applies also to the first rise in fossil abundance (Kow 106-114). The first part of the second major peak of abundance (Kow 128-142) is, however, characterised by a drop in arthropod abundance. In bed Kow 140, we found no arthropod remains. Then, in the second part of this peak (Kow 142-152), their abundance rises following the general trend, but with a much lower amplitude. In the last major peak, the number of arthropod specimens again drops gradually, with some minor peaks in beds Kow 172 (beginning of the total peak), Kow 180 and Kow 190; these changes, however, follow the opposite trend, i.e. the arthropod number rises while at decreasing sample size. organic geochemical data. The results are presented in Table 2 and Fig. 6. Total organic carbon values are in a rather narrow range from 1.2% wt. to 4.6% wt. in our samples. In general, samples that are rich in carbonate are slightly depleted in organic carbon, but differences in TOC content between marls and shales are minor. The section is characterised by a low total sulphur concentration between 0% and 0.5% wt., but in most of the samples, it is below 0.1% wt.
Steranes to hopanes ratio values are in the range of 0.5 to 0.8 (Table 2) which is similar to other parts of the Kowala section, including Annulata and Dasberg 42,43 , and lower than values noted for the Hangenberg secion 5 . These data imply that both algae and bacteria were important organisms contributing to the kerogen formation. Gammacerane, an indicator of water column stratification, was found in the all samples with relative concentration also compatible to the other sections from the Kowala quarry 42,43 . Despite the small differences in the TOC www.nature.com/scientificreports www.nature.com/scientificreports/ content, important changes in isorenieratane concentrations were found (Fig. 6). This compound is a biomarker of green sulphur bacteria and acts as an indicator of euxinic conditions in the photic zone of the water column 3,38 . Two maxima in isorenieratane concentrations were identified (Table 2 and Fig. 6). The first is in the lower part of the section (around bed Kow 100), reaching 25 µg/g TOC and the second, smaller maximum (±10 µg/g TOC) in the upper part of the section. Moreover, isorenieratane closely correlates with the 2,3,6−/3,4,5-diaryl isoprenoid, i.e., the so called palaeorenieratane (R 2 = 0.95), which usually co-occurs with isorenieratane in Palaeozoic sedimentary rocks (e.g. 4,[42][43][44][45][46][47][48][49]. Additionally, there is a strong correlation with the concentration of aryl isoprenoids, which are compounds that are degradation products of isorenieratane (e.g. 3,48,50 In almost all samples the maleimides (1H-pyrrole-2,5-diones), decomposition products of chlorophylls and bacteriochlorophylls, were identified as the abundant group of compounds from the polar fraction. Between maleimides, those treated as of bacteriochlorophyll origin (with 2-methyl-3-iso-butyl-configuration; Me,i-Bu) were present in almost all samples. However, the Me,i-Bu to Me,Et (2-methyl-3-ethyl-) maleimide ratio 40,51,52 does not correlate with the isorenieratane and palaeorenieratane concentrations.
inorganic geochemical data. The Th/U ratio in almost all carbonate-rich/limestone samples is >1, which is indicative of oxic conditions. Only three samples from the lower part of the investigated section reached Th/U ratio values below 1, which imply dysoxic bottom water conditions. The lower values of the Th/U ratio in shales (<3) are indicative of oxygen depleted conditions in the whole analysed section (Fig. 6). The V/Cr ratios range from 1.21 to 9.03 (Table 3); these values are indicative of oxic through dysoxic to anoxic conditions (e.g. 5,53 ). The V/Cr ratio shows a good correlation with Mo as well as the isorenieratane contents. Additionally, the C org /P ratio  www.nature.com/scientificreports www.nature.com/scientificreports/ also closely corresponds with these data (Fig. 6 and Table 3). The maximum values of the C org /P ratio around bed Kow 101 (first postulated anoxic interval) reached from 116.92 to 212.47, while in the second anoxic interval located between Kow 158 and Kow 162, the ratios reached values from 193.16 to 768.5 C org /P, respectively. All results are presented in Table 3 and in Fig. 6.  Table 2), organic and inorganic proxies indicate significant changes in depositional conditions and abundance of some fossil groups. There are two visible spikes of isorenieratane, palaeorenieratane and aryl isoprenoid concentrations ( Fig. 6; Table 1). The first is located in the lowermost part of the section, around sample Kow 101, and the second is in the upper part, between samples Kow 158 and Kow 162 (Fig. 6). Similar patterns were observed for the Mo concentration and V/Cr ratio, which reached their maxima at the same stratigraphic levels (Fig. 6). The third, smaller isorenieratane spike that correlated with higher concentrations of Mo, V/ Cr and C org /P ratios is observable in the lowermost part of the section in sample Kow 75 (Fig. 6). Conversely, inorganic proxies based on the uranium (Th/U and U autig ) and maleimide ratios are not in agreement with the above parameters. The discrepancy between the U proxies and Mo concentration and the V/Cr ratio in the upper part of the section is unclear, especially since all these parameters work closely together in the other Upper Devonian sections from the Kowala quarry, such as those containing the Annulata or Hangenberg events 5,42 . Notably, part of the uranium was transported to the basin with a detrital fraction, which, in consequence, changed the ratio between the detrital Th and U connected with primary organic matter. However, the lack of correlation between U and Al (R 2 = 0.11) does not confirm this mechanism as a reliable factor. The other, more likely explanation of disagreement between Th/U and the other proxies is based on the assumption, that the euxinic zone was located in the water column but did not reach, or only periodically reached, the sea-floor (as it was shown in 43 ). Uranium was partially diluted from the sediment after deposition during oxic periods, while Mo connected with pyrite framboids 54   www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ shuttle cannot be excluded as a process driving Mo concentration, but this element does not correspond to the correlation between inorganic and organic indicators.
The lack of correlation between the isorenieratane concentration and the Me,i-Bu to Me,Et maleimide ratio is much easier to explain. The origin of Me,i-Bu-maleimides is connected with green sulphur bacteria while Me,Et-maleimides are degradation products of both chlorophyll and bacteriochlorophyll 40,51 . This implies that Me,i-Bu to Me,Et maleimide ratio values are not only dependent on green sulphur bacteria (GSB) blooms but also on the flowering of algae and other marine microorganisms that generate chlorophyll. Thus, Me,i-Bu-maleimides are useful indicators of GSB occurrence, but the Me,i-Bu to Me,Et maleimide ratio does not necessarily illustrate the intensity of euxinia.
Gammacerane, an indicator of water column stratification 55 was found in the all investigated samples and the values of G/H ratio are quite stable across the section (Table 2). This implies, that anaerobic ciliates, which are precursors of gammacerane 55 , were constantly present in water column and most possibly fed on green sulphur bacteria. The presence of gammacerane also indicates a permanent stratification of water column in the aftermath of F/F crisis.
The next proxy that is in accordance with the other proxies used here is the C org /P ratio. The values of the C org /P ratio, ranging from 30 to 150, are characteristic for high productivity and periodically oxygen-restricted conditions (see 56 ), while the higher values of the C org /P ratio (>150) are indicative for high-productivity and permanent anoxic conditions on the seafloor. In the case of the investigated section, higher values of C org /P are in agreement with the Mo, V/Cr and isorenieratane values, thus confirming a higher productivity and seafloor euxinia (Fig. 6).
Based on isorenieratane, palaeorenieratane and aryl isoprenoid concentrations, as well as the Mo concentration, V/Cr and C org /P ratio values, the following scenario of depositional conditions during the triangularis zone of the early Famennian can be presented. At the first stage of deposition, conditions were dysoxic to anoxic on the seafloor with photic zone euxinia (PZE) reaching (periodically?) the bottom waters. Then, the conditions became more aerobic, which took place between sedimentation of the layers K112 and K151. The second event saw anoxia/euxinia occurring between the formation of the layers K158 and K167. During sedimentation of the uppermost part of the section, oxygenation of the bottom waters prevailed again, while the water column above may have still witnessed some PZE. The episodes with domination of euxinia in the bottom part of the water column are shaded on Fig. 6.

Relationship between redox changes and fossil abundance. Depositional conditions are an
important factor controlling the state of fossil preservation 57 , and, euxinia/anoxia/dysoxia and water column stratification played an important role in the exceptional preservation of past organisms 2,58 . At Kowala, the lower Famennian section investigated here is regarded as an example of a conservation deposit, possessing well-preserved, abundant and diverse assemblages of arthropods, fish and non-biomineralised macroalgae 9,13,16 , even with sporadic cases of soft tissue preservation 59 . Interestingly, this fossiliferous interval falls within the lowermost Famennian triangularis Zone that marks the immediate aftermath of the Frasnian-Famennian biotic crisis. During this time interval, oxygen-deficient conditions in the Kowala basin occurred as confirmed through gamma-ray spectrometry 60 , as well as geochemical and petrographic studies 43 . Therefore, taking these findings into account, we have expected that the highest number of exceptionally preserved fossils (especially phosphatised arthropods) occur in horizons characterised by elevated anoxia/euxinia, especially since the PZE has been considered as an important factor in preserving fossils in Upper Devonian Lagerstätte deposits such as the Australian Gogo Formation 2 .
