Upper Silurian) Oceanic Geobioevents – Coordination of Changes in Conodont, and Brachiopod Faunas, and Stable Isotopes

Gondwana Research 51 (2017) 272–288

Contents lists available at ScienceDirect

Gondwana Research

journal homepage: www.elsevier.com/locate/gr

Integrated record of Ludlow (Upper Silurian) oceanic geobioevents – Coordination of changes in conodont, and brachiopod faunas, and stable isotopes

Andrej Spiridonov a,⁎, Robertas Stankevič a,TomasGečas a,TomasŠilinskas a, Antanas Brazauskas a, Tõnu Meidla b, Leho Ainsaar b,PetrasMusteikisa, Sigitas Radzevičius a a Institute of Geosciences, Faculty of Chemistry and Geosciences, Vilnius University, M. K. Čiurlionio 21/27, LT-03101 Vilnius, Lithuania b Department of Geology, University of Tartu, Ravila 14a, 50411 Tartu, Estonia article info abstract

Article history: The Ludlow Epoch (Silurian) was marked by several globally recognized but mechanistically poorly understood Received 16 May 2017 biotic events. The most pronounced of them was the Lau Event, which strongly decimated conodont, graptolite, Received in revised form 4 August 2017 and brachiopod faunas. Additionally, this event coincides with the largest positive stable carbon isotopic anomaly Accepted 7 August 2017 in the whole Phanerozoic, as well as the resurgence of the so-called “anachronistic” microbial facies that were fre- Available online 1 September 2017 quently encountered during survival episodes of the major mass extinction events. In this contribution, based on č Handling Editor: R.D. Nance the analysis of the outer shelf facies succession (Milai iai-103 core), from the Lithuanian part of the Silurian Baltic Basin, as integrated quantitative record of conodont, brachiopod and δ13C changes across most of the Ludlow is Keywords: presented. The succession was subdivided into four conodont zones that served as a stratigraphic framework Conodonts for analyzing the δ13C and palaeoecological trends. The depth constrained cluster analysis revealed successions Brachiopods of three statistically distinct conodont and six statistically distinct brachiopod assemblages that replace each Stable carbon isotopes other near the change points in the stable carbon isotopic curve. The application of a newly developed mathemat- Lau event ical technique based on the analysis of recurrence patterns of the fossil assemblages revealed that both conodonts Time anomalous communities and brachiopods are represented by highly time specific assemblages in the aftermath of the Lau event (O. snajdri Joint recurrence analysis Interval Zone). The anomalous interval is confined to the transgressive and highstand phases of the 3rd order post-Lau transgression. The discussed interval is coeval with the extensive development of stromatolitic communities in the nearshore environments around the world. The results allow for the first time to quantify the pro- found ecosystem-wide geobiological impact of the mid-Ludfordian event that lasted up to the latest Ludfordian. © 2017 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.

1. Introduction (the last one at the very end of the Epoch), based on the integrated stratigraphical analysis of the distribution of conodont species, changes in sed- It is now widely recognized that the Silurian Period was a time of a re- imentary successions and stable carbon isotope trends in carbonates markable evolutionary turnover, extinction and biotic radiation events, (Jeppsson and Aldridge, 2000; Jeppsson et al., 2012; Melchin et al., accompanied by shifts in biogeochemical cycles of carbon and other nu- 2012). The Linde and Lau events most probably correspond closely to trients (Cooper et al., 2014; Crampton et al., 2016; Lenton et al., 2016). the leintwardinensis and kozlowski events, respectively (Jeppsson and The Ludlow epoch is characterized by up to five (Kaljo et al., 1996)glob- Aldridge, 2000; Jeppsson et al., 2012). Brachiopods experienced a signifi- ally recognized ecological and evolutionary perturbations of benthic, pe- cant drop in global genus level diversity in the mid-Ludfordian, an epi- lagic ecosystems (Talent et al., 1993; Urbanek, 1993; Jeppsson and sode of extinction called the “Pentamerid event” (Talent et al., 1993) Aldridge, 2000; Eriksson et al., 2009; Munnecke et al., 2012). Two major that is temporally equivalent to the Lau Event (Jeppsson and Aldridge, events are detected in the graptolite record of the Ludlow, namely the 2000). The data from the Prague Basin show that a significant reorganiza- early Ludfordian leintwardinensis Event and the mid-Ludfordian kozlowski tion of the benthic (trilobite, brachiopod and bivalve) communities took Event (Urbanek, 1993; Štorch, 1995; Štorch et al., 2014). Three oceanic place in mid-Ludfordian (Manda and Frýda, 2014). This all shows clearly events – the Linde, Lau and Klev events - are recognized in the Ludlow that the Ludlow Epoch was one of the most dramatic time intervals of the mid-Palaeozoic era (Calner, 2008). The most prominent geobiological turnover episode was the short- ⁎ Corresponding author. termed Lau Event (Jeppsson, 1998; Cramer et al., 2015). It was associated E-mail address: [email protected] (A. Spiridonov). with the largest positive stable carbon isotopic excursion of the whole

http://dx.doi.org/10.1016/j.gr.2017.08.006 1342-937X/© 2017 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved. A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 273

Phanerozoic eon (Calner, 2008), as well as the global resurgence of 2. Geological setting microbially built structures (Calner, 2005a) and oceanic carbonate whitings due to the supersaturation of carbonates in the sea water The studied Milaičiai-103 core section is located in the western Lith- (Kozłowski, 2015). uanian part of the Baltic (or Baltoscandian) Silurian Basin, in the transi- The described patterns of community collapse and the tional zone between the shallow water carbonate facies mostly palaeoenvironmental change in Ludlow were explained by characterized by high abundance of shelly taxa and the deep water fa- interchanging oceanic and climatic states (humid vs. arid periods) cies characterized by high abundance of graptolites and dominance of that influenced the rates and modes of oceanic circulation regimes argillaceous sediments – mudstones and organic rich clays (Fig. 1). (Jeppsson, 1987; Samtleben et al., 1996; Jeppsson et al., 2012)oras The Silurian Baltic Basin was an epeiric sea deepening toward the mod- a result of effects of sudden, high magnitude sea level changes ern south-west direction and spanning the territories of southern Scan- (Loydell et al., 2001), most probably of glacioeustatic origin (Calner dinavia, Baltic countries, Kaliningrad District (Russia), eastern Poland and Eriksson, 2006; Lehnert et al., 2007). Better understanding on and westernmost part of the Ukraine (Einasto et al., 1986; Paškevičius, causes of the observed global perturbations is hampered by the 1997; Lazauskienė et al., 2003; Jeppsson et al., 2006; Koren' and lack of quantitative palaeoecological studies based on integrated Suyarkova, 2007; Kaljo et al., 2014). The greatest thickness of strata palaeontological and geochemical records. This contribution pre- and the most complete successions are found in the deeper part of the sents a high-resolution integrated stratigraphical–quantitative basin comprising Poland, western Lithuania, and the Kaliningrad Dis- palaeoecological study of conodonts, brachiopods, and stable carbon trict. Usually very good to excellent preservation of conodonts and isotopes of the Ludlow in the shelf succession of the Lithuanian part very low colour alteration index (≈1) (Männik and Malkowski, 1998) of the Baltic Silurian Basin. The main objective of this paper is to show that the Silurian strata of this basin are unaltered and therefore quantitatively test the influence of the oceanic events on the com- suitable for palaeoenvironmental and palaeobiological research. munity dynamics of conodonts and brachiopods using constructed The Dubysa, Mituva and Ventspils formations are distinguished in comprehensive stratigraphical framework. In order to achieve this the investigated interval of the Milaičiai-103 borehole (Paškevičius et goal, a new numerical technique that quantifies and tests the signif- al., 2012) and this lithostratigraphic subdivision is used in the present icance of temporal recurrence patterns of palaeocommunities was study (Fig. 2). The geological section of the investigated interval is main- developed and tested on the present data set. ly composed of dolomitized marlstones, nodular limestones and wavy-

Fig. 1. Palaeogeographic map (Bassett et al., 1989) showing the location of the Milaičiai-103 borehole. 274 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288

Fig. 2. Distribution of graptolites, and conodonts, and δ13C trend. From left to right: series, stages, regional stages, formations and biozones. δ13C chemostratigraphical zones were named using nomenclature of Frýda and Manda (2013): R-Zone - the zone of the rising values; S-Zone - the zone of the stably high values; and F-Zone is the Zone of long exponentially falling values. bedded marlstones. There is a pervasive decreasing trend in clayey ma- Bohemograptus praecornutus Urbanek, B. cornutus Urbanek and long- terial content upward in the studied interval. ranging P. frequens, B. b. bohemicus, Pseudomonoclimacis tauragensis (Paškevičius), B. b. tenuis (Bouček) is interpreted as the leintwardinensis 3. Overview of graptolite biostratigraphy in the core Biozone. Previously this interval was subdivided in to the incipiens, praecornutus and cornutus biozones (Paškevičius et al., 2012). The ﬁrst The graptolite biostratigraphy based on Paškevičius et al. (2012) is appearance of S. incipiens is related to the upper part of the scanicus revised in this study. The graptolite fauna is rare in the studied interval Biozone and the species ranges into the middle part of the of the Milaičiai–103 well and the graptolite biozones are not very well leintwardinensis Biozone in Wales (Zalasiewicz et al., 2009), Bohemia constrained. (Štorch et al., 2014) and the Kaliningrad District (Koren' and In the lower part of the investigated section, Saetograptus chimaera Suyarkova, 2007). B. praecornutus appears in the upper part of the chimaera (Barrande), Lobograptus ex. gr. scanicus (Tullberg) and long scanicus Biozone in the southern Tien Shan, Kyrghizstan (Koren' and ranging Pristiograptus frequens (Jaekel), Bohemograptus bohemicus Sujarkova, 2004) and in the Kaliningrad District, but occurs in the bohemicus (Barrande), Pseudomonoclimacis haupti (Kühne), leintwardinensis–tenuis interval in Bohemia (Štorch et al., 2014). Ac- Neodiversograptus beklemishevi Urbanek and Colonograptus sp. were cording to Urbanek (1970), B. praecornutus appears in the uppermost identiﬁed. This assemblage is referred to the progenitor-scanicus part of the leintwardinensis Biozone of early Ludfordian age. He also dis- Biozone of Gorstian Age, with depth interval between 1301 and tinguished the new praecornutus Biozone within the former 1269 m (Fig. 2). leintwardinensis Biozone (Urbanek and Teller, 1997). In Arctic Canada, The interval with Saetograptus cf. incipiens (Wood), S. B. praecornutus appears above the leintwardinensis Biozone and the in- leintwardinensis (Lapworth), Pristiograptus tumescens (Wood), terval is regarded as the praecornutus – tenuis Biozone (Lenz and A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 275

Kozłowska-Dawidziuk, 2004). The short ranged B. cornutus is related to Brachiopods were sampled more densely than conodonts, on aver- the tenuis Biozone in Bohemia (Štorch et al., 2014). Urbanek (1970) distin- age every 0.54 m (σ±0.61 m) in the depth interval between 1301 guished the cornutus Biozone that represents a part of the tenuis interval in and 1160.6 m, comprising a series of 259 samples and spanning a slight- the Mielnik IG–1 well (Poland). According to the data from southern Tien ly shorter interval (in the upper part of the section) than the conodont Shan and Kyrghizstan (Koren' and Sujarkova, 2004), the situation is sampling interval (Fig. 3). All other sampling procedures were similar different: B. cornutus appears in the linearis (=leintwardinensis) Biozone to the conodont sampling procedures. The mass of bulk rock samples and ranges through the tenuis–kozlowskii interval. Although the details was at the order of several hundred grams, being attempted to keep are not in a good agreement, the interval with the mentioned graptolite the mass of the individual samples in a narrow range (although the taxa represents the leintwardinensis Biozone, here distinguished in the exact masses were not recorded during the time of sampling during interval between 1269 and 1254 m. the late 1990s). Argillaceous samples were macerated using 10–15%

