An introduction to ice ages, climate dynamics and biotic events: the Late Pennsylvanian world

The Middle–Late Pennsylvanian chronostratigraphic scale consists of two series, the Middle and Upper Pennsylvanian, and three so-called ‘global’ stages (in ascending order), Moscovian, Kasimovian and Gzhelian (Fig. 1). The Kasimovian Stage is the lower stage of the Upper Pennsylvanian Series. In reviewing the timescale associated with the Kasimovian Stage, Wang et al. (2023) note that the Variscan Orogeny and the late Paleozoic ice ages caused geographical isolation of marine faunas, and this makes difficult global correlation of Middle–Upper Pennsylvanian marine biostratigraphy. Another important factor was the closure of the Rheic Ocean, when Gondwana and Laurussia were finally amalgamated to make Pangaea during the Serpukhovian, which made impossible the movement of marine organisms around Pangaea via Tethys (Fig. 2).

Wang et al. (2023) review the major fossil groups, such as conodonts, fusulines and some macrofossils, that provide the basis for a tentative global correlation of the Kasimovian. The index taxon (primary signal) for the base of the Kasimovian Stage has not been selected, but the lowest occurrence of the conodont Idiognathodus heckeli might be the best marker for the base of the Kasimovian Stage because it has a wide geographical distribution, and it has a clear taxonomic definition within a hypothesized phylogenetic lineage. According to Wang et al. (2023), the fusuline Montiparus might be regarded as an auxiliary marker to define the base of the Kasimovian given its wide distribution (Ueno 2022). The GSSP candidates for the base of the Kasimovian Stage could be the Naqing section in South China, the Usolka section in the South Urals, and the Afanasievo section in the Moscow Basin.

Wang et al. (2023), nevertheless, do not address a pressing problem of Pennsylvanian chronostratigraphy, the currently different ways in which the base of the Upper Pennsylvanian (Kasimovian) is defined. Thus, three different definitions of that boundary are currently in use (Fig. 3): (1) the base of the Kasimovian Stage as it is defined in the Moscow basin, which is the base of the Krevyakinian substage (Alekseev et al. 2009); (2) the base of the North American Missourian Stage as it has long been defined by the lycospore extinction event and corresponding vegetational change, correlateable from the North American midcontinent to the Ukrainian Donets basin (Peppers 1996); and (3) the base of the Missourian defined by the lowest occurrence of the conodont Idiognathodus eccentricus, which is the base of the Exline cyclothem and thus stratigraphically higher than the lycospore extinction event (Heckel et al. 2002). This last boundary is correlated to the base of the middle substage, the Khamovnikian, of the Russian Kasimovian (Fig. 3). Two of us (SGL and WD) favour the second boundary, which has as its primary signal a substantial turnover event in the palaeoflora and some important changes in the marine biota (Fig. 3), instead of the other two boundaries, which are based on what Lucas (2018, 2019) has referred to as conodont ‘non-events’. This issue needs to be resolved by the Subcommission on Carboniferous Stratigraphy.

A long debate existed in the European literature regarding a gap between the top of the Westphalian (Middle Pennsylvanian) type section, in the German Saar coalfield, and the bottom of the Stephanian (Late Pennsylvanian) type section in the French/German Massif Central. Both type sections were characterized by floral successions and lacked marine indicators. One of the great twentieth century students of Carboniferous macrofloras and their use in biostratigraphy, Robert Wagner (1927–2018), was of the opinion that a temporal gap between these type sections was substantial. Therefore, he identified a large temporal hiatus between the Westphalian and Stephanian in Europe and thus between the Middle and Late Pennsylvanian as these concepts had been extended into the coal basins of eastern and Midwestern North America (Fig. 4). Wagner claimed that this hiatus was not present in northwestern Spain, and proposed that floras there bridged the gap in a continuous succession. On this basis he proposed the Cantabrian Stage (now used as a substage) to fill the perceived gap.

This modified the plant biostratigraphy that had been in use for decades. By positioning this new chronostratigraphic unit between the Westphalian and Stephanian, supposedly filling a gap existing everywhere else in Euramerica, Wagner and associates argued that the Cantabrian was the base of the Stephanian. Later work, however, by Wagner and his associates, correlated the base of the Cantabrian with strata outside of Spain, previously correlated with the type Westphalian in Europe and the Middle Pennsylvanian in the USA. Thus, there indeed were strata in the proposed gap elsewhere, not only in the USA, but in Central Europe and Britain. As a consequence of these correlations, the Cantabrian base disassociated the base of the Stephanian from the base of the Late Pennsylvanian, with which it had long been correlated in the USA. In this sense, the Cantabrian did not fill a gap, because many of the plant-fossil-characterized time horizons thought unique to NW Spain in fact occurred in other areas across the Euramerican part of Pangaea. Wagner also proposed type sections for the post-Cantabrian Barruelian stage (adopted by the ICS) and a Saberian stage (not yet adopted) to be equivalent to the intervals previously referred to as Stephanian A and B (Fig. 4).

Knight et al. (2023) argue the case for the Cantabrian Substage. They propose that a continuous sedimentary succession across the Middle–Upper Pennsylvanian boundary exists in the Cantabrian region of Spain, an extensive southern foreland to the Variscides (Besly and Cleal 2021). In this area of northwestern Spain the wetland biome persisted through the Late Pennsylvanian, and Knight et al. (2023) argue that it supports biostratigraphical correlations as far east as the Donbas region of the Ukraine. Four new high precision U–Pb CA-ID-TIMS radioisotopic ages from northern Spain support preliminary scaled dates for the Spanish succession: base of Asturian c. 310.7 Ma; base of the Cantabrian c. 307.5 Ma; base of the Barruelian c. 304.9 Ma; base of Saberian c. 303.5 Ma. An important point is that the Cantabrian as used by Knight et al. (2023) is equivalent to the Odontopteris cantabrica assemblage zone of Wagner (1984). However, the most significant floral event of the Middle–Late Pennsylvanian, the lycospore extinction event, is not evident in the Spanish section, which greatly limits its utility in correlation.

Nelson et al. (2023) critique the Cantabrian Substage, first reviewing the arguments made by Nelson and Lucas (2021), which maintain that the Cantabrian was never adequately defined, so it should never have been formalized as a chronostratigraphic unit. Correlations of the Cantabrian have been based solely on megafloral biozones, thus on assemblages considered typical but with variable species first and last occurrences, some before, some after and some within the biozone. No megafloral signal, therefore, corresponds uniquely to the base of the Cantabrian (e.g. Cleal et al. 2003).

Thus, at most, the Cantabrian corresponds to an assemblage zone of plant fossils in northern Spain, where the term is of local use. It does not serve as a chronostratigraphic unit with isochronous boundaries recognizable elsewhere in Europe, let alone outside of Europe.

Pfefferkorn (2023) also critiques the Cantabrian as a formal stage or substage. He observes that major biotic changes often took place over relatively short time intervals, and that well-defined stage boundaries often represent such times. The Cantabrian Stage was proposed in 1969 based on the assumption that floristic change occurs gradually, leaving a record of continuous species turnover. As a consequence of this assumption, a drastic change in terrestrial floras that characterized the Westphalian–Stephanian boundary in Europe was thought to indicate that strata were missing, and the resulting presumed ‘gap’ had to be closed by finding strata representing this ‘missing’ time elsewhere (Fig. 4). This conclusion reflects assumptions about the nature of evolution and ecological change that Pfefferkorn argues are incorrect, so that the Cantabrian Stage does not exist and cannot be recognized, either in its type area or elsewhere.

Important physical changes characterized the Late Pennsylvanian world, including the assembly of Pangaea, the accompanying tectonism of the Variscan, Appalachian and ancestral Rocky Mountain orogenies and the climate changes of the late Paleozoic ice ages. Iannuzzi et al. (2023) review the Pennsylvanian on the western rim of Gondwana to reveal that, across the South American continent, glaciomarine deposits and peat-forming environments to the south coexisted with marine carbonate platforms and aeolian dune fields further north. Thus, they present an overview of the main climatic-environmental events that took place across South America during the Pennsylvanian. These changes are associated with the floristic changes that occurred in the emergent lands based on palaeobotanical and palynological information.

Griffis et al. (2023) review the late Paleozoic glaciogenic records of the Paraná and Kalahari basins of southern Gondwana, which form one of the largest, best-preserved and well-calibrated records of this glaciation. The widespread deposition of marine sediments indicates that these basins transitioned from subglacial reservoirs to ice-free conditions during the latest Carboniferous. High-precision U–Pb zircon CA-TIMS dates from volcanic ash beds indicate that subglacial conditions in these regions were limited to the Carboniferous. The loss of ice in these regions is congruent with a late Carboniferous peak in pCO₂ and widespread marine anoxia. Griffis et al. (2023) stress that identifying a definitive driver for the greenhouse gases of the late Paleozoic ice ages, such as large igneous volcanism, sustained volcanic activity, or an increased biological pump, remains unresolved. The proposed Carboniferous peak for the high-latitude late Paleozoic ice age in Gondwana does not match the record from the low-latitude tropics, where evidence of an early Permian peak has been identified.

Heckel (2023) reviews the classic cyclothems of the North American Midcontinent, which are stratigraphic sequences deposited during glacial-eustatic transgressive–regressive depositional cycles. Heckel notes that major cyclothems are most complete on lower shelves and become separated by exposure surfaces higher on the shelf. Minor cyclothems extend only onto lower shelves, or are parasequences reflecting the reversal of sea-level during transgression or regression.

Control of glacial eustasy by interaction of Earth's orbital parameters, together with radioisotopic dates, allows subdivision of the cyclothem succession into inferred c. 400 kyr groupings, which reflects the longest eccentricity cycle and facilitates global correlation. Delineation of basinward extents of regression from the disappearance of exposure surfaces elucidates the history of glacial intensity and, in some instances, basin-margin development. Heckel notes that major cyclothems sometimes underwent reversals of a general sea-level trend, something found also in Pleistocene glaciations, and display ‘splays’ of lesser cycles onto the higher shelf. Geochemical cycles ‘nested’ within deepest-water shales on low shelves appear as thin nearshore facies between exposure surfaces at their high-shelf shorelines.

Cyclothems are the basis of an astrochronology that permits Paleozoic time to be deciphered at high resolution, as has been demonstrated in deep- and quiet-water deposits (e.g. Heckel 2013). Nevertheless, rigorous testing of astronomical forcing in low-latitude cyclothemic successions, which have a direct link to higher latitude glaciogenic records through inferred glacioeustasy, still needs a comprehensive approach that integrates new techniques with further optimization and additional independent age constraints (Montañez 2022).

Gulbranson and Tabor (2023) present a study of the Late Pennsylvanian palaeosol record that begins by noting that the Kasimovian was a time of ecological upheaval and large magnitude changes in palaeoclimate. Referred to as the ‘collapse’ of the palaeotropical rainforests, the Kasimovian is marked by rapid changes in megafloral plant communities and associated ecosystem effects on vertebrates and invertebrates. The pCO₂ variation that coincided with these ecological changes varied from pre-industrial levels (PAL) to 2 × PAL on 10⁵ year timescales. Deciphering the carbon cycle perturbations that led to these changes in CO₂ and the connection of these climate forcings to the terrestrial upheaval of palaeotropical rainforests remain a challenge. Gulbranson and Tabor (2023) assess the effects of palaeosol accumulation and/or degradation on the terrestrial carbon cycle during the Kasimovian. They survey palaeosols in ice-free depositional basins on Pangaea that are assessed for palaeolandscape equilibrium. An orbital framework is developed to understand the relationships of palaeosols, the carbon cycle and insolation. Based on these analyses a key time interval is identified in the early Kasimovian that records a shift in palaeolandscape equilibria, terrestrial carbon cycling and orbital forcing. The carbon cycling and landscape equilibria are eccentricity paced. Gulbranson and Tabor (2023) conclude that the dominance of short eccentricity throughout this interval indicates that the changes in palaeosols and the locus of carbon burial may have acted as a stochastic process.

Soreghan et al. (2023) discuss palaeo-loess and silt-rich aeolian-marine deposits long known in the Carboniferous–Permian strata of western equatorial Pangaea (USA), and increasingly recognized in eastern equatorial Pangaea (Europe). They note that aeolian-transported dust and loess appear earlier (Late Devonian) in the west, become common during the late Carboniferous, and predominate by the Permian. Indeed, the thickest loess deposits in Earth history (more than 1000 m thick) date from this time interval, and archive unusually dusty conditions in equatorial latitudes, especially compared to the scarcity of equatorial dust and loess in the modern world. These upper Paleozoic deposits record widespread silt generation, aeolian emission and transport, and accumulation in epeiric seas. Silt was sourced from contemporaneous orogenic belts, and glaciation in these mountains is one scenario capable of explaining the initial production of such voluminous material. In western Pangaea, large rivers drained the highlands and transported silt westward. The increase in dust deposition in post-Kasimovian time records widespread aridification that progressed across Pangaea, from west to east. According to Soreghan et al. (2023), abundant volcanism, especially at equatorial latitudes, created acidic atmospheric conditions and enhanced the nutrient reactivity of dust, likely affecting Earth's carbon cycle. The late Paleozoic was thus Earth's largest dust bowl, and this dust is an important archive and agent of both climate and climate change.

Opluštil and Schneider (2023) review Late Pennsylvanian stratal and fossil records in Europe, North Africa and Asia Minor, where the continental deposits of Pennsylvanian sedimentary basins are either intercalated with shallow marine strata or were deposited in exclusively continental settings. Lengthy and extensive research on these deposits has led to the definition of regional stages and substages based on marine fauna and terrestrial flora, later augmented by terrestrial and freshwater faunal biostratigraphy. Isochroneity in these strata can be established by glacioeustatically driven marine bands, but where such bands are missing only non-marine biostratigraphic control exists. Precise correlation between the upper Paleozoic strata in the basins in the region examined is limited by the resolution of biostratigraphic zonations combined with gaps in sedimentary successions and the variable quality of the fossil record.

No articles in this volume specifically address the trace fossil (ichnological) record across the Middle–Late Pennsylvanian boundary, so we briefly review this record here. There is no evidence of a substantial change in marine or non-marine invertebrate ichnology across the Middle–Late Pennsylvanian boundary. No substantial change in ichnodiversity and no major behavioural innovations are indicated by the invertebrate trace fossil record. Thus, in a review of Phanerozoic trace fossil ichnodiversity, Buatois and Mángano (2018, table DR1) listed the temporal ranges of 428 ichnogenera of bioturbators and 106 bioerosion ichnogenera. Of the bioturbation ichnogenera, only two (Acripes, Huilmuichnus) have Middle Pennsylvanian first appearances, and three ichnogenera (Microspherichnus, Mirandarichnium and Oklahomaichnus) have Late Pennsylvanian first appearances. Two ichnogenera (Arthrophycus and Trichophycus) have Late Pennsylvanian last appearances. Of the bioerosion ichnogenera, only two (Polyactina, Rhoipalia) have Middle Pennsylvanian first appearances, and only one ichnogenus (Bascomelia) has a Late Pennsylvanian last appearance. And, other than Arthrophycus, these are all rare ichnogenera, some of questionable validity (such as Oklahomaichnus, a likely synynom of Diplichnites: Lucas and Lerner 2001).

Indeed, the ichnodiversity compilation of Buatois and Mángano (2018, table DR2) shows essentially no change in generic ichnodiversity across the Middle–Late Pennsylvanian boundary, in either continental or marine settings (also see Minter et al. 2016). There is also little to no change in generic ichnodiversity across the Early–Middle Pennsylvanian boundary; only in the early Permian is there a pronounced jump in continental generic ichnodoversity, from 55 Late Pennsylvanian ichnogenera to 91 in the early Permian.

Buatois and Mángano (2018) noted that during the Phanerozoic, behavioural innovations recorded by new ichnotaxa took place in pulses, with particularly large changes during the early Cambrian (Terraneuvian), corresponding to the so-called ‘Cambrian explosion’, and during the Ordovician, overlapping what has been called the ‘great Ordovician biodiversification event’. What has been called the Paleozoic evolutionary fauna refers to seafloors that were dominated by articulate brachiopods, crinoids, and rugose and tabulate corals (Sepkoski 1981). Established by the end of the Ordovician, the Paleozoic evolutionary fauna persisted until the Permo-Triassic extinctions ushered in seafloors with mollusc-dominated communities. During most of the duration of the Paleozoic evolutionary fauna (including during the Pennsylvanian) there appears to have been a limited increase in overall ichnodiversity and a general lack of behavioural innovations in the marine ichnofauna.

This is not a reflection of a lack of record or a lack of understanding of Pennsylvanian marine ichnoassemblages, which are numerous and have been well studied. For example, in Kansas, USA, marine ichnofaunas span the whole Pennsylvanian Subsystem and have been the subject of diverse research (see review by Mángano et al. 2002). Mángano et al. (2002) studied Pennsylvanian tidal-flat ichnofaunas and note that they are bivalve dominated, in contrast to the trilobite-dominated ichnofanuas of the early–middle Paleozoic. This does not change until the Mesozoic, when infaunalization produced relatively deep burrowing in crustacean-dominated tidal-flat ichnoassemblages. A high diversity of shallow tier traces characterizes late Paleozoic tidal-flat ichnofaunas, whereas a lesser diversity of deep tier ichnoassemblages characterizes the Mesozoic–Cenozoic. The change seems to have taken place primarily across the Permo-Triassic boundary when extinctions reset the seafloor community.

A more granular comparison of, for example, the tidal-flat Late Pennsylvanian Wavery ichnoassemblage in Kansas with older Pennsylvanian ichnoassemblages from similar environmental settings reveals much overlap in ichnotaxa and almost complete overlap of ichnoguilds and architectural designs (Mángano et al. 2002; Buatois et al. 2005). Thus, there is no evidence of major ichnological changes in the marine realm across the Middle–Late Pennsylvanian boundary.

As noted above, on land, across the Middle–Late Pennsylvanian boundary, a profound transformation took place in the vegetation, but it was not a single event, having begun during the Moscovian and extending through the Kasimovian, taking place at different times in different parts of Pangaea (e.g. DiMichele et al. 2023). The original claim that this caused a substantial tetrapod extinction and drove tetrapod endemism has not withstood subsequent analysis (Dunne et al. 2018; Lucas 2023). There still were, however, important changes in the tetrapod biota during the Middle–Late Pennsylvanian transition. Nevertheless, there are no clear and substantial changes in the tetrapod trace-fossil record across the Middle–Late Pennsylvanian boundary.

The change in the tetrapod footprint record that may correspond to the base of the Kasimovian is reflected in the Notolacerta–Dromopus tetrapod footprint biochron boundary. This has been thought to coincide with the Kasimovian–Gzhelian transition (Fillmore et al. 2012; Voigt and Lucas 2018; Lucas 2019; Schneider et al. 2020; Lucas et al. 2022), though it could be older, within the Kasimovian. Thus, note that there are two possible Kasimovian records of Dromopus, in Morocco and in England (Hmich et al. 2006; Meade et al. 2016). Gzhelian (Virgilian) occurrences of Dromopus include its type material from the Howard Limestone (Wabaunsee Group) in Kansas (Marsh 1894) and the Thuringian Forest basin in Germany (Voigt 2005, 2012). The Dromopus biochron extends through a large part of the early Permian (Voigt and Lucas 2018), and Gzhelian records include other characteristic early Permian tetrapod footprint ichnogenera, including Batrachichnus, Dimetropus, Ichniotherium and Limnopus.

However, other than Dromopus, most of the tetrapod ichnogenera of the Dromopus biochron have older Pennsylvanian (and, in some cases, Mississippian) records. There are also much older records of diadectomorph footprints (ichnogenus Ichniotherium) and of eupelycosaur footprints (ichnogenus Dimetropus) from the Bashkirian of Germany (Voigt and Ganzelewski 2010). These records substantially predate the oldest body-fossil records of their inferred trackmakers, which are no older than Kasimovian. The Bashkirian Ichniotherium are particularly interesting as possibly representing the oldest fossils of a tetrapod high-fibre herbivore. These Bashkirian footprint records thus diminish the extent of any Kasimovian event in tetrapod evolutionary history by pushing back the record of two significant tetrapod groups (diadectomorphs and eupelycosaurs minus varanopids) thought to have first appeared during the Kasimovian. A much more extensive tetrapod footprint record from Bashkirian and Moscovian strata may push back the supposed Kasimovian or Gzhelian first appearances of other tetrapod ichnotaxa. The tetrapod footprint record thus seems to diminish the apparent evolutionary turnover of tetrapods during the Kasimovian indicated by their body-fossil record.

There is a major change in the terrestrial flora across the Middle–Late Pennsylvanian boundary that has been hyperbolized as the ‘Carboniferous rainforest collapse’ (Sahney et al. 2010). That term should be abandoned, as the floristic changes were prolonged and complex, and they represent the replacement of one kind of rainforest by another (Fig. 5), not a disappearance of rainforest vegetation, something that occurred multiple times during the Pennsylvanian.