However, we found increased amounts of thylacocephalan arthropods at intervals where conditions were rather suboxic and PZE did not reach the seafloor (Figs 4, 5). These intervals of higher abundance of arthropods especially embrace the section from samples K112 to K128, K142 to K150, and K171 to K175 (non-shaded parts of the section in Figs 4 and 5). In the case of other fossils having originally phosphatic shells such as orbiculoid brachiopods, which are the most abundant fossils occurring throughout the studied section (Fig. 4), such tendencies were not observed. It is possible that these organisms could have colonised the sea-bottom only during short oxygenated pulses in the otherwise suboxic bottom waters (e.g. 5,61,62 ,). However, considering their large abundance throughout the studied section, irrespective of the anoxic events, it is even more probable that the orbiculoids were epiplanktonic 63 ; these orbiculoids would have drifted in the surface waters attached to algae occurring in the same deposits 16 , or would have functioned as opportunistic bottom dwellers able to thrive in stressed, oxygen-deficient conditions as other linguloid brachiopods known from the Devonian 64 .
Our data show that arthropods (and especially dominant thylacocephalans) were common only during oxic/ suboxic periods (Figs 5, 6) with a thin PZE located in the water column above 43 , while more restricted conditions were unfavourable for this group of organisms.
The mode of life of thylacocephalans is still problematic. They have been interpreted either as nektonic predators 65 , benthic ambush predators, or benthic scavengers [66][67][68] . For some species (mostly protozoeids), nektonic mode of life was proposed on the basis of such features as their small size, elongated carapace and hyperthrophied eyes 69,70 . A nektobenthic mode of life of thylacocephalans instead, has been proposed by various authors (e.g. 67,71 ) based on such features as reduced posterior trunk appendages, lack of a flexible abdomen (see 11,70,72 ) and their rather thin, poorly mineralised cuticles 68,73 . However the last feature is shared with some nektonic arthropods, like amphipods 74 , so except the last one, these features can be observed in modern nektobenthic crustaceans (both swimming and seafloor-dwelling 75 ). Although the anatomy of both raptorial and posterior appendages of the Kowala thylacocephalans remains unknown, the anatomy and architecture of their carapaces were described by Broda and Zatoń 10 . The authors pointed out an additional important feature: the presence of "sensory belts" in the upper and lower margins of the carapace. The authors assumed that these zones, consisting of many organule canals, are the remnants of a highly developed sensory system that can monitor the surrounding space. Such a (2019) 9:16996 | https://doi.org/10.1038/s41598-019-52784-4 www.nature.com/scientificreports www.nature.com/scientificreports/ developed set of sensors could allow these predators to easily find their prey either in the surrounding water column, or hiding on the sea bottom. Such sensors could have also served as anti-predatory "alarm device". However, taking all these known features into account, we are still not sure about the exact thylacocephalan mode of life. Depending on species, it would probably be either nektonic or necto-benthic organism.
The lack of arthropod fossils in euxinic intervals (Figs 3 and 5), however doesn't have to be necessarily connected with unfavourable living conditions. Based on modern examples from the Santa Monica Basin, the Black Sea, the Baltic Sea and other basins, anoxic conditions at the seafloor promote phosphorous flux out of the sediment [76][77][78] . This is in accordance with P concentrations throughout the section, showing its lowermost values exactly within the euxinic levels ( Fig. 6). At the same levels, C org /P ratio values are also the highest. As shown by Zatoń et al. 9 , exoskeletal remains of all arthropods from the Late Devonian of the Kowala quarry are phosphatic. Apparently, a lack of sufficient phosphorous in sediments during the euxinic periods precluded arthropod preservation and in consequence controlled the overall fossilisation processes of the arthropod exoskeletons. The abundance of arthropod exoskeletons in rocks characterised by suboxic conditions during their sedimentation coincides with much higher values of P (Figs 4 and 5). Phosphatisation requires a special microenvironment characterised by a specific pH, redox conditions and a sufficient concentration of P [79][80][81][82] . Thus, all parameters promoting phosphatisation of the arthropod cuticle appear to have been present in the Kowala suboxic environment.
The alternative hypothesis, which assumes mass mortality events that may have occurred during euxinic events (e.g. 8 ), is not supported in our case. This is primarily because the PZE intervals in the studied section are depleted in fossil arthropods. In conclusion, the post-Frasnian-Famennian crisis interval in the Kowala quarry is rich in the opportunistic benthic linguloid brachiopod Orbiculoidea, orthoconic nautiloids and phosphatised remains of arthropods, among which Thylacocephala dominate. As linguloid brachiopods occur throughout the studied section, the arthropods appear to be absent in intervals when PZE was detected. The simultaneous drop in phosphorous content in the euxinic intervals indicates that the absence of arthropods resulted rather from their non-preservation due to low P content than from changes in palaeoenvironmental conditions. Thus, in this case, euxinic/anoxic conditions negatively influenced the preservation of arthropod exoskeleton via phosphatisation instead of promoting their fossilisation as has been suggested in an earlier study 2 . Apparently, PZE conditions were not universal for fossil preservation through phosphatisation. Thus, any interpretations concerning the abundance of fossils within euxinic horizons should be treated with caution, and taphonomic causes may be crucial, primary factors in controlling the presence or absence of fossils in the rock record.