In Clay County, of the north-central region of Texas, USA, Pennsylvanian and Permian rock strata predominantly consist of continental deposits (DiMichele et al., 2018b). Alluvial sediments accumulated in the eastern portion of a coastal plain that bordered the eastern platform of the Midland Basin. The geological units identified in Clay County include the Markley Formation (Late Pennsylvanian to lowermost Permian), Archer City Formation (lower Permian, Wolfcampian Series), Nocona Formation (lower Permian, Wolfcampian Series), and Petrolia Formation (lower Permian, Leonardian Series) (Hentz, 1988). Some of these stratigraphic units are time transgressive.

The Archer City Formation, together with the Markley Formation, encompass the Bowie Group (Pennsylvanian–Permian) ( Figure 1 ). The lithological composition of the Archer City Formation primarily consists of clastic deposits, mainly of claystone, with smaller components of siltstone, sandstone, and conglomerate (Hentz, 1988). The thickness of the formation varies from 107 to 122 m and includes strata of sandstone with thicknesses of approximately 15 m. The plants are well-preserved but not to the extent of the somewhat older Williamson Drive material (Xu et al., 2018). Nonetheless, SR plant fossils exceed the threshold for the detection of the most inconspicuous of interactions, punctures made by piercing-and-sucking insects. Compared to other examined plant assemblages from the latest Pennsylvanian to early middle Permian of Texas, the preservation at Sanzenbacher Ranch is somewhat less informative than Williamson Drive (Xu et al., 2018), more informative than Coprolite Bone Bed (Labandeira and Allen, 2007) and South Ash Pasture (Maccracken and Labandeira, 2020), and of similar preservational style as Taint (Beck and Labandeira, 1998), Colwell Creek Pond (Schachat et al., 2014), and Mitchell Creek Flats (Schachat et al., 2015).

The Sanzenbacher Ranch plant assemblage deposit is situated in the middle region of the Archer City Formation ( Figure 1 ) and is pegged to the Asselian Stage of the early Cisuralian. Based on the field notes of DiMichele et al. (2018a), the Sanzenbacher Ranch plant assemblage is predominantly composed of pteridosperms, followed by conifers. Within the pteridosperms, the callipterid-peltasperms are the most abundant plant group at the outcrop, where Autunia conferta is the most prevalent plant, followed by Rhachiphyllum schenkii and cf. Lodevia. The most abundant medullosan is Neurodontopteris auriculata, along with Odontopteris spp. A lyginopterid is represented by Sphenopteridium germanicum. Coniferopsids constitute the second most abundant plant group, although the identification process of these specimens is challenging. Coniferopsids include Walchia sp. 1, Cordaites sp. 1 and fragments of conifer scales, cones, and bracts (DiMichele et al., 2018a).

Sphenopsids are represented by the calamitaleans Annularia spicata, Annularia carinata, and Asterophyllites equisetiformis; sphenophylls are represented by the Sphenophyllum verticillatum, Sphenophyllum thonii, Sphenophyllum angustifolium, and Sphenophyllum oblongifolium. The reproductive organs for calamitaleans are identified as Calamostachys sp. 1 and Palaeostachya sp. 1 (DiMichele et al., 2018a). The two groups of ferns occurring at the outcrop are arborescent marattialeans (Marattiales) and herbaceous filicaleans (Filicales). Due to the degree of fragmentation of the material, identifications for sterile specimens are provisional, designated, for example, as Pecopteris sp. 1 or as marattialean foliage. Fertile specimens of Asterotheca sp. 1 are also present (DiMichele et al., 2018a). Some specimens have been identified as Pecopteris monyi, Pecopteris hemiltelioides, Pecopteris permica, Oligocarpia gutbieri, Oligocarpia cf. gutbieri, and Sphenopteris sp. 1 (DiMichele et al., 2018a). Other fern fossils also are found in the Sanzenbacher Ranch plant assemblage, with some lacking a well-defined taxonomic affinity, and others possibly associated with seed plants. These other plant fossils include Taeniopteris sp. 1, a likely seed plant, and Yuania, a noeggerathialean, recently shown to be a pteridophyte (Yang and Wang, 2012). Also present are reproductive structures such as Trigonocarpus sp. 1 and various unaffiliated winged seeds (DiMichele et al., 2018a).