DiMichele et al. (2023) address these changes by noting that three floristic events (changes) can be identified leading up to and across the Moscovian–Kasimovian boundary (Montañez 2016). Two significant changes took place during the Middle Pennsylvanian, first in the middle Moscovian (Atokan–Desmoinesian; ∼Bolsovian–Asturian), the second in the late Moscovian (mid-Desmoinesian; mid-Asturian), and the third, a threshold-like vegetational change in tropical wetlands, occurred in the early Kasimovian (the US Desmoinesian–Missourian boundary) (Phillips et al. 1974; Pfefferkorn and Thomson 1982; Cleal 1984, 1997; Wilson 1984; DiMichele and Phillips 1996; Peppers 1997; Cleal et al. 2003; DiMichele et al. 2011). These changes took place during a time period of dynamic and complex physical change in Euramerican Pangaea driven by changes in polar ice volume and accompanying changes in sea-level, atmospheric circulation, rainfall and temperature (Phillips and Peppers 1984; Cecil 1990; Heckel 1991, 2008, 2023; Rygel et al. 2008; Montañez 2016; Matthaeus et al. 2022). During the early Kasimovian turnover, the so-called ‘Carboniferous rainforest collapse’ (actually a threshold-like ‘turnover’; DiMichele et al. 2009), lycopsid dominance of (mostly peat) swamps changed to marattialean tree-fern and medullosan pteridosperm dominance (Fig. 5), and there was a decrease in biodiversity. This change took place quickly, during one glacial–interglacial cycle, and included vegetational turnover in other wetland habitats. For several subsequent glacial–interglacial cycles peatland dominance varied (based on palynology; Peppers 1984, 1996; Wilson 1984; Kosanke and Cecil 1996) before stabilizing. These vegetational changes likely reflect climatic events driving unidirectional, non-reversible wetland vegetational changes, during the cooler, wetter parts of glacial–interglacial cycles (DiMichele et al. 2009).

Schachat et al. (2023) analyse vegetational change during the Middle–Late Pennsylvanian transition in far western equatorial Pangaea (New Mexico, USA). The study is based on the largest database assembled from this region: more than 40 quantitatively analysed floras from 26 stratigraphic levels. Most of the sampled floras, both below and above the Middle–Upper Pennsylvanian boundary, are ‘mixed’ (sensu Bashforth et al. 2021), including both hygromorphic and xeromorphic taxa. Schachat et al. (2023) recalibrated the taxonomic data morphometrically by focusing on foliar traits of lamina width and venation, and the data were examined using stratigraphic credible intervals, capture–mark–recapture analyses, and resampling analyses. The results indicate that no substantive taxonomic turnover occurred across the Middle–Late Pennsylvanian boundary in this western Pangaean region, in marked contrast to patterns in mid-Pangaean coal basins where there was a large wetland vegetational turnover (e.g. Phillips et al. 1974; Opluštil et al. 2022). However, because the plant data are quantitative rather than just presence–absence, changes in proportional composition can be tracked. In New Mexico, immediately following the Middle–Late Pennsylvanian boundary, and for approximately half of the Missourian Stage, floras previously dominated by mesomorphs and hygromorphs became overwhelmingly dominated by xeromorphic taxa. Although expressed differently, the timing of the western Pangaean physical and palaeobotanical changes parallels those from mid-Pangaean coal basins and suggests a widespread though diachronous environmental change across the Middle–Late Pennsylvanian boundary.

Wilson et al. (2023) further discuss the series of vegetational changes that took place in tropical ecosystems during the Pennsylvanian. Using palynology and plant macrofossils, the most notable change can be recognized at the Middle–Late Pennsylvanian boundary in the Illinois Basin. This is the extirpation of certain lineages of arborescent lycopsids, followed by their replacement by stem-group marattialean tree ferns. The leading hypothesis suggests a sudden drought event as the cause of this change (Phillips and Peppers 1984; Winston 1990). To test this hypothesis, Wilson et al. (2023) examined the vascular anatomy and physiology of key lineages of Pennsylvanian plants and provide new data on the vascular systems of these plants, quantifying their physiological capacity and drought resistance. According to Wilson et al. (2023), three Pennsylvanian plant lineages – the medullosans, lycopsids and the facultative groundcover/vine Sphenophyllum – had high hydraulic conductivity but were vulnerable to drought-induced damage, whereas others are resistant, including stem-group tree ferns and coniferophytes. Relative abundance changes among these plants were likely driven by drought; differences in water-use efficiency would have amplified drought events as plant communities changed. Thus, Wilson et al. (2023) conclude that the interaction of physiological selectivity and positive feedback between aridity and drought tolerance likely played a role in floral changes across the Middle–Late Pennsylvanian boundary.

Three chapters in this book are devoted to the invertebrate fossil record across the Middle–Late Pennsylvanian boundary; two on conodonts (Barrick et al. 2022b; Hu et al. 2023) and one on insects and other terrestrial arthropods (Donovan et al. 2022). In addition, patterns in some marine invertebrate groups across the Middle–Late Pennsylvanian boundary are discussed in Heckel (2023). Missing are reviews of fusulines and the shelly marine benthos, so we offer short summaries here.

The beginning of the Late Pennsylvanian marks an important turnover in the evolutionary history of fusulines, the dramatic decline of the Fusulinidae (which dominated fusulinid assemblages during the Early–Middle Pennsylvanian) and the origin of the Schwagerinidae (Fig. 6). The schwagerinids had a key evolutionary innovation, the keriothecal (honeycombed) test wall, presumably to allow photosynthetic symbionts (algae) to live in the fusulinid test. This is the beginning of the diversification of the schwagerinid Triticites, which came to dominate Late Pennsylvanian fusulinid assemblages in North America. The first schwagerinid genus, Montiparus, appears at the base of the Khamovnikian (Fig. 6). The lowest occurrence of this distinctive and widespread fusulinid genus has been advocated as a primary signal for defining the base of the Kasimovian (Villa and Task Group 2001, 2004; Ueno 2022).

At the beginning of the Late Pennsylvanian (the beginning of the Khamovnikian subage and Missourian age; the base of the Missourian and the base of the Kasimovian, as presently defined, do not coincide, and thus represent two different locations for the Middle–Late Pennsylvanian boundary), important evolutionary turnover in some marine taxa took place, notably fusulinids, brachiopods, sponges, ammonoids and conodonts, and there have been claims of slow rates of origination and extinction, so-called ‘sluggish evolution’ (e.g. Boardman et al. 1989; Stanley and Powell 2003; Powell 2005). However, none of these changes can be characterized as a major origination, extinction or evolutionary event outside of relatively small clades.

In North America, extinctions of some key marine invertebrate taxa took place across the Middle–Late Pennsylvanian boundary (Fig. 4), including extinction of the fusulinid Beedeina, of the ‘sponge’ Chaetetes, of the chonetid brachiopod Mesolobus, and of the ammonoids Wellerites, Goniolyphoceras, Wewokites and Eothalassoceras (Boardman et al. 1989, 1990; Barrick et al. 1996; Heckel 2002, 2023). Among the conodonts, Neognathodus and Swadelina go extinct. The Late Pennsylvanian saw the first appearances of the fusulinids Eowaeringella and Triticites, the microgastropod Plocezyga costata and the ammonoids Parashumardites and Pennoceras.

Fan et al. (2020) and Shi et al. (2021) recently presented an analysis of a Chinese database (Geodiversity Database) of marine invertebrate diversity during the late Paleozoic that identifies what they call the ‘Carboniferous–Permian biodiversification event’. This increase in diversity rivals the great Ordovician biodiversification ‘event’ in magnitude, and is not consistent with ideas of sluggish evolution. This Carboniferous–Permian ‘event’ was punctuated by a diversity drop that began before the Kasimovian and mainly involved losses among the foraminiferans and brachiopods according to the Chinese database (Shi et al. 2021). The global Paleobiology Database shows a similar drop in diversity, not only in brachiopods and foraminiferans, but also in gastropods and cephalopods. Shi et al. (2021) propose that this diversity drop is related to climate changes that drove the vegetation changes on land across the Middle–Late Pennsylvanian boundary. However, the lack of changes in most marine invertebrate groups and the lack of substantial ichnological changes suggest that this drop is likely biased by the sheer abundance of foraminiferal and brachiopod taxa and does not reflect an important extinction.

It is important to note that recently published reviews of various Carboniferous marine macroinvertebrate records for biostratigraphic purposes identified no substantial evolutionary events across the Middle–Late Pennsylvanian boundary (ammonoids; Nikolaeva 2022; bivalves: Amler and Silantiev 2022; brachiopods: Angiolini et al. 2022; crinoids: Ausich et al. 2022; rugose corals: Wang et al. 2022). A similar review of smaller (non-fusuline) foraminiferans also identified no substantial evolutionary events across the Middle–Late Pennsylvanian boundary (Vachard and LeCoze 2022). Thus, there seems to be little evidence that climate change across the Middle–Late Pennsylvanian boundary drove any substantial changes in marine animal communities.

Barrick et al. (2022b) review the Late Pennsylvanian evolution of conodonts. They begin by noting that Late Pennsylvanian conodont faunas were dominated by idiognathodids historically assigned to Idiognathodus (untroughed) or Streptognathodus (troughed). Recent work suggests that conodont clades arose iteratively, through time, from unrelated ancestors in different geographical regions. The end-Desmoinesian extinction event terminated two conodont genera, Swadelina (troughed) and Neognathodus (long carina), and comparable new morphotypes developed in the early Kasimovian from surviving Idiognathodus species, especially in North America (Rosscoe and Barrick 2013). A second, early Gzhelian, Eurasian radiation produced new forms that dominated Gzhelian faunas globally. After a low diversity interval in the mid-Gzhelian, a new major radiation occurred, which led to redevelopment of Idiognathodus-like elements in the Cisuralian (Barrick et al. 2022a).

Hu et al. (2023) review the late Moscovian–early Gzhelian conodonts present in abundance in a newly discovered slope section, the Shanglong section in southern Guizhou, China. This conodont fauna is dominated by P1 elements of Idiognathodus, and less common are Swadelina, Streptognathodus, and Heckelina. A total of 62 species are identified and assigned to eight genera by Hu et al. (2023). A numerical cluster technique identifies four sub-biofacies of the slope settings.

In a switch to the terrestrial realm, Donovan et al. (2022) review the record of plant–arthropod interactions in order to analyse origination and extinction rates of insects during the Middle–Late Pennsylvanian transition. The major plant turnovers during this time period raise the question of what happened to other groups of terrestrial organisms. The analysis of Donovan et al. (2022) indicates that plant–arthropod associations broadly persist through the Middle–Late Pennsylvanian boundary, and new damage types and host-plant associations first appeared during the Late Pennsylvanian. Their analysis suggests that insects did not suffer major extinctions or turnover during the Middle–Late Pennsylvanian, despite short- and long-term changes in climate and environmental conditions.

Lucas (2023) analyses the Middle–Late Pennsylvanian fossil record of tetrapod vertebrates, amphibians and amniotes (synapsids plus reptiles). The Late Pennsylvanian was a critical juncture in tetrapod evolution when many terrestrially adapted taxa first appeared. During the Late Pennsylvanian, body fossils of several important new tetrapod taxa appeared, including the Eryopidae, Cochleosauridae, Dissorophidae, Discosauriscidae?, Trimerorhachidae, Diadectidae, Petrolacosauridae, Varanopidae, Sphenacodontidae and Edaphosauridae (e.g. Reisz 1986; Milner 1987; Berman et al. 1997; Lucas 2004, 2022; Carroll 2009; Lucas et al. 2018; Schneider et al. 2020). Before the Late Pennsylvanian, tetrapod-body fossils indicate no direct consumption of high-fibre plant foods by tetrapods. Then, during the Late Pennsylvanian, the first obligate high-fibre herbivores appear in the tetrapod body-fossil record, the diadectomorphs, bolosaurids, caseasaurs, and edaphosaurid eupelycosaurs (e.g. Hotton et al. 1997; Sues and Reisz 1998; Reisz and Sues 2000; Reisz and Fröbisch 2014; Modesto et al. 2015; Spindler et al. 2016; Lucas et al. 2018).

The Moscovian tetrapod record reflects a taphonomic megabias that favoured the preservation, collection and study of aquatic tetrapods that lived in wetland palaeoenvironments (‘coal swamps’). However, the Kasimovian tetrapod record is limited to seven localities, all but one in the USA, and two of which are singleton records, so it is less abundant, diverse or widespread than earlier Moscovian and later Gzhelian tetrapod records. According to Lucas, this ‘Kasimovian bottleneck’ hinders interpretation of tetrapod evolutionary events across the Middle–Late Pennsylvanian boundary. Significant changes did take place across that boundary, but a granular examination of the tetrapod fossil record indicates that they were spread out over Moscovian through Gzhelian time. Many of the perceived changes in tetrapods across the Middle–Late Pennsylvanian boundary are largely artefacts of facies changes and the Moscovian tetrapod taphonomic megabias. Therefore, there is no simple link between Late Pennsylvanian tetrapod evolutionary events and changes in climate and vegetation.

All of the topics discussed in this book need more data and analysis. A GSSP for the base of the Kasimovian (Upper Pennsylvanian) is a needed step to stabilize and standardize the chronostratigraphy of the Middle–Upper Pennsylvanian boundary. The confusion caused by the Cantabrian can be overcome by recognizing that the Cantabrian is not a chronostratigraphic unit of value to broad correlations, but instead is a floristic assemblage zone of some value to correlations in Spain and other parts of Europe. More precise time constraints are needed to better understand the Pennsylvanian–early Permian ice ages. And, loess and palaeosols provide important climate archives and proxies that merit more extensive analysis. The idea of a ‘rainforest collapse’ across the Middle–Late Pennsylvanian boundary needs to be abandoned. Patterns of evolution of marine and non-marine animals show no major origination or extinction events across the Middle–Upper Pennsylvanian boundary that can be connected to changes in climate or vegetation.

We thank all the contributors to this volume for their patience and dedication. We also are grateful to the reviewers whose comments improved the content and clarity of the manuscript.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.