The interval with Ps. tauragensis and B. b. tenuis in the upper part of H2O2 solution. The limestones and dolostones were processed using the Dubysa Formation (Fig. 2) is interpreted as the tenuis Biozone. Grap- mechanical disintegration techniques (Musteikis and Paškevičius, tolites are rare above the tenuis Biozone of mid- Ludfordian Age. 1999). The abundance of brachiopods was measured as the number of Slovinograptus balticus (Teller) has been identiﬁed in the middle part specimens with both valves plus the number of more common and Monograptus valleculosus Tsegelniuk in the upper part of the Mituva disarticulated valves (dorsal or ventral). Although differences in rock Formation. This interval represents the formosus Biozone. composition undoubtedly inﬂuence the original abundance signal and The upper boundary of the Ludfordian Stage and the Ludlow could increase the noise level, it was shown that this factor doesn't preclude not be drawn on the basis of graptolites in the Milaičiai–103 well, be- detection of the most robust patterns (Spiridonov et al., 2016; cause of lack of a lack diagnostic graptolite species. The studied interval Spiridonov, 2017). Additionally, the abundance information is studied comprises the Dubysa and Pagėgiai regional stages. The Dubysa/ from the standpoint of compositional taxic changes in this paper and Pagėgiai boundary in the Lithuanian part of the Baltic Silurian Basin is this is far less impacted by the sample size variation than the ap- marked by the upper boundary of the tenuis Biozone (here distin- proaches which analyze trends in absolute abundance (Musteikis and guished at depth of 1234 m) and corresponds to the last appearance Paškevičius, 1999). The conodont and brachiopod collections used in of conodont Polygnathoides siluricus (Brazauskas, 1993; Brazauskas et this study are stored in the Geological Museum of the Vilnius University. al., 2004; Martma et al., 2005) and to the lithostratigraphic boundary The drill core examined in this study is stored in the Vievis core facility between the Dubysa and Mituva formations (Lapinskas et al., 1985; in Vievis city of the Lithuanian Geological Survey. The detail information Paškevičius et al., 1994). on the composition of conodont and brachiopod assemblages can be accessed in the Supporting Information. 4. Materials and methods

4.1. Palaeontological data 4.2. Stable carbon isotopic data

Overall, 126 conodont samples were taken on average every 1.16 m For the purpose of stable carbon and oxygen isotopic (σ±0.52 m) in the depth interval between 1301 and 1155 m of the chemostratigraphy, 143 bulk carbonate samples were analyzed, taken Milaičiai-103 core, spanning most of the Ludlow, except for its lower- from the depth interval between 1301 and 1155 m. The samples were most part where the core was missing (Fig. 2). Samples for the conodont taken from the micritic matrix avoiding secondary structures (i.e. preparation were obtained by cutting core fragments in half. The diam- veins). For each sample, approximately 2 g of rock was powdered. Sam- eter of the core was 7.7 cm. The thickness of the sampled interval at each ples were analyzed using mass spectrometer Delta V Advantage using sampling point was in order of 10 to 15 cm. The sample weight ranged GasBench II by Thermo Scientiﬁc for preparation of gases. The interna- from 235 up to 820 g (average 565 g and σ±126 g). The processing of tional standards that were used for raw data calibration are: NBS 18, samples was performed using the standard buffered formic acid (10%) NBS 19 and LSVEC. The accuracy of analyses is in order of σ=0.02‰. dissolution technique (Green, 2001). The dried residues were divided The mass spectrometry analyses were conducted in the Department of in several granulometric fractions. The conodont elements were manu- Geology, the University of Tartu. Only the stable carbon isotopic data ally picked. SEM micrographs of stratigraphically important conodonts are applied in this contribution. The stable carbon and oxygen isotopic were taken in the Nature Research Centre (Vilnius) using the open ac- data can be found in the Supporting Information. cess research facility. Conodont abundance is an excellent proxy that can be used in deciphering complex patterns of biotic, palaeoclimatic and 4.3. Depth constrained clustering analyses and zonation of palaeoceanographical changes in the mid-Palaeozoic (Girard et al., palaeocommunities 2014; Girard and Renaud, 2007; Huang and Gong, 2016; Spiridonov et al., 2015; Spiridonov et al., 2016). Although hydrodynamic and In order to reveal the compositional evolution of conodont and bra- bioturbational processes can bias the composition and abundances of chiopod assemblages stratigraphically constrained cluster analysis was conodonts (and other fossils) by spatially and temporally homogenizing performed using the CONISS algorithm (Grimm, 1987). Additionally, (smoothing) the original composition (McGoff, 1991; Purnell and testing of the statistical signiﬁcance of the inferred clusters was execut- Donoghue, 2005; Olszewski, 2012), the original conodont micro- ed using the Monte Carlo procedure (Bennett, 1996). In order to stabi- biofacies can be traced even at local scales (Vierek and Racki, 2011). lize variance and reduce the effects of sampling inequality before the

Moreover, the between-sample volatility of conodont abundance in analysis, the count data were transformed using log10(N+1) function, lithologically and biofacially monotonous sections is usually high and where N isthenumberofconodontelementsornumberofbrachiopod this rejects facies control as a significant biasing factor at the sampling individuals of a given taxon in a sample. The constrained clustering and resolution employed here. The distance between neighbouring samples the statistical testing of distinctness of clusters were performed in the ‘R’ (N1 m) makes the influence of bioturbation rather unlikely. The number computational environment using package ‘rioja’ (Juggins, 2015; R of conodont elements per apparatus in different species varies in a nar- Development Core Team, 2015). Inferred statistically distinct clusters row range (15 to 19) (Purnell and Donoghue, 2005; Dzik, 2016). There- were later described as discrete fossil assemblages and named and fore, frequency of conodont elements in a sample likely reflects (is numbered according to the order of their stratigraphical appearance proportional to) the original abundance and thus the community struc- and taxonomic identity (i.e. LC1 - Ludlow conodont assemblage N°1; ture of this group. LB1 - Ludlow brachiopod assemblage N°1). 276 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288

Fig. 3. Stratigraphic distribution of brachiopods, stratigraphy, lithology and δ13C trend.

4.4. Recurrence and joint recurrence plots and recurrence quantification where R is a squared recurrence matrix, dij =1−Cij is a compositional analyses distance between two compared samples at the depths i and j, θ(⋅)is the Heaviside step function, which assigns “1” (black in the plot) if the In order to describe the compositional evolution of conodont and bra- difference is smaller or equal to the threshold value (ε)or“0” (white chiopod fossil assemblages, so-called recurrence plot and recurrence quan- in the plot) otherwise. Compositional distance is based on the modified tification analysis were employed. In the simplest form (as used here) the Morisita-Horn compositional overlap index (Cij)(Horn, 1966). Modifi- recurrence plots are square matrices in which all states in a time series (or cation is introduced in order to handle barren samples. If at least stratigraphical depth series) are compared to all other states using the one of two compared samples in our modification have a zero diversity predefined distance measure and a filtering threshold (Marwan et al., (S = 0), the index has the “0” value. Alternatively it could be coded that 2007; Webber and Marwan, 2015). White points in a plot show that two the samples with zero diversity can be similar, but it would imply a sim- states are different, and black points show that the compared states (e.g. ilarity which is based on the absence of positive sightings, which is two conodont associations from different depths) are very similar or recur- much easier to reach from any starting composition of an assemblage ring (e.g. Fig. 9A, B). The darker areas in a recurrence plot reveal a greater (therefore this strategy is much more prone for false positives). On the similarity to other assemblages. The recurrence and cross recurrence tech- other hand, here we adopted a concept of similarity which is based on niques were successfully applied using conodont abundance in the positive evidence. Therefore, if two assemblages do not have any deciphering the complex dynamical patterns of the Mulde event and si- species, and therefore don't have shared species, they are dissimilar. multaneously correlating the geographically widespread conodont time This index varies between “0” if there is no similarity between sam- series (Spiridonov, 2017), thus proving the potential of this approach in ples and “1” when samples are completely similar. The threshold quantifying and testing the effects of global events on the regional biota. values (ε) in the calculation of the recurrence plots were automati- Mathematically, the recurrence of one compositional state at depth i cally optimized in order to achieve a 20% global recurrence rate (or to a compositional state at the depth j is calculated using the formula: RR, which is the percentage of recurrence points out of all points in the matrix) – apercentagesufficiently large in order to perceive recurrence patterns in compositionally non-stationary sequence of ðÞ¼ε θε− ; ; ¼ ; …; ð Þ Ri; j dij for i j 1 N 1 assemblages. A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 277

For the purpose of comparing compositional trajectories of cono- In an analogous way, we can define the centrality of recurrence for dont and brachiopod palaeocommunities we employed the so-called the whole matrix (CENM): joint recurrence plot methodology. The recurrence plots compare X the similarity in compositional changes of a single group of organ- N NCRj ¼ vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffij¼1 ; isms, while the joint recurrence plots check whether or not in the CENM u u XN XN two different recurrence plots of equal dimensions, recurrence 2u 1 2 N tX t −t ; ; occurs jointly between the same compared temporal points (hence N CRij C diag j ð4Þ NCRj −1 j¼1 i¼1 the name). Mathematically (based on the formalism presented in j¼1 Marwan et al., 2007, the joint recurrence plot (or squared matrix XN XN for N N diag ! ! ! ! CRj Rj x ; y x y ¼ ¼ JR i; jðε ; ε Þ ) is described as a Hadamard product of two j 1 j 1 recurrence plot matrices:

Here NCRj is the number of recurrence points in a column j, tCRij is the row number of recurrent points in a column j, and t is the row ! ! ! ! ! ! ! ! C,diag ,j x ; y x y x x y y P JR ; ε ; ε ¼ θε −d θε −d ; for i; j ¼ 1; …; N ð2Þ N i j ij ij number of a diagonal point of the recurrence matrix in a column j. j¼1

diagRj is the number of diagonal recurrence points in the matrix. If there are no samples with zero abundance in the studied section, the later measure equals to the N, otherwise to the number of productive ! ! samples. Here ε x is the filtering threshold of the firstrecurrenceplotandε y is ! The localized application ofcentralityofrecurrenceCENC can elucidate fi x the ltering threshold of another recurrence plot, analogously d ij and which fossil assemblages are transiently stable (since this measure in- ! y creases if neighbouring assemblages are very similar), and simultaneous- d ij are distances of two compared recurrence plots between the given time points. A joint recurrence plot is an ideal tool for revealing ly non-recurring in their exact composition at later times (since the similarities in the dynamical behaviour of two very different systems measure decreases if the average distance between recurring assem- (i.e. conodont and brachiopod palaeocommunities), which simulta- blages increases). The assemblages that are recurring after transient dis- neously unfold in two directly incommensurable state spaces (i.e. cono- appearances would have high RRC (recurrence rate in the column) and dont element abundances vs. brachiopod individual abundances). Since low CENC. The third category of assemblages is anomalous assemblages fi conodonts were sampled more sparsely than brachiopods, in the joint (extremely time speci c) having low RRC and also low CENC. The comple- „ ” recurrence analysis, their abundance values were interpolated for bra- mentary CENM value shows overall compactness of recurrence for the fi chiopod sample depths. The original 20% recurrence rate was used for whole set of assemblages and reveals the degree of time-speci city of as- the conodont and brachiopod recurrence matrices as input of the joint sociations for the whole studied time period. Higher values of CENM indi- recurrence analysis. cate a high degree of short-term autocorrelation and non-stationarity and “ ” If the fossil assemblages experience long term trends or irrevers- non-repeatability of an observed process (i.e. its evolutionarity ). fi ible turnovers in composition, we should expect that most of the re- In order to test the signi cance of CENC and CENM estimates for cono- currence points will occur near the diagonal of the plot (assemblages dont and brachiopod recurrence matrices and their joint recurrence ma- of very different age will have diminishingly smaller probability of trix, a null modelling procedure using the Monte Carlo approach was being similar). On the other hand, if perturbations in the composi- performed. From the time series of a species, lag-1 autocorrelation coef- fi ρ fl tion of a fossil community are temporary and reversible, we should cient ( ) was estimated, then the time series was shuf ed, later on expect a more even distribution of recurrence (black) points in a re- detrended (to make values symmetrically distributed around zero), currence plot. For this purpose, a new recurrence quantification sta- and this new sequence of values (x) was used to construct a new ρ fi tistic is introduced here – the so called centrality of recurrence (CEN) autocorrelated time series: yt = yt−1 +xt. At the nal stage, in order ≥ metric. It shows how much recurrence there is and how tightly it is to ensure that all abundances in the simulated time series are ( 0), distributed near the main diagonal of a recurrence plot. It increases the modulus (or absolute value) of the smallest number |ymin |was with the recurrence rate, showing that a given state is similar with added to all members of the time series. This procedure was performed other states, and decreases if those similar states are far away in for all species in the matrices, in order to simulate a group of time series time (or stratigraphical depth) from the given state. In other that is random although parameterized (histograms of abundances and words, it is the product of recurrence rate, with the reciprocal of autocorrelations are used in the simulation) by the data. Overall 100 the standard deviation of recurrence points from the main diagonal random species occurrence matrices were generated for each dataset of recurrence matrix. The centrality of recurrence in a single column, according to the described model. All null model matrices were scaled of the recurrence or joint recurrence matrix is mathematically de- to have 20% recurrence rate (the same as for the empirical matrices). fined as: For each random matrix, the vector of CENC values and CENM value were estimated and their assemblages were used in testing non-ran- domness of revealed empirical patterns. If the fossil assemblages of spe- ¼ vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiNCR ; N ; ¼ cies are truly temporally organized and their abundance shifts are not CENC u for NCR 1 otherwise CENC 0 u XN just a passive result of coinciding random walks, they should have t 1 − 2 N tCRi tC; diag higher values of centrality of recurrence than the shuffled and NCR−1 i¼1 autocorrelated (and thus temporally non-coherent) data. ð3Þ Recurrence and joint recurrence matrices, as well as centrality estimates were calculated using code written in the Perl programming language. The code and the instructions for its use can be found in the Here CENC is the centrality of recurrence for a given column of Supporting Information. recurrence matrix, NCR is the number of recurrence points in a studied column, N is a length of a vector (number of rows in a column), tCRi is 4.5. Measuring cosmopolitanisms of fossil assemblages the row number of a recurrent points in a column, and tC,diag is the row number of a diagonal point of the recurrence matrix in a given In order to explore possible determinants of recurrence changes, column. the rich brachiopod dataset was explored in the light of 278 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 palaeobiogeography. Each brachiopod fossil sample was evaluated for graptolite Biozone. This zone is also characterized by two other time-re- the genus level cosmopolitanism. The proposed measure (cosmopoli- stricted conodont species, Coryssognathus dubius (although, on Sardinia tanism of a sample) is mathematically defined as follows: it ranges into the Přidoli, Serpagli et al., 1996) and massive Silurognathus maximus (Fig. 4). The latter species was previously suggested to be en- − 1 Κ ¼ S 1 1− P ð5aÞ demic to Gotland and restricted to the lower or middle part of the P. n β i¼1 pi i siluricus Zone (Jeppsson, 2005; Jeppsson et al., 2006). In the present material, this species occurs in the upper part of the P. siluricus Zone (Fig. ¼ − 1 ð Þ 2). Similarly, a probable fragment of S. maximus was found in the S 1 β 5b max upper part of the P. siluricus Zone in the Coral Garden Formation in northeastern Australia (Jeppsson et al., 2007). The present material Here, Κ (Kappa) stands for the genus level cosmopolitanism of the agrees with a broader geographical and chronological range of this spe- sample; n - the number of genera in the assemblage; pi - the proportion cies and confirms its relationship to the P. siluricus Zone. Contrast of the given genus in the assemblage; βi -thenumberof Panderodus recurvatus is shown in compilations by Jeppsson et al. palaeobiogeographic provinces in which a given genus is detected; S is (2012) as having its last occurrence in the upper P. siluricus Zone (Lau the scaling factor needed in order to ensure that the metric Κ will vary event) but it ranges into the uppermost part of the Ludlow in the section in the predetermined range ∈[0; 1]; and βmax is the number of provinces studied here (Figs. 2 and 5). into which the global distribution of a studied taxon is subdivided (in The Ozarkodina snajdri Interval Zone is distinguished in the depth in- case of the late Silurian brachiopods, βmax =6). If an assemblage is terval 1236–1180 m. In the studied section, the nominal taxon is present monoprovincial, the proposed metric is equal to zero. In an opposite only in the uppermost part of this interval. The upper part of O. snajdri case, very high levels of cosmopolitanisms when all genera occupy all Interval Zone completely encompasses the whole observed range of provinces, this measure converges to the unity. The numbers of Ozarkodina scanica in the depth interval 1200.7–1190.1 m. In the stud- palaeobiogeographical provinces, in which the given genera were de- ied section and elsewhere in Lithuania (Brazauskas, 1989; Paškevičius tected, were determined according to the latest compilation of et al., 1994; Brazauskas et al., 2004) and Sweden, O. scanica is restricted palaeobiogeography of the upper Silurian brachiopods (Jia-Yu et al., to the upper part of the O. snajdri Interval Zone (Jeppsson, 1974). There- 1995). The late Silurian palaeobiogeographical spread of genera, not fore, the interval between 1200.7 and 1180 m is described here as the O. listed in this compilation, was determined using the Paleobiology Data- scanica Subzone of the O. snajdri Interval Zone. base (accessed March 31, 2017). In order to obtain a long term trend, The uppermost interval of the stratigraphic succession is assigned to the time series of brachiopod assemblage cosmopolitanism was fitted the Ozarkodina baccata–Ozarkodina crispa Zone. This Zone was distin- to non-parametric local regression smoothing (LOWESS) performed in guished in the depth interval between 1180 and 1155 m, based on the the R environment (R Development Core Team, 2015). Similar analyses distribution of O. baccata. Although O. baccata appears slightly earlier were not performed for conodonts since to this date there is no compre- than O. crispa in the Bohemian section (Slavik and Carls, 2012), it is hensive studies of their paleogeography in the late Silurian. apparently restricted to the uppermost Ludlow in the British sections and is contemporaneous with O. crispa (Miller and Aldridge, 1997; 5. Results Märss and Miller, 2004). This species is also restricted to the uppermost Ludfordian in the Vilkaviškis-134 section in Lithuania 5.1. Conodont stratigraphy (Radzevičius et al., 2016) and in Estonian sections (Märss and Männik, 2013). Due to the absence of indicative species (i.e. O. crispa The abundant conodont material (3596 conodont elements) yielded and undescribed species of the O. remscheidensis s.l. plexus: in the course of processing of the Ludlow section of the Milaičiai-103 (Jeppsson and Aldridge, 2000)) in the present section, it is impossi- core enabled a robust chronostratigraphic subdivision of the section ble to locate the exact position of the Klev Event and the lower (Figs. 2, 4, and 5). The interval between 1301 m and 1246 m is assigned boundary of the Přidoli. here to the Kockelella variabilis Interval Zone corresponding to the Gorstian and the lowermost part of the Ludfordian stages, equivalent 5.2. δ13C stratigraphy to the international K. crassa, K. variabilis, and A. ploeckensis conodont zones (Jeppsson et al., 2006; Melchin et al., 2012), and corresponds to The δ13C trend (Fig. 2) in the Milaičiai-134 core is very similar to the progenitor-scanicus, leintwardinensis and lower part of tenuis grapto- other relatively complete offshore carbonate platform records in the lite Biozones. The composition of the conodont fauna in the lower part Ludlow (Lehnert et al., 2003; Bickert et al., 1997; Kaljo et al., 1997; of the studied interval (1301–1270 m) certainly points to a Gorstian Barrick et al., 2010; Kozłowski and Sobień, 2012), particularly in the age for the sedimentary rocks. The time-specifictaxaareKockelella Viduklė-61 core in western Lithuania (Martma et al., 2005). The lower stauros and Wurmiella inflata; they accompanied by Kockelella ortus part of the curve (up to the end of the W. inflata Subzone) between absidata, which ranges up to the depth of 1237 m (into the Ludfordian 1301 and 1267 m is characterized by δ13Cvaluesfluctuating around Stage). The disappearance of both time-specific species slightly pre- 0‰. This interval is followed by a sharp decline to −1.68‰ in the cedes the overall sharp drop in conodont and brachiopod numbers depth interval 1267–1247 m. This interval of low stable carbon isotopic and an increase in graptolite abundance and diversity in the depth inter- values is characterized by very low conodont and brachiopod abun- val 1265–1253 m. This disappearance interval is most probably related dance (most samples are barren) and moderate evolutionary turnover to the Linde Conodont Extinction Event in the middle part of the of conodont species (see Fig. 2). The minor positive carbon isotopic ex- Kockelella variabilis Interval Zone (Jeppsson and Aldridge, 2000). The cursion (up to 0.51‰) between 1272 and 1266 m (Fig. 2) and perturba- short interval between 1271 and 1270 m is distinguished here as the tion in the conodont assemblages probably correspond to the Linde Wurmiella inflata Subzone of the Kockelella variabilis Interval Zone event. Similar excursion which is associated with the Linde event was (upper-most part of the progenitor-scanicus graptolite Biozone). W. detected in Gotland (Samtleben et al., 2000; Cramer et al., 2011), Latvia inflata apparently has stratigraphic significance as its range ends at the and Lithuania (Martma et al., 2005; Kaljo and Martma, 2006), and beginning of the Ancoradella ploeckensis Zone (early Ludfordian) in the Ukraine (Kaljo et al., 2007). The interval of negative δ13Cvalues Sardinian sections (Corriga et al., 2009). (1275–1254 m) most probably marks the so-called Etelhem Secundo The Polygnathoides siluricus Zone is distinguished in the depth inter- Episode of Jeppsson and Aldridge (2000). val 1246–1236 m based on the observed distribution of the nominal The episode of negative δ13C was followed by the mid-Ludfordian species. It corresponds to the central and the upper parts of the tenuis (Lau) positive carbon isotopic excursion that is confirmed by the A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 279

Fig. 4. SEM micrographs of conodonts from the Milaičiai-103 core. A, Ozarkodina baccata: A), Pa element, no. VU-CON-MIL103-001, depth 1173 m. B, E, J, Wurmiella excavata:B)Pbelement, no. VU-CON-MIL103-002, depth 1174 m, E) Pa element, no. VU-CON-MIL103-003, depth 1238 m, J) Pb element, no. VU-CON-MIL103-004, depth 1238 m. C, R, U, Polygnathoides siluricus:C) Pa element, no. VU-CON-MIL103-005, depth 1246 m; R) Pb element, no. VU-CON-MIL103-006, depth 1236 m; U) Pa element, no. VU-CON-MIL103-007, depth 1237m.D,G,Ozarkodina snajdri: D) Pa element, no. VU-CON-MIL103-008, depth 1182 m; G) Pa element, no. VU-CON-MIL103-009, depth 1172 m. F, N, Oulodus elegans: F) Sc element, no. VU-CON-MIL103-010, depth 1173 m; N) Sa element, no. VU-CON-MIL103-011, depth 1173 m. H, Wurmiella inﬂata: H) Pa element, no. VU-CON-MIL103-012, depth 1170 m. I, Kockelella stauros:I)Paelement,no. VU-CON-MIL103-013, depth 1170 m. K, L, P, Q, Kockelella variabilis: K) Sb element, no. VU-CON-MIL103-014, depth 1176 m; L) M element, no. VU-CON-MIL103-015, depth 1176 m; P) Pa element, no. VU-CON-MIL103-016, depth 1176 m; Q) Pb element, no. VU-CON-MIL103-017, depth 1176 m. M, Coryssognathus dubius: M) Sc element, no. VU-CON-MIL103-018, depth 1237 m. O, Silurognathus maximus: O) Pa element, no. VU-CON-MIL103-019, depth 1238 m. S, T, Ozarkodina scanica: S) Pa element, no. VU-CON-MIL103-020, depth 1197.4 m; T) Pa element, no. VU-CON-MIL103-021, depth 1197.4 m.

distribution of the zonal conodont species, i.e., P. siluricus (Fig. 2). In the or the F-Zone (1226–1188 m). The maximum of this excursion (5.02‰) chemostratigraphic classiﬁcation proposed by Frýda and Manda (2013), occurs at the depth 1229 m. the stable carbon isotopic anomaly was subdivided into three stable car- The mid-Ludfordian stable carbon isotopic excursion is followed by a bon isotopic zones: the rising or the R-Zone in this section corresponds decrease in δ13C values in the depth interval 1188–1177 m. The interval to the 1247–1232 m interval; the zone of stably high values or the S- between 1169 m and 1155 m is characterized by the δ13C values ﬂuctu- Zone (1232–1226 m); and the zone of long exponentially falling values ating around 0‰. 280 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288

The constrained cluster analysis of brachiopod collections revealed six statistically distinct assemblages (Fig. 7). The first assemblage (LB1) is tied to the K. variabilis Zone (1301–1246 m). The most important species of this assemblage are Glassia obovata (36.51%), Dayia navicula (31.71%), Aegiria grayi (13.98%), and Coelospira prunum (6.99%). This association represents a BA5 type benthic assemblage, indicative of a quiet water (low turbulence) possibly poikiloaerobic offshore environment (Boucot, 1975; Musteikis and Paškevičius, 1999). The upper part of this stratigraphic interval (1264.8–1253.1 m) is characterized by very low abundance of brachiopods, containing barren intervals that suggesting possible anoxic episodes that were unfavourable for the benthic fauna. Overall at least 18 different brachiopod species were found in this assemblage. The second brachiopod assemblage (LB2) almost matches the P. siluricus Zone, reaching slightly higher in the section (1246– 1233.5 m). This low diversity (in comparison to neighbouring clusters) assemblage is heavily dominated by Cromatrypa? pubes, constituting 75.69% of specimens. Other important taxa are Aegiria grayi (7.83%) Fig. 5. Translucent light micrographs of conodonts from the genus Panderodus from the and an undescribed smooth-shelled brachiopod species (5.77%). The Milaičiai-103 core. A) P. recurvatus, no. VU-CON-MIL103-022, depth 1196 m; B) P. aff. dominant species can be found in a wide range of palaeodepths (BA5 gracilis, no. VU-CON-MIL103-023, depth 1180 m; C) P. unicostatus, no. VU-CON-MIL103- to BA2), but it is most abundant in the BA4, and it is a characteristic 024, depth 1238 m; D) P. equicostatus, no. VU-CON-MIL103-025, depth 1221 m. taxon of restricted dysaerobic environments with increased flux of ter- Abbreviations: ewt – end of the white tip; ebc – end of the basal cavity. rigenous material (Musteikis and Modzalevskaya, 2002). Overall at least 17 species were detected in this interval. The third brachiopod cluster (LB3) is confined to the lower part of 5.3. Conodont and brachiopod assemblages the O. snajdri Interval Zone (1233.5–1226.3 m). The numerically important species in this assemblage are Dayia navicula (38.01%), Isorthis The depth constrained clustering analysis revealed three composi- amplificata (23.96%), Coelospira prunum (8.05%) and Protochonetes tionally distinct conodont fossil assemblages, LC1, LC2 and LC3 (Fig. 6). stoniskensis (7.23%). The Dayia-Isorthis community inhabits environ- The first Ludlow conodont assemblage LC1 is recorded in the interval ments of moderate terrigenous material supply, lowered turbidity and 1301–1237 m and corresponds to the K. variabilis Interval Zone and normal oxygen supply, occupying BA3–4 benthic assemblage zones most of P. siluricus Zone, being equivalent to the Gorstian and lower (Musteikis and Paškevičius, 1999). At least 24 brachiopod lineages Ludfordian stages. Although this assemblage contains also short ranged were detected in this assemblage. conodont taxa, it is strongly dominated by four long-ranging open ma- The fourth brachiopod assemblage (LB4) occurs in a stratigraphically rine species: Wurmiella excavata (39.7%), Panderodus unicostatus very narrow interval (1226.3–1223 m) corresponding to the lower O. (21.5%), Ozarkodina confluens (15.25%), and Panderodus recurvatus snajdri Interval Zone. This community is characterized by the domi- (10.13%). Of all time restricted taxa in this cluster, species of the nance of Dicoelosia biloba (50.9%). Other important taxa are Isorthis genus Kockelella (K. variabilis - 5.39%, K. ortus absidata - 1.9%) are the amplificata (17.97%) that was also an important component in the pre- most abundant. The structure of LC1 assemblage reveals three major vious assemblage, Isorthis clivosa (6.65%), Coelospira prunum (6.15%) subclusters bounded by the disappearances of W. inflata, K. stauros and Skenidioides lewisii (5.49%). The Dicoelosia-dominated communities and Decoriconus fragilis at the end of the W. inflata Subzone, and are assigned to the BA4 benthic assemblage zone characterized by a low appearances of Belodella resima, P. siluricus, S. maximus and level of turbulence and sufficient oxygen supply (Musteikis and Coryssognathus dubius in the upper most K. variabilis and also P. siluricus Paškevičius, 1999). In the LB4 assemblage there was a similar number zones (Fig. 6). of brachiopod lineages (23) as in the assemblage LB3. The second conodont assemblage LC2 (1236–1183 m) corresponds The fifth brachiopod assemblage (LB5) spans the middle and the to the peak and falling limb of the Lau δ13C excursion in the middle upper parts of the O. snajdri Zone (1223–1180 m). This assemblage is Ludfordian Stage — the uppermost part of the P. siluricus Zone and dominated by Isorthis amplificata (24.35%), Skenidioides lewisii most of the Ozarkodina snajdri Interval Zone. This assemblage is highly (15.26%), Dicoelosia biloba (11.5%), and Shaleriella delicata (11.09%). distinct relative to all neighbouring assemblages, being heavily domi- The dominance of Isorthis and Skenidioides in a community suggests nated by Panderodus equicostatus (67.53%). Some subdominant taxa that the given assemblage formed under moderately turbulent condi- are almost the same as in the LC1: Wurmiella excavata (10.53%), tions and normal oxygen levels near the BA3–BA4 boundary Panderodus recurvatus (4.83%). Panderodus unicostatus accounts only (Musteikis and Paškevičius, 1999). At least 49 brachiopod lineages for 1.03% of conodonts in this interval and the abundance of Ozarkodina were detected in this fossil assemblage, making it the most diverse in confluens reaches 3.79%. The LC2 assemblage is characterized by the the studied interval. lowest conodont diversity of all three assemblages. The sixth brachiopod assemblage (LB6) corresponds to the O. The third conodont assemblage LC3 (1182–1155 m) comprises the baccata–O. crispa Zone (1180–1160 m). The dominant species of this as- upper Ludfordian part of the section (O. baccata–O. crispa Zone), being semblage is Isorthis ovalis (53.14%), accompanied by Dayia navicula characterized by the δ13Cvaluesfluctuating around 0‰. The upper (16.9%), and Shaleriella delicata (6.76%). Isorthis ovalis is typical of BA 3 Ludfordian conodont assemblage of this cluster is strongly dominated benthic assemblage that is characterized by moderate levels of water by Panderodus unicostatus (67.73%), which was also the second domi- turbulence and abundant oxygen supply (Musteikis and Paškevičius, nant species in the LC1 cluster and a subordinate species in the LC2 clus- 1999). This assemblage was relatively species poor with only 15 bra- ter. Other abundant taxa in this assemblage are Wurmiella excavata chiopod species. (9.06%), Ozarkodina ambigua (9.06%), Ozarkodina confluens (5.08%) An additional comparison of Shannon-Weaver diversity indices H and Oulodus elegans (4.57%). Similarly to the assemblages LC1 and (or distribution entropy) for six brachiopod assemblages reveals trends LC2, the zonal and other conodont species are numerically subordinate similar to that of species diversity: LB1 (H = 1.73), LB2 (H = 1.02), LB3 to these long ranging taxa. (H = 2.02), LB4 (H = 1.62), LB5 (H = 2.74), LB6 (H = 1.67). The A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 281

Fig. 6. Depth constrained cluster diagram of conodonts from the Milaičiai-103 core. Diameters of circles correspond to the abundance of a given taxon in a sample; horizontal lines represent the boundaries of statistically distinct conodont assemblages (LC1, LC2, and LC3). maximum entropy of the species abundance distribution was found in of conodonts and brachiopods is highly distinct from the random the assemblage LB5, which confirms its greatest diversity. model. CENM of the conodont and brachiopod joint recurrence matrix (Fig. 8, C) differs by 22σ (standard deviations) from the average of the 5.4. Patterns of recurrence of conodont and brachiopod communities random models. According to the Chebyshev's inequality this translates to the p≤0.002.This shows that the joint recurrence pattern of conodont The matrix centrality analysis revealed that Ludlow conodont com- and brachiopod faunas is highly time specific and strongly coordinated munities have CENM that is indistinguishable overall from the random between the groups. model, with slight tendency toward higher centrality (Fig. 8, A). This More details can be gathered from recurrence plots and time series suggests that we can detect compositionally similar assemblages, dom- of time specific centralities (CENC) in comparison with the random inated by the same species, throughout the intervals. On the other hand, models and stratigraphic boundaries (Fig. 9). The conodont recurrence the brachiopod assemblages have much higher centrality (CENM)than plot and the time series of CENC (Fig. 9, A, C) show that most of the O. the random model (Fig. 8, B). Moreover, the joint recurrence pattern snajdri Interval Zone (except its lowermost and uppermost parts) was 282 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288

Fig. 7. Depth constrained cluster diagram of brachiopods from the Milaičiai-103 core. Brachiopods were seriated according to the patterns of species co-occurrences. Diameter of circle corresponds to the abundance of a given taxon in a sample; horizontal lines represent the boundaries of statistically distinct brachiopod assemblages (LB1, LB2 … LB6). characterized by an anomalous, very time speciﬁc and compositionally interval starts slightly earlier and ended slightly later in brachiopods. stable conodont assemblage. Exactly the same pattern can be observed The anomalous assemblages dominated the peak and falling limb of in the brachiopod recurrence plot (Fig. 9B, D), although the anomalous the Lau δ13C excursion (Fig. 9E, F). The reason why conodonts had

Fig. 8. Comparison of A) conodont; and B) brachiopod; and C) joint-recurrence CENM with ensembles of CENM of null models. Arrows show observed empirical values of CENM,histograms with the ﬁtted Gaussian probability density function shows results of the null model. A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 283

Fig. 9. Compositional similarity recurrence plots for depth ranked samples of A) the Ludlow conodonts; and B) brachiopods from the Ludlow part of the Milaičiai-103 core. CENC time series (in red) for C) conodonts; and D) brachiopods with null models (95% conﬁdence intervals; therefore the one sided (values greater than the null model) signiﬁcance 97.5%), which are represented as series of box and whisker plots. For the purpose of clarity of presentation null model, outlying values were deleted (the same in the Fig. 10). The “horseshoe” appearances of the null model ensembles are caused by the edge effect bias, which originates when at the beginning of the recurrence plot there is lower opportunity for a sample 13 composition to recur in close neighbourhood backward in time, and at the end of the plot to recur forward in time (therefore proportionally lower CENC). δ C values for E) conodont; and F) brachiopod sample depths. Since, in the lower part of the O. snajdri Interval Zone brachiopods were sampled more densely, corresponding part of the δ13C trend appears deformed (the same in the Fig. 10).

overall lower centrality (CENM) than brachiopods seems to arise from compositional trending of assemblages and their long-term non-repeat- the fact that brachiopods were sampled much more densely in the ability (“evolutionarity”). high centrality region (in the middle of the O. snajdri Interval Zone). In order to test the possibility of significance of local faunal “outages” Even more evident time uniqueness of the conodont and brachiopod (sensu (Morris et al., 1995; Brett and Baird, 1997) evidenced by the bar- assemblages in the interval of the Lau event can be observed in their ren samples, the two sided test of equality of proportions (barren sam- joint recurrence plot and the time series of their joint recurrences—CENC ples to productive samples) was performed in the R environment (R (Fig. 10). The most distinct feature is the recurrence interval in the O. Development Core Team, 2015). The observed patterns were compared snajdri Interval Zone characterized by very high joint recurrence rate to the frequency of barren samples from the Monte Carlo simulated fos- and essentially no similarity with other intervals. Similarly to the previ- sil assemblages. The results show that the probability of observing the ously described analysis using CENM, the evolutive analysis using the given frequency of barren conodont samples (19.8%) as a result of random CENC metric suggests that the assemblage dynamics of nektobenthic coincidences of absences of the individual species is diminishingly small conodonts and benthic brachiopods was highly coordinated. A great (P=2.2⋅10−16). The same probability was obtained for brachiopods majority of the joint recurrence was concentrated near the diagonal of (with 30.5% of barren samples in the sample series). These results confirm the plot. Almost all values of CENC for fossiliferous samples are statisti- the significance of co-recurrence of conodonts and brachiopods but also cally distinct (much higher) from the null model. This shows general the pattern of their joint absences that are highly coincident as well. 284 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288

5.5. Patterns of cosmopolitanism in brachiopod assemblages

The analysis of changes in generic cosmopolitanism of brachiopod assemblages revealed a remarkable pattern that is closely tied to the Lau event. The Gorstian and early Ludfordian (LB1 and LB2 brachiopod assemblages) were dominated by remarkably provincial compositions (average Κ=0.38). The lowest value of cosmopolitanism was reached in the upper part of the K. variabilis and in the P. siluricus zones, with two long-lasting episodes of brachiopod and conodont minima (Fig. 11). The brachiopod cosmopolitanism changed from low to high state (average Κ=0.84) at the depth 1231 m in the lower part of the O. snajdri Interval Zone (or LB3 brachiopod assemblage) and remained high until the end of Ludlow (Fig. 11).

6. Discussion

The conodont succession of the studied section accords well with the Laurentian (Barrick et al., 2010; Barrick et al., 2011), Australian (Jeppsson et al., 2007; Mathieson et al., 2016), Carnic Alps (Corradini et al., 2015), and especially Gotlands (Jeppsson, 1974; Jeppsson et al., 2006; Jeppsson and Aldridge, 2000; Melchin et al., 2012), Estonian (Märss and Männik, 2013) and Lithuanian (Martma et al., 2005) stratigraphical schemes of the Ludlow. Although, some zonal species (Ancoradella ploeckensis, Ozarkodina crispa), are absent, the presence of other time-speciﬁc taxa (i.e. O. baccata, Oulodus elegans), as well as the shape of the δ13C curve, suggest that the section is rather complete. As evidenced by the numerical analyses, several reorganizations in both conodont and brachiopod communities took place in the Ludlow. The most distinct pattern in the recurrence plots of conodonts and brachiopods is the appearance of highly time speciﬁc communities in the aftermath of the Lau extinction. The most pervasive feature of the conodont assemblage of the anomalous interval (corresponding to most of the O. snajdri Interval Zone ≈ LC2 conodont assemblage) is the dominance of the coniform P. equicostatus. This species (and comparable forms), based on the detailed studies on Gotland (Jeppsson, 2005), Lith- uanian (Radzevičius et al., 2014; Radzevičius et al., 2016), Latvian (Loydell et al., 2010) and Polish (Jarochowska and Munnecke, 2015) materials, thrived in and later disappeared from the Baltic Basin in the late Homerian, reaching high concentration in the post-extinction strata of the Mulde Event (Jeppsson and Calner, 2002). Later on, in the late Ludfordian, P. equicostatus re-appeared in substantial quantities in the sections of Baltica (Jeppsson, 1998; Jeppsson and Aldridge, 2000; Jeppsson, 2005; Jeppsson et al., 2012) and Australia (Jeppsson et al., 2012). A similar general pattern is recorded in the mid-North American sections where the species is absent in the lower and the middle parts of Ludlow but re-appears in the condensed sections of Laurentia (Barrick et al., 2010; Barrick et al., 2011). The re-appearance is slightly delayed,

Fig. 10. A) Compositional joint recurrence plot of ranked samples for brachiopods and conodonts interpolated to the depths of brachiopod samples. B) Comparison of CENC for the ranked samples of the joint recurrence plot with the null model, which is represented by the series of box and whisker plots (the same legend as in the Fig. 9). C)

Comparison of CENC for brachiopod and conodont recurrence plots interpolated to brachiopod sample depths. D) δ13C values for the brachiopod sample depths. Fig. 11. The LOWESS regression trend (1) fitted to the trend in brachiopod cosmopolitanisms (2), compared with the Ludlow brachiopod assemblage zones, and conodont zones, regional stages and global series. There is sharp transition to the much higher (almost two times) levels of cosmopolitanism in the brachiopod genus level data in the lower most part of the O. snajdri Interval Zone which later spans up to the end of the Ludlow. A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 285 compared to Baltica data, looking at the re-appearance levels relative to match very closely the proposed model of 3rd order (≈1–2Maindura- the δ13C curves. The differences could be explained by environmental, tion) sea level changes, that is based on stratigraphic and stable oxygen climatic, trophic or migrational factors. isotopic data from the Gotland area (Lehnert et al., 2007). The maximal A similar pattern was detected in the strata corresponding to the af- transgression during the pulse of brachiopod cosmopolitanism is also termath of another extinction episode – the Ireviken Event on Gotland. confirmed by the brief appearance of the deep water (Armstrong, P. equicostatus appeared after the first extinction datum (Jeppsson, 1996) conodont Dapsilodus obliquicostatus (Fig. 2), appearing at the 1998). Higher up, in the lower K. ranuliformis Zone, P. equicostatus was samelevelalsoinBohemia(Slavik and Carls, 2012). Therefore it is high- a dominant species (Jeppsson, 1997). The coeval pattern is similar also ly probable that global fluctuations in the sea level and climate states in the Lithuanian sections (Spiridonov et al., 2015). The dominance of were responsible for the coordinated responses of conodont and bra- P.exgr.equicostatus was also documented in the Late Ordovician chiopod assemblages, as well as changes in cosmopolitanism observed (mid-Caradoc) after-event strata in Estonia (Männik, 2004). This sug- in the Baltic data (see Fig. 12 for the stratigraphical, sea level, and gests that P. equicostatus and related forms are characteristic of dis- palaeobiological synthesis of the inferred patterns). turbed and highly deviating Ordovician-Silurian environments, The very high centrality (CENC and CENM) of the joint dynamics of particularly in the Ludlow. conodonts and brachiopods shows that the intercalation of ‘typical’ ver- Besides the extinction of many Gorstian to lower Ludfordian species sus time specific (epiboles of exotic taxa (Brett, 1995, 1998; Brett and during the Lau Event and later transient dominance of P. equicostatus, Baird, 1997)) assemblages in both groups is highly non-random and co- there is another important feature that can be demonstrated in the cur- ordinated. The ‘typical’ conodont assemblages are associated with the rent and formerly published material. This is the temporary decline of ‘typical’ brachiopod assemblages, and anomalous assemblages in one taxa that are usually frequent in this interval – P. unicostatus, W. group are mostly synchronous with the anomalous assemblages in an- excavata and O. confluens (see, for example (Jeppsson, 1974; Barrick et other group (especially in the aftermath of the Lau Event). Even the ab- al., 2010; Slavik et al., 2010; Slavik and Carls, 2012). This numerical de- solute magnitudes of time specificity (centrality) for conodonts and crease accompanied with the appearance of the “exotic” P. equicostatus brachiopods track each other closely in the interpolated dataset (Fig. makes the conodont composition of the middle part of the O. snajdri 10,C). Interestingly though, on qualitative grounds it was proposed Zone so exceptional for Ludlow. The strong decrease in formerly domi- that the post-event shelly benthic communities in Bohemia were also nant species that usually form a “backbone” of communities defining represented by the” time specificfacies” such as monospecific trilobite many of their functions (Gaston, 2010; Connolly et al., 2014; assemblages (Manda and Frýda, 2014), which supports a wider spatial Hannisdal et al., 2016) is indicative of a significant restructuring of nek- extent of the revealed phenomenon. tonic and nektobenthic communities in the close aftermath of the Lau It was proposed that mid-Silurian to the mid-Devonian regional ma- extinction. The post-extinction intermittently stable conodont commu- rine communities from the Appalachian Basin experienced sharply nities were replaced by communities of more usual compositions after a bounded stable (equilibrial) states in their taxonomic composition, as relatively short time, on the order of several hundreds of Ka (assuming well as in their patterns of ecological dominance — the pattern called co- Silurian time scale of 2012 (Melchin et al., 2012)). ordinated stasis (Brett and Baird, 1995; Brett et al., 1996; Brett, 2012), in The brachiopods of the Milaičiai-103 section demonstrate a similar some respect analogous to the morphological stasis of single lineages in anomalous composition in the O. snajdri Interval Zone. On the other the macroevolutionary theory of punctuated equilibria (Eldredge and hand, there is a significant difference between these two groups. The Gould, 1972; Gould, 2002). The present study wasn't designed to an- local conodont communities were strongly impoverished by the Lau swer the question of the taxonomic stasis, which is actively debated event, whilst brachiopods, at least at the local scale in the aftermath (Morris et al., 1995; Miller, 1997; Bonuso et al., 2002; Ivany et al., (similarly as in the aftermaths of Ireviken and Mulde event (Bassett 2009), of the coordinated stasis hypothesis. However it supports the and Cocks, 1974; Munnecke et al., 2003) experienced a rise in species view that there is high degree of temporal coordination between the diversity, despite the fact that the event had a global negative impact communities of two taxonomically different and ecologically important on genus level diversity (Talent et al., 1993). The palaeobiogeographical groups occupying disparate ecological guilds. analysis of brachiopod cosmopolitanism, together with the analysis of Both conodonts and brachiopods are represented by the assem- other proxies, gives a clue for the reasons of the observed pattern. The blages of anomalous composition in the O. snajdri Interval Zone. On lowest cosmopolitanism values observed in the lower to middle parts the other hand the causes of this patterns were likely different. In the of the section (upper K. variabilis Interval Zone and P. siluricus Zone) case of conodonts, the respective assemblages were characterized by point at restricted dispersal of benthic fauna in the early Ludlow, i.e. very strong dominance of a single taxon whilst in brachiopods the re- strong regression associated with the Lau Event (Calner and Eriksson, spective dominants comprised a group of short-ranged taxa. Although, 2006; Lehnert et al., 2007). The recent developments in the close association of the change points in the centrality graphs with palaeobiogeographic macroevolutionary theory suggest that a major in- the major biogeochemical and turnover events points to an external crease in diversity should be expected during transgressions following common forcing mechanism (see Peters and Foote, 2002) of the coordi- major regressive events (Stigall et al., 2016). On the other hand, the nated response (though group-specific in detail), deviating composi- sea level drop could facilitate both the extinction and origination of tions, and outages of faunas. new endemic taxa through geographic range “implosion” vicariance The impact of the Lau Event and the spread and significance of the (Radzevičius et al., 2016) and the sea level rise theoretically should in- assemblages of anomalous composition can be evaluated only in the crease alpha (local level) diversity through geodispersal and further light of time/environment analysis (Holland, 1997). The physical nature promote the diversification through the local adaptation (Abe and of the forcing in the aftermath of the Lau Event is supported by the sed- Lieberman, 2009; Stigall, 2015; Stigall et al., 2016). The major increase imentological and palaeoecological studies of carbonate shelf deposits. in the cosmopolitanism of brachiopod assemblages in the lower O. The upper Ludfordian nearshore strata on Gotland witnessed significant snajdri Interval Zone, observed in the studied data set, points to the im- expansion of stromatolites, oncoids, oolites and flat-pebble conglomer- pulse of geodispersal and possibly biotic homogenization. A new fauna ates, all of which signal unsuppressed development of microbial com- of almost cosmopolitan brachiopod genera was established during this munities and pointing at the development of so called “anachronistic” episode. A very similar pattern was observed in the aftermath of the facies (Calner, 2005a, 2005b; Eriksson and Calner, 2008). Very similar, late Ordovician extinction event, during the earliest Silurian recovery contemporaneous microbially mediated facies were detected in a near- of brachiopod communities — the extinction was followed by the trans- shore sections of the Mituva Formation in Lithuania (Lapinskas, 2000; gression-mediated invasion of species (Sheehan, 1975; Sheehan and Lapinskas, 2004), but also in Podolia (Ukraine) (Jarochowska and Coorough, 1990). The pulse of brachiopod (and conodont) dispersal Kozłowski, 2014) and in northeastern Australia (Jeppsson et al., 2007), 286 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288

Fig. 12. The summary and the interpretation of the biogeochemical (δ13C), palaeoecological and palaeobiogeographic perturbations as described in the present study compared to the sea level trend (red curve) proposed by Lehnert et al. (2007),andEriksson and Calner (2008).

likely reflecting a global extent of anomalous conditions. It was even specificity of the fossil assemblages using recurrence plots revealed proposed, based on the stratigraphical distribution of oncoids and stro- that conodonts and brachiopods jointly experienced a transient matolites, that they are a generic feature of the Silurian events anomalous episode in the aftermath of the Lau extinction event, (Antoshkina, 2015). The widespread stromatolite biostromes in the which ended at the beginning of the O. baccata–O. crispa Zone. Phanerozoic are generally confined to the “event horizons” (Eagan • The anomalous interval corresponds to the previously proposed and Liddell, 1997) and thought to indicate, in the majority of cases, re- “anachronistic period” characterized by the resurgence of microbially stricted (for complex multicellular life), or at least peculiar environ- mediated sedimentary structures (stromatolites, oncoids etc.) in the ments (Garrett, 1970; Schubert and Bottjer, 1992; Yao et al., 2016). shallow water successions. This fact reveals the deep although tran- Recent sedimentological and geochemical evidence shows a high car- sient structural impact of the Lau event for the whole ecosystem. bonate saturation of the ocean water during the Ludfordian Lau Event (Kozłowski, 2015). It is likely that carbonate supersaturation at least partially promoted the nearshore development of stromatolites that Acknowledgments are thought to be limited by the availability of carbonate through the most of the Phanerozoic (Grotzinger, 1990). The same factor could en- We would like to thank Carlton E. Brett (University of Cincinnati) able the flourishing of offshore shelly benthos described here, and pos- and two anonymous reviewers for their constructive suggestions and sibly its better preservation due to chemical buffering effects against linguistic corrections. L.A. and T. M. were financed by the project dissolution. IUT20-34 “The Phanerozoic journey of Baltica: sedimentary, geochemi- The fact that there is high congruence between different data sets cal and biotic signatures of changing environment – PalaeoBaltica” from and regions suggests a significant, although transient, ecosystem scale the Estonian Research Agency. This is a contribution to IGCP-652: state shift in the aftermath of the Lau extinction. Moreover the present “Reading geologic time in Palaeozoic sedimentary rocks: the need for application of recurrence plots and centrality analysis allowed the rec- an integrated stratigraphy”. This paper is Paleobiology Database official ognition of the anomalous compositions of conodont and brachiopod publication number 291. communities. The compositional data are intrinsically multivariate and therefore difficult to interpret using intuition and expert judgement Appendix A. Supplementary data only. Some time specific patterns are obviously outlying against the background in the normal (metazoan dominated) Phanerozoic succes- Supplementary data to this article can be found online at http://dx. sions (for example stromatolite bioherms), and others can be formed doi.org/10.1016/j.gr.2017.08.006. as an intricate and persistent combination of ordinary features (for example common species in highly deviating proportions in the context References of the neighbouring samples). In the latter case, the recognition of an anomaly can be very complicated. The objective approach outlined Abe, F.R., Lieberman, B.S., 2009. The nature of evolutionary radiations: a case study involv- here, combined with null modelling, can be used for other groups of or- ing Devonian trilobites. Evolutionary Biology 36, 225–234. Antoshkina, A., 2015. Ooid-stromatolite association as a precursor of bioevents (Silurian, ganisms, as well as for any other (e.g. coded lithofacies) variables in Timan–northern Ural region). Palaeoworld 24, 198–206. testing the hypothesis of time specificity of a set of studied features. Armstrong, H., 1996. Biotic recovery after mass extinction: the role of climate and ocean- state in the post-glacial (Late Ordovician-Early Silurian) recovery of the conodonts. Geological Society, London, Special Publications 102, 105–117. 7. Conclusions Barrick, J.E., Kleffner, M.A., Gibson, M.A., Peavey, F.N., Karlsson, H.R., 2010. The mid- Ludfordian Lau Event and carbon isotope excursion (Ludlow, Silurian) in southern Based on the comprehensive bio-chemostratigraphical and Laurentia–preliminary results. Bollettino della Società Paleontologica Italiana 49, – palaeoecological analysis of the environmental and biotic change in 13 33. Barrick, J.E., Klapper, G., Kleffner, M.A., Karlsson, H.R., 2011. Conodont biostratigraphy and the dynamic Ludlow succession in the Milaičiai-103 core section, the stable isotope chemostratigraphy of the lower Henryhouse Formation (Gorstian- following conclusions are drawn: early Ludfordian, Ludlow, Silurian), southern Oklahoma, USA. Australasian Palaeontological Memoirs 51. • Bassett, M.G., Cocks, L.R.M., 1974. A review of Silurian brachiopods from Gotland. Fossils Constrained cluster analysis revealed three major conodont assem- and Strata 3, 1–56. blages and six distinct brachiopod assemblages in the Ludlow. Bassett, M.G., Kaljo, D., Teller, L., 1989. The Baltic region. In: Holland, C., Bassett, M.G. • The changes of assemblages are coordinated with the changes in δ13C (Eds.), A Global Standard for the Silurian System. National Museum of Wales Geolog- ical Series Cardiff, pp. 158–170. and eustatic sea level curves. Bennett, K.D., 1996. Determination of the number of zones in a biostratigraphical • A newly developed mathematical technique for the detection of time- sequence. New Phytologist 132, 155–170. A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288 287

Bickert, T., Pätzold, J., Samtleben, C., Munnecke, A., 1997. Paleoenvironmental changes in Girard, C., Renaud, S., 2007. Quantitative conodont-based approaches for correlation of the Silurian indicated by stable isotopes in brachiopod shells from Gotland, Sweden. the Late Devonian Kellwasser anoxic events. Palaeogeography, Palaeoclimatology, Pa- Geochimica et Cosmochimica Acta 61, 2717–2730. laeoecology 250, 114–125. Bonuso, N., Newton, C.R., Brower, J.C., Ivany, L.C., 2002. Does coordinated stasis yield tax- Girard, C., Cornee, J.-J., Corradini, C., Fravalo, A., Feist, R., 2014. Palaeoenvironmental onomic and ecologic stability?: Middle Devonian Hamilton Group of central New changes at Col des Tribes (Montagne Noire, France), a reference section for the York. Geology 30, 1055–1058. Famennian of north Gondwana-related areas. Geological Magazine 151, 864–884. Boucot, A.J., 1975. Evolution and Extinction Rate Controls. Elsevier, Amsterdam. Gould, S.J., 2002. The Structure of Evolutionary Theory. Harvard University Press, Cam- Brazauskas, A., 1989. About the Ludlovian and Pridolian boundary according to conodonts bridge, MA. in Lithuanian sections [in Russian]. Geologija 10, 72–78. Green, O.R., 2001. A Manual of Practical Laboratory and Field Techniques in Palaeobiology. Brazauskas, A., 1993. Silurian Conodonts and Biostratigraphy of Lithuania [in Russian]. Springer, Dordrecht. Dissertation. Vilnius University, p. 336. Grimm, E.C., 1987. CONISS: a FORTRAN 77 program for stratigraphically constrained clus- Brazauskas, A., Musteikis, P., Paškevičienė,L.,Paškevičius, J., Radzevičius, S., Jankauskas, T., ter analysis by the method of incremental sum of squares. Computers & Geosciences Karatajūtė-Talimaa, V., Gritytė, J., Valiukevičius, J., 2004. Taxonomic composition and 13, 13–35. recognition of the analyzed remains. Silurian. In: Baltrūnas, V. (Ed.), Litosfera: Grotzinger, J.P., 1990. Geochemical model for Proterozoic stromatolite decline. American Lietuvos žemėsgelmių raida ir ištekliai. Institute of Geology and Geography, Vilnius Journal of Science 290, 80–103. University, Vilnius, pp. 173–194. Hannisdal, B., Haaga, K.A., Reitan, T., Diego, D., Liow, L.H., 2016. Common species link glob- Brett, C.E., 1995. Sequence stratigraphy, biostratigraphy, and taphonomy in shallow ma- al ecosystems to climate change. bioRxiv, 043729. rine environments. PALAIOS 597–616. Holland, S.M., 1997. Using Time/Environment Analysis to Recognize Faunal Events in the Brett, C.E., 1998. Sequence stratigraphy, paleoecology, and evolution; biotic clues and re- Upper Ordovician of the Cincinnati Arch. Paleontological Event Horizons: Ecological sponses to sea-level fluctuations. PALAIOS 13, 241–262. and Evolutionary Implications. Columbia University Press, New York, pp. 309–334. Brett, C.E., 2012. Coordinated stasis reconsidered: a perspective at fifteen years. Earth and Horn, H.S., 1966. Measurement of “overlap” in comparative ecological studies. The Amer- Life. Springer, pp. 23–36. ican Naturalist 100, 419–424. Brett, C.E., Baird, G.C., 1995. Coordinated stasis and evolutionary ecology of Silurian to Huang, C., Gong, Y., 2016. Timing and patterns of the Frasnian–Famennian event: evi- Middle Devonian faunas in the Appalachian Basin. In: Erwin, D.H., Anstey, R.L. dences from high-resolution conodont biostratigraphy and event stratigraphy at (Eds.), New Approaches to Speciation in the Fossil Record. Columbia University the Yangdi section, Guangxi, South China. Palaeogeography, Palaeoclimatology, Pa- Press, New York, pp. 285–315. laeoecology 448, 317–338. Brett, C.E., Baird, G.C., 1997. Epiboles, Outages, and Ecological Evolutionary Bioevents: Ivany, L.C., Brett, C.E., Wall, H.L.B., Wall, P.D., Handley, J.C., 2009. Relative taxonomic and Taphonomic, Ecological, and Biogeographic Factors. Paleontological Events: Strati- ecologic stability in Devonian marine faunas of New York State: a test of coordinated graphic, Ecological, and Evolutionary Implications. Columbia University Press, New stasis. Paleobiology 35, 499–524. York, pp. 249–284. Jarochowska, E., Kozłowski, W., 2014. Facies development and sequence stratigraphy of Brett, C.E., Ivany, L.C., Schopf, K.M., 1996. Coordinated stasis: an overview. the Ludfordian (Upper Silurian) deposits in the Zbruch River Valley, Podolia, Western Palaeogeography, Palaeoclimatology, Palaeoecology 127, 1–20. Ukraine: local facies overprint on the δ13Ccarb record of a global stable carbon isotope Calner, M., 2005a. A Late Silurian extinction event and anachronistic period. Geology 33, excursion. Facies 60, 347–369. 305–308. Jarochowska, E., Munnecke, A., 2015. Late Wenlock carbon isotope excursions and associ- Calner, M., 2005b. Silurian carbonate platforms and extinction events—ecosystem chang- ated conodont fauna in the Podlasie Depression, eastern Poland: a not-so-big crisis? es exemplified from Gotland, Sweden. Facies 51, 584–591. Geological Journal 51 (5), 683–703. Calner, M., 2008. Silurian global events–at the tipping point of climate change. In: Elewa, Jeppsson, L., 1974. Aspects of late Silurian conodonts. Fossils and Strata 79. A. (Ed.), Mass Extinction. Springer, Berlin, pp. 21–57. Jeppsson, L., 1987. Lithological and conodont distributional evidence for episodes of Calner, M., Eriksson, M.J., 2006. Evidence for rapid environmental changes in low latitudes anomalous oceanic conditions during the Silurian. In: RJ, A. (Ed.), Palaeobiology of during the Late Silurian Lau Event: the Burgen-1 drillcore, Gotland, Sweden. Geolog- Conodonts. Ellis Horwood Ltd, Chichester, pp. 129–145. ical Magazine 143, 15–24. Jeppsson, L., 1997. A new latest Telychian, Sheinwoodian and Early Homerian (Early Silu- Connolly, S.R., MacNeil, M.A., Caley, M.J., Knowlton, N., Cripps, E., Hisano, M., Thibaut, L.M., rian) standard conodont zonation. Earth and Environmental Science Transactions of Bhattacharya, B.D., Benedetti-Cecchi, L., Brainard, R.E., 2014. Commonness and rarity the Royal Society of Edinburgh 88 (2), 91–114. in the marine biosphere. Proceedings of the National Academy of Sciences 111, Jeppsson, L., 1998. Silurian oceanic events: summary of general characteristics. In: 8524–8529. Landing, E., Johnson, M. (Eds.), Silurian Cycles: Linkages of Dynamic Stratigraphy Cooper, R.A., Sadler, P.M., Munnecke, A., Crampton, J.S., 2014. Graptoloid evolutionary With Atmospheric, Oceanic and Tectonic Changes. New York State Museum, New rates track Ordovician–Silurian global climate change. Geological Magazine 151, York, pp. 239–257. 349–364. Jeppsson, L., 2005. Conodont-based revisions of the Late Ludfordian on Gotland, Sweden. Corradini, C., Corriga, M.G., Männik, P., Schönlaub, H.P., 2015. Revised conodont stratigra- GFF 127, 273–282. phy of the Cellon section (Silurian, Carnic Alps). Lethaia 48, 56–71. Jeppsson, L., Aldridge, R., 2000. Ludlow (late Silurian) oceanic episodes and events. Jour- Corriga, M.G., Corradini, C., Ferretti, A., 2009. Silurian conodonts from Sardinia: an over- nal of the Geological Society 157, 1137–1148. view. Rendiconti della Società Paleontologica Italiana 3, 95–107. Jeppsson, L., Calner, M., 2002. The Silurian Mulde Event and a scenario for secundo– Cramer, B.D., Brett, C.E., Melchin, M.J., Maennik, P., Kleffner, M.A., McLaughlin, P.I., Loydell, secundo events. Earth and Environmental Science Transactions of the Royal Society D.K., Munnecke, A., Jeppsson, L., Corradini, C., Brunton, F.R., 2011. Revised correlation of Edinburgh 93, 135–154. of Silurian Provincial Series of North America with global and regional chrono- Jeppsson, L., Eriksson, M.E., Calner, M., 2006. A latest Llandovery to latest Ludlow high- stratigraphic units and δ13Ccarb chemostratigraphy. Lethaia 44 (2), 185–202. resolution biostratigraphy based on the Silurian of Gotland—a summary. GFF 128, Cramer, B.D., Schmitz, M.D., Huff, W.D., Bergström, S.M., 2015. High-precision U–Pb zircon 109–114. age constraints on the duration of rapid biogeochemical events during the Ludlow Jeppsson, L., Talent, J.A., Mawson, R., Simpson, A.J., Andrew, A.S., Calner, M., Whitford, D.J., Epoch (Silurian Period). Journal of the Geological Society 172 (2), 157–160. Trotter, J.A., Sandström, O., Caldon, H.-J., 2007. High-resolution Late Silurian correla- Crampton, J.S., Cooper, R.A., Sadler, P.M., Foote, M., 2016. Greenhouse–icehouse transition tions between Gotland, Sweden, and the Broken River region, NE Australia: litholo- in the Late Ordovician marks a step change in extinction regime in the marine plank- gies, conodonts and isotopes. Palaeogeography, Palaeoclimatology, Palaeoecology ton. Proceedings of the National Academy of Sciences 113, 1498–1503. 245, 115–137. Dzik, J., 2016. Evolutionary roots of the conodonts with increased number of elements in Jeppsson, L., Talent, J.A., Mawson, R., Andrew, A., Corradini, C., Simpson, A.J., Wigforss- the apparatus. Earth and Environmental Science Transactions of the Royal Society of Lange, J., Schönlaub, H.P., 2012. Late Ludfordian correlations and the Lau event. Edinburgh 29–53. Earth and Life. Springer, pp. 653–675. Eagan, K.E., Liddell, W.D., 1997. Stromatolite biostromes as bioevent horizons: an example Jia-Yu, R., Boucot, A.J., Yang-Zheng, S., Strusz, D.L., 1995. Biogeographical analysis of Late from the Middle Cambrian Ute Formation of the Eastern Great Basin. Paleontological Silurian brachiopod faunas, chiefly from Asia and Australia. Lethaia 28, 39–60. Events: Stratigraphic, Ecological, and Evolutionary Implications 285–308. Juggins, S., 2015. rioja: analysis of Quaternary science data. R package version 0.9-5. URL. Einasto, P.E., Abushik, A.F., Kaljo, D.L., Koren', T.N., Modzalevskaya, T.L., Nestor, H.E., https://cran.r-project.org/web/packages/rioja/index.html. Klaamann, E., 1986. Osobiennosti silurskogo osadkonakopleniya i associacii fauny v Kaljo, D., Martma, T.O.N., 2006. Application of carbon isotope stratigraphy to dating the kraevych basseinach Pribaltiki i Podolii. In: Kaljo, D.L., Klaamann, E. (Eds.), Teoria i Baltic Silurian rocks. GFF 128 (2), 123–129. opyt ekostratigrafii. Valgus, Tallin, pp. 65–72. Kaljo, D., Boucot, A.J., Corfield, R.M., Le Herisse, A., Koren', T.N., Kriz, J., Männik, P., Märss, Eldredge, N., Gould, S.J., 1972. Punctuated equilibria: an alternative to phyletic gradual- T., Nestor, V., Shaver, R.H., Siveter, D.J., Viira, V., 1996. Silurian bio-events. In: Walliser, ism. In: Schopf, T. (Ed.), Models in Paleobiology. Freeman and Company, San O.H. (Ed.), Global Events and Event Stratigraphy in the Phanerozoic. Springer, Berlin, Francisco, pp. 82–115. pp. 173–224. Eriksson, M.J., Calner, M., 2008. A sequence stratigraphical model for the Late Ludfordian Kaljo, D., Kiipli, T., Martma, T., 1997. Carbon isotope event markers through the Wenlock– (Silurian) of Gotland, Sweden: implications for timing between changes in sea level, Pridoli sequence at Ohesaare (Estonia) and Priekule (Latvia). Palaeogeography, palaeoecology, and the global carbon cycle. Facies 54, 253–276. Palaeoclimatology, Palaeoecology 132, 211–223. Eriksson, M.E., Nilsson, E.K., Jeppsson, L., 2009. Vertebrate extinctions and reorganizations Kaljo, D., Grytsenko, V., Martma, T., Mõtus, M.A., 2007. Three global carbon isotope shifts during the Late Silurian Lau Event. Geology 37, 739–742. in the Silurian of Podolia (Ukraine): stratigraphical implications. Estonian Journal of Frýda, J., Manda, Š., 2013. A long-lasting steady period of isotopically heavy carbon in the Earth Sciences 56 (4), 205–220. late Silurian ocean: evolution of the δ13C record and its significance for an integrated Kaljo, D., Grytsenko, V., Kallaste, T., Kiipli, T., Martma, T., 2014. Upper Silurian stratigraphy δ13C, graptolite and conodont stratigraphy. Bulletin of Geosciences 88. of Podolia revisited: carbon isotopes, bentonites and biostratigraphy. GFF 136, Garrett, P., 1970. Phanerozoic stromatolites: noncompetitive ecologic restriction by graz- 136–141. ing and burrowing animals. Science 169, 171–173. Koren', T.N., Sujarkova, A.A., 2004. The Ludlow (late Silurian) neocucullograptid fauna Gaston, K.J., 2010. Valuing common species. Science 327, 154–155. from the southern Tien Shan, Kyrghizstan. Alcheringa 28, 333–387. 288 A. Spiridonov et al. / Gondwana Research 51 (2017) 272–288

Koren', T.N., Suyarkova, A.A., 2007. Silurian graptolite biostratigraphy of the Kaliningrad Olszewski, T.D., 2012. Remembrance of things past: modelling the relationship between district, northwest Russia. Acta Palaeontologica Sinica 46 (Suppl), 232–236. species' abundances in living communities and death assemblages. Biology Letters Kozłowski, W., 2015. Eolian dust influx and massive whitings during the kozlowski/Lau 8, 131–134. Event: carbonate hypersaturation as a possible driver of the mid-Ludfordian Carbon Paškevičius, J., 1997. The Geology of the Baltic Republics. Vilnius University and Geological Isotope Excursion. Bulletin of Geosciences 90, 807–840. Survey of Lithuania (387 p). Kozłowski, W., Sobień, K., 2012. Mid-Ludfordian coeval carbon isotope, natural gamma Paškevičius, J., Lapinskas, P., Brazauskas, A., Musteikis, P., Jacyna, J., 1994. Stratigraphic re- ray and magnetic susceptibility excursions in the Mielnik IG-1 borehole (Eastern vision of the regional stages of the Upper Silurian part in the Baltic Basin. Geologija Poland)—dustiness as a possible link between global climate and the Silurian carbon (Vilnius) 17, 64–87. isotope record. Palaeogeography, Palaeoclimatology, Palaeoecology 339, 74–97. Paškevičius, J., Klimantavičius, V., Radzevičius, S., 2012. Lithostratigraphy, graptolites and Lapinskas, P., 2000. Structure and Petroliferosity of the Silurian in Lithuania. Institute of brachiopods communities of the Ludlow (Silurian) of the Eastern slope of the Baltic Geology, Vilnius. Syneclise. Geologija 54, 75–88. Lapinskas, P., 2004. Features of paleogeographical change. Silurian [in Lithuanian]. In: Peters, S.E., Foote, M., 2002. Determinants of extinction in the fossil record. Nature 416, Baltrūnas, V. (Ed.), Litosfera: Lietuvos žemėsgelmių raida ir ištekliai. Institute of Ge- 420–424. ology and Geography, Vilnius University, Vilnius, pp. 229–232. Purnell, M.A., Donoghue, P.C., 2005. Between death and data: biases in interpretation of Lapinskas, P., Paškevičius, J., Jacyna, J., 1985. New data on the stratigraphy and correlation the fossil record of conodonts. Special Papers in Palaeontology 73, 7–25. of the South-east Baltic Wenlockian and Ludlovian (Silurian) [in Russian]. Geologija R Development Core Team, 2015. R: A Language and Environment for Statistical Comput- 29–40. ing. Version 3.1.3. R Foundation for Statistical Computing, Vienna. Lazauskienė,J.,Šliaupa, S., Brazauskas, A., Musteikis, P., 2003. Sequence stratigraphy of the Radzevičius, S., Spiridonov, A., Rimkus, A., Meidla, T., Ainsaar, L., Brazauskas, A., Dankina, Baltic Silurian succession: tectonic control on the foreland infill. Geological Society, D., 2014. Integrated δ13C isotope and graptolite stratigraphy of Upper Wenlock and London, Special Publications 208, 95–115. Lower Ludlow of the Vilkaviškis-134 core, Lithuania. Geologija 56, 32. Lehnert, O., Fryda, J., Buggisch, W., Manda, S., 2003. A first report of the Ludlow Lau event Radzevičius, S., Spiridonov, A., Brazauskas, A., Dankina, D., Rimkus, A., Bičkauskas, G., from the Prague Basin (Barrandian, Czech Republic). Proceedings of the 7th Interna- Kaminskas, D., Meidla, T., Ainsaar, L., 2016. Integrated stratigraphy, conodont turn- tional Graptolite Conference and Field Meeting of the International Subcommission over and paleoenvironments of the Upper Wenlock and Ludlow of the Vilkaviškis- on Silurian Stratigraphy: Tucumán, pp. 139–144. 134 core (Lithuania). Newsletters on Stratigraphy 49 (2), 321–336. Lehnert, O., Eriksson, M.J., Calner, M., Joachimski, M., Buggisch, W., 2007. Concurrent sed- Samtleben, C., Munnecke, A., Bickert, T., Pätzold, J., 1996. The Silurian of Gotland (Swe- imentary and isotopic indications for global climatic cooling in the Late Silurian. Acta den): facies interpretation based on stable isotopes in brachiopod shells. Geologische Palaeontologica Sinica 46, 249. Rundschau 85, 278–292. Lenton, T.M., Dahl, T.W., Daines, S.J., Mills, B.J.W., Ozaki, K., Saltzman, M.R., Porada, P., Samtleben, C., Munnecke, A., Bickert, T., 2000. Development of facies and C/O-isotopes in 2016. Earliest land plants created modern levels of atmospheric oxygen. Proceedings transects through the Ludlow of Gotland: evidence for global and local influences on of the National Academy of Sciences 113 (35), 9704–9709. a shallow-marine environment. Facies 43 (1), 1–38. Lenz, A.C., Kozłowska-Dawidziuk, A., 2004. Ludlow and Pridoli (Upper Silurian) Grapto- Schubert, J.K., Bottjer, D.J., 1992. Early Triassic stromatolites as post-mass extinction disas- lites from the Arctic Islands, Canada. National Research Council of Canada, Ottawa ter forms. Geology 20, 883–886. (141 pp). Serpagli, E., Corradini, C., Oliviri, R., 1996. Occurrence and range of the Silurian conodont Loydell, D.K., Jeppsson, L., Aldridge, R.J., 2001. Discussion on Ludlow (late Silurian) oceanic Coryssognathus dubius (Rhodes, 1953) in Southern Sardinia. Bollettino-Societa episodes and events journal, Vol. 157, 2000, 1137–1148. Journal of the Geological So- Paleontologica Italiana 35, 239–244. ciety 158, 731–732. Sheehan, P.M., 1975. Brachiopod synecology in a time of crisis (Late Ordovician-Early Si- Loydell, D.K., Nestor, V., Männik, P., 2010. Integrated biostratigraphy of the lower Silurian lurian). Paleobiology 205–212. of the Kolka-54 core, Latvia. Geological Magazine 147, 253–280. Sheehan, P.M., Coorough, P.J., 1990. Brachiopod zoogeography across the Ordovician-Silu- Manda, Š., Frýda, J., 2014. Evolution of the late Ludlow to early Lochkovian brachiopod, tri- rian extinction event. Geological Society, London, Memoirs 12, 181–187. lobite and bivalve communities of the Prague Basin and their link with the global car- Slavik, L., Carls, P., 2012. Post-Lau Event (late Ludfordian, Silurian) recovery of conodont bon cycle. GFF 136, 179–184. faunas of Bohemia. Bulletin of Geosciences 87, 815–832. Männik, P., 2004. Recognition of the Mid-Caradoc event in the conodont succession of Es- Slavik, L., Kříž, J., Carls, P., 2010. Reflection of the mid-Ludfordian Lau Event in conodont tonia. WOGOGOB-2004 8th Meeting on the Working Group on the Ordovician Geol- faunas of Bohemia. Bulletin of Geosciences 85, 395–414. ogy of Baltoscandia May 13–18, 2004. Estonia Organising Comittee, Tallinn and Tartu, Spiridonov, A., 2017. Recurrence and cross recurrence plots reveal the onset of the Mulde p. 63. event (Silurian) in the abundance data for Baltic conodonts. The Journal of Geology Männik, P., Malkowski, K., 1998. Silurian conodonts from the Goldap core, Poland. Pro- 125, 381–398. ceedings of the Sixth European Conodont Symposium (ECOSVI). Palaeontologia Spiridonov, A., Brazauskas, A., Radzevičius, S., 2015. The role of temporal abundance struc- Polonica, pp. 141–151. ture and habitat preferences in the survival of conodonts during the mid-early Siluri- Märss, T., Männik, P., 2013. Revision of Silurian vertebrate biozones and their correlation an Ireviken mass extinction event. PloS One 10, e0124146. with the conodont succession. Estonian Journal of Earth Sciences 62, 181–204. Spiridonov, A., Brazauskas, A., Radzevičius, S., 2016. Dynamics of abundance of the mid- to Märss, T., Miller, C.G., 2004. Thelodonts and distribution of associated conodonts from the late Pridoli conodonts from the eastern part of the Silurian Baltic Basin: multifractals, Llandovery–lowermost Lochkovian of the Welsh Borderland. Palaeontology 47, state shifts, and oscillations. American Journal of Science 316, 363–400. 1211–1265. Stigall, A.L., 2015. Speciation: expanding the role of biogeography and niche breadth in Martma, T., Brazauskas, A., Kaljo, D., Kaminskas, D., Musteikis, P., 2005. The Wenlock- macroevolutionary theory. Macroevolution. Springer, pp. 301–327. Ludlow carbon isotope trend in the Vidukle core, Lithuania, and its relations with oce- Stigall, A.L., Bauer, J.E., Lam, A.R., Wright, D.F., 2016. Biotic immigration events, speciation, anic events. Geological Quarterly 49, 223–234. and the accumulation of biodiversity in the fossil record. Global and Planetary Change Marwan, N., Romano, M.C., Thiel, M., Kurths, J., 2007. Recurrence plots for the analysis of 148, 242–257. complex systems. Physics Reports 438, 237–329. Štorch, P., 1995. Biotic crises and post-crisis recoveries recorded by Silurian planktonic Mathieson, D., Mawson, R., Simpson, A.J., Talent, J.A., 2016. Late Silurian (Ludlow) and graptolite faunas of the Barrandian area (Czech Republic). Geolines 3, 59–70. Early Devonian (Pragian) conodonts from the Cobar Supergroup, western New Štorch, P., Manda, Š., Loydell, D.K., 2014. The early Ludfordian leintwardinensis graptolite South Wales, Australia. Bulletin of Geosciences 9. Event and the Gorstian–Ludfordian boundary in Bohemia (Silurian, Czech Republic). McGoff, H.J., 1991. The hydrodynamics of conodont elements. Lethaia 24, 235–247. Palaeontology 1–41 (Early view). Melchin, M.J., Sadler, P.M., Cramer, B.D., 2012. The Silurian Period. In: Gradstein, F.M., Ogg, Talent, J.A., Mawson, R., Andrew, A.S., Hamilton, P.J., Whitford, D.J., 1993. Middle J.G., Schmitz, M. (Eds.), The Geologic Time Scale. Elsevier, Amsterdam, pp. 525–558 Palaeozoic extinction events: faunal and isotopic data. Palaeogeography, 2012, 2-volume set. Palaeoclimatology, Palaeoecology 104, 139–152. Miller, A.I., 1997. Coordinated stasis or coincident relative stability? Paleobiology 23, Urbanek, A., 1970. Neocucullograptinae n. subfam. (Graptolithina)—their evolutionary 155–164. and stratigraphic bearing. Acta Palaeontologica Polonica 15. Miller, C.G., Aldridge, R.J., 1997. Ozarkodina remscheidensis plexus conodonts from the Urbanek, A., 1993. Biotic crises in the history of Upper Silurian graptoloids: a upper Ludlow (Silurian) of the Welsh Borderland and Wales. Journal of palaeobiological model. Historical Biology 7, 29–50. Micropalaeontology 16, 41–49. Urbanek, A., Teller, L., 1997. Graptolites and stratigraphy of the Wenlock and Ludlow se- Morris, P.J., Ivany, L.C., Schopf, K.M., Brett, C.E., 1995. The challenge of paleoecological sta- ries in the East European Platform. Palaeontologia Polonica 56, 23–57. sis: reassessing sources of evolutionary stability. Proceedings of the National Acade- Vierek, A., Racki, G., 2011. Depositional versus ecological control on the conodont distri- my of Sciences 92, 11269–11273. bution in the Lower Frasnian fore-reef facies, Holy Cross Mountains, Poland. Munnecke, A., Samtleben, C., Bickert, T., 2003. The Ireviken Event in the lower Silurian of Palaeogeography, Palaeoclimatology, Palaeoecology 312, 1–23. Gotland, Sweden–relation to similar Palaeozoic and Proterozoic events. Webber, C.L.J., Marwan, N., 2015. Recurrence Quantification Analysis: Theory and Best Palaeogeography, Palaeoclimatology, Palaeoecology 195 (1), 99–124. Practices. Munnecke, A., Delabroye, A., Servais, T., Vandenbroucke, T.R.A., Vecoli, M., 2012. System- Yao, L., Aretz, M., Chen, J., Webb, G.E., Wang, X., 2016. Global microbial carbonate prolif- atic occurrences of malformed (teratological) acritarchs in the run-up of Early eration after the end-Devonian mass extinction: mainly controlled by demise of skel- Palaeozoic δ13C isotope excursions. Palaeogeography, Palaeoclimatology, Palaeoecol- etal bioconstructors. Scientific Reports 6. ogy 367, 137–146. Zalasiewicz, J.A., Taylor, L., Rushton, A.W.A., Loydell, D.K., Rickards, R.B., Williams, M., Musteikis, P., Modzalevskaya, T.L., 2002. Some Silurian brachiopods from Lithuania and 2009. Graptolites in British stratigraphy. Geological Magazine 146, 785–850. their palaeobiogeographical significance. Palaeontology 45, 595–626. Musteikis, P., Paškevičius, J., 1999. Brachiopod communities of the Lithuanian Silurian. In: Boucot, A.J., Lawson, J.D. (Eds.), Paleocommunities - A Case Study From the Silurian and Lower Devonian. Cambridge University Press, Cambridge, pp. 305–326.