The studied material comes from the Sanzenbacher Ranch locality ( Figure 1 ), and the plant assemblage from this locality is reposited in two collections with three locality numbers: NPL 1161, USNM 40600, and USGS 9999. The specimens were collected during three separate field expeditions. The first collection was made between 1940 and 1941 as part of the Statewide Paleontological–Mineralogical Survey conducted by the Works Progress Administration (WPA), led by Adolph H. Witte (DiMichele et al., 2018a). These specimens are currently at the Non-vertebrate Paleontology Laboratory at the University of Texas at Austin, under the locality number NPL 1161. The second collection, made during the 1960s, was conducted by Sergius H. Mamay and Arthur D. Watt and is cataloged under the locality number USGS 9999. Last, a field trip in 1991, including Mamay, William DiMichele, Dan Chaney, and Robert Hook collected fossils at the same locality as the second collection. Specimens from this collection are cataloged under locality number USNM 40600. The USNM (40600) and the USGS (9999) specimens are housed at the Smithsonian’s National Museum of Natural History, in Washington, DC.

Paleoecological studies have revealed differences among the three collections. The USGS (United States Geological Survey) collection is dominated by a single species, A. conferta, which comprises more than 30% of the specimen occurrences. By contrast, the USNM (United States National Museum, later renamed the National Museum of Natural History or NMNH) collection is dominated by three species with >10% of the occurrences attributable individually to A. conferta, Neurodontopteris auricular, and collectively several species of Pecopteris. The last collection, the NPL material, encompasses four species each accounting for >10% of the occurrences: A. spicata, Calamites stems, Walchia sp. 1, and A. conferta (DiMichele et al., 2018a).

All plant specimens present in the three collections were analyzed, including leaves, fronds, reproductive structures (seeds and cones), stems and other axes, and wood impressions. Photography and data collection were conducted in a Paleobiology laboratory at the NMNH and at the Non-vertebrate Paleontology Laboratory (The University of Texas at Austin) using a Leica StereoZoom 6 Plus stereomicroscope to identify and record instances of herbivory and pathogens. An important attribute in evaluating this herbivory was the presence of a reaction rim and other tissue deformations attributable to herbivory and teratologic structures such as galls (Meyer and Maresquelle, 1983). All specimens were photographed using a Canon EOS 50D and a Canon EOS Rebel T3 camera with a Canon EF-S 60mm macro lens for documentation and assembling additional data for quantitative analyses.

We recorded presence–absence data for damage types (DTs) consistent with production by insects, mites, and pathogens. The damage was classified into functional feeding groups based on the available literature (Labandeira et al., 2007 and subsequent addenda). Nine functional feeding group categories were found: hole feeding, margin feeding, surface feeding, piercing and sucking, oviposition, galling, seed predation, borings, and pathogens. For rare unassignable damage, descriptions were provided. The number of feeding event occurrences (FEOs) were tallied, which indicate how many times a feeding event of a particular DT occurs in on single plant organ such as a leaf (Xiao et al., 2022a, b, c).

The program ImageJ 1.53t (Image processing and analysis in Java Wayne Rasband and contributors, National Institutes of Health, USA) was used to assess the surface areas of the plant specimens. Photos of the specimens were used to calculate the total surface area (TSA) and total herbivorized surface area (THSA) for each specimen and collectively for the plant assemblage. These two measurements, each summed over the entire plant assemblage, were employed to calculate the herbivory index (HI) for the plant assemblage. To evaluate the herbivory index, we applied the method used by Beck and Labandeira (1998):

And for the proportion of herbivorized leaves (PHL):

Herbivory was assessed in terms of richness and intensity of interactions. Richness was determined by DT diversity for the total plant assemblage and for each plant taxon and morphotype. Intensity was determined by how often, or the frequency of, the DTs that were present in the plant assemblage and for each taxon in the plant assemblage.

We evaluated the relationships between functional feeding groups and taxa in the six other plant assemblages from the uppermost Pennsylvanian to early middle Permian interval of north-central Texas. Ordination methods, consistent with those employed by previous authors, were used (Schachat et al., 2015; Xu et al., 2018; MacCracken and Labandeira, 2020; also see Schachat et al., 2022). We constructed a dissimilarity matrix obtained from a Bray–Curtis index that was inputted into a Non-metric Multidimensional Scaling (NMDS) ordination analysis and carried out using the vegan package (Oksanen et al., 2022) in R (R Development Core Team, 2013). This analysis was performed using the “metaMDS” function. The analysis resulted in two separate NMDS plots. The first plot illustrates the relationship between the FFGs and leaf taxa. The second plot shows the relationships of the most herbivorized functional feeding groups and plant taxa for each plant assemblage present during the late Paleozoic interval in Texas.

We analyzed 1,361 specimens from the SR plant assemblage ( Supplementary Table S1 ) across three collections: NPL 1161 collection with 531 specimens, USNM 40600 with 311 specimens, and USGS 9999 with 519 specimens. Each of the three collections exhibited a significantly different taxonomic composition. The NPL collection displayed an elevated presence of calamitaleans, the USNM collection was dominated by A. conferta, and the USGS collection had the highest percentage of morphotype seed sp. 9. Collectively, pteridosperms were the most abundant group, of which A. conferta was the most abundant pteridosperm taxon, followed by R. schenckii. In addition, all three collections contained a significant number of reproductive organs and stems, such as seed morphotype sp. 9 and calamitalean axes, respectively.

Pteridosperms were the most abundant group, accounting for 29.76% of the total plant assemblage organs, and included lyginopterids, medullosans, callipterid peltasperms, and an unidentified pteridosperm ( Supplementary Table S2 ). Peltasperms comprised 19.25% of the plant organs of the total plant assemblage that included callipterid leaves of A. conferta, R. schenkii, cf. Lodevia, cf. and Comia sp., and female reproductive structures of Peltaspermum sp. Medullosans, represented by the genera Neurodontopteris, Odontopteris, and cf. Eusphenopteris made up 6.47%. Lyginopterids, represented solely by Sphenopteridium germanicum, accounted for 3.97%.

Seeds accounted for 25.5% of the total plant assemblage, with all specimens unattached to reproductive structures. The seeds have been classified by the authors into nine seed morphotypes and one identifiable Trigonocarpus sp. 1. Seed sp. 9 was the most abundant morphotype at the Sanzenbacher plant assemblage. Notably, small, isolated, rounded seeds were prevalent, along with two other morphotypes with greater representation: seed sp. 6 and seed sp. 5.

Sphenopsids, mainly composed of whorls foliage of Annularia and stems of Calamites sp. 1, accounted 15.43% of the occurrences, consisting of 210 specimens. The third most abundant group of the Sanzenbacher plant assemblage was comprised of calamitalean stems, followed by A. spicata, which was notably abundant in the NPL 1161 collection. Marattialean, filicalean, and other ferns also made a notable contribution to the plant assemblage, representing 7.42% of all specimens. The most abundant fern was Pecopteris consisting of several species treated here as a single group, followed by Asterotheca sp. 1, and Sphenopteris sp. 1.

Other groups were less represented, each having fewer than 10 specimens. These included cycadopsids, ginkgopsids, and Noeggerathiales. Unidentified specimens found in the USGS 9999 and NPL 1161 localities could not be identified due to fragmentary material and whitish surfaces on some specimens, making it impossible to discern venation or other important foliar features. Lycopsids were absent.

The SR plant assemblage had 152 specimens with at least one instance of a DT caused by insects, mites, or pathogens (11.17%). We observed 48 different DTs, 205 DT occurrences, and 547 feeding event occurrences (FEOs) across 31 plant morphotypes ( Table 1 ). The callipterids collectively hosted nearly half of the recorded DTs, comprising 49.34% of the total ( Supplementary Table S2 ). The taxon with the greatest DT richness was A. conferta (18 DTs and 58 DT occurrences), followed by R. schenkii (16 DTs and 41 DT occurrences). The lyginopterid S. germanicum was the third group with the greatest richness (11 DTs and 17 DT occurrences), with 8.55% of the total. Cordaites sp. 1 and N. auriculata presented nine DTs, whereas cf. Lodevia sp. had eight DTs, represented by 7.24%, 6.58%, and 3.95%, respectively. The Sanzenbacher plant assemblage had a total surface area of 8,687.86 cm² and an herbivorized surface area (HSA) of 22.48 cm² ( Supplementary Table S1 ). The taxon with the greatest surface area was A. conferta, with 5,212.77 cm², 60% of the total, and an HSA of 6.11 cm², comprising 27.2% of the total herbivorized area. In the second place is R. schenckii, which had 551.43 cm² (6.35%) and an HSA of 5.811 cm² (25.84%). (More information regarding the SA and HSA are available at the Supplementary Table S5 ).

This study reports the presence of nine functional feeding groups, and 48 DTs for a plant assemblage with a herbivory index of 0.26. The same functional feeding groups have been described for the nearby early Kungurian, Colwell Creek Pond (CCP) plant assemblage ( Figure 11 ), where margin feeding, hole feeding, surface feeding, piercing and sucking, oviposition, galling, seed predation, and borings were present (Schachat et al., 2014). The Williamson Drive (WD) plant assemblage presented a similar suite of functional feeding groups, except for the absence of seed predation and the additional presence of rare skeletonization (Xu et al., 2018). The WD and CCP plant assemblages consist of two suites that have a high richness of DTs during the late Gzhelian to early Kungurian stages. The Sanzenbacher Ranch (SR) plant assemblage is of intermediate late Asselian age and is consistent with a pattern of a high DT richness from the latest Pennsylvanian to the early Permian.

External foliage feeding, consisting of the three functional feeding groups of hole feeding, margin feeding, and surface feeding, constituted the most abundant source of herbivory at SR, compared to the remaining functional feeding groups. Other Permian plant assemblages displayed a similar pattern with these three externally feeding functional feeding groups, although there is a different composition regarding DTs within these plant assemblages. The younger plant assemblages of Mitchell Creek Flats (MCF), CCP, and South Ash Pasture (SAP) present the same external foliage feeding functional feeding groups as shown at SR. Alternatively, the three older plant assemblages of WD, Coprolite Bone Bed (CBB), and Taint displayed the presence of all four functional feeding groups of external foliage feeding, which notably includes skeletonization ( Figure 11 ). There is a similarity regarding the prominent presence of margin feeding in most Permian plant assemblages.

The individual functional feeding groups of external foliage feeding shows several patterns. Margin feeding had the highest scores for the SAP, SR, MCF, and CBB plant assemblages ( Table 2 ). The CBB plant assemblage notably documented 70.07% of all DT occurrences as margin feeding (Labandeira and Allen, 2007). By contrast, the SR plant assemblage had 42.44% of all DT occurrences allocated to margin feeding, accounting for less than half of all DT occurrences. The SAP and MCF plant assemblages had margin feeding percentages of 36.59% and 22.41%, respectively (Schachat et al., 2015; MacCracken and Labandeira, 2020), whereas the CCP and WD plant assemblages displayed 11.08% and 4.04% margin feeding, respectively. For all late Paleozoic plant assemblages, the most common damage was DT12 and DT13, consisting of a broad (level 1) damage-type functional breadth, formerly considered as “generalized”. The updated host specificity categories are defined as follows: broad damage-type functional breadth (level 1), intermediate damage-type functional breadth (level 2), and narrow damage-type functional breadth (level 3), based on a revision of the categories of generalist–intermediate–specialist (Labandeira et al., 2007) used previously. In the present study of SR, we report the presence of DT162 and DT348 on Taeniopteris sp. and N. auriculata, whose damage is classified as level 3, or “specialized”, or more properly narrow damage-type functional breadth. Hole feeding at SR was more representative of the CBB plant assemblage (24.09%), followed by SAP and MCF, which had lower numbers of damaged plants (12.19% and 12.07%). Sanzenbacher Ranch and remaining plant assemblages had values under 9%, indicating that of all the DT occurrences ( Table 2 ), SR had fewest hole feeding DT occurrences compared to margin feeding. Surface feeding had an elevated presence at WD, with 34.19% of all occurrences, contrasting with less hole feeding and margin feeding. The SAP, MCF, and CCP plant assemblages had <10% of all DT occurrences as hole feeding, and the same pattern was reported in this study at SR. Notably, SR, MCF, CCP, and SAP lacked evidence of skeletonization ( Figure 4 ).

Contrasting with external foliage feeding, galling for a few plant assemblages was the first or second functional feeding group most represented, based on DT occurrences. The MCF locality had galling that ranked as the first FFG, representing 29.31% of all DT occurrences ( Table 2 ), and also with the greatest richness. Our study showed that galling as the second group in occurrences and the first in DT diversity, accounting for nine out of 48 DTs (18.75%) ( Table 1 ). Schachat et al. (2015) indicated that the high diversity of galls could be indicative of aridification in the MCF environment, and the SR locality has drought-tolerant taxa, such as the conifer Walchia sp., the medullosan Neurodontopteris, and callipterid peltasperms (DiMichele et al., 2018a). Such a condition could be indicative of the expansion of aridification in Euramerica due to glacial–interglacial cycles. The Archer City Formation has a low percentage of carbonates that would have formed the original soil, which may be indicative of soil moisture deficiency and water stress (Griffis et al., 2023). A current example is Neotropical Dry Forest, where there is an association between gall-inducing insects and aridity, involving factors such as ontogenetic stage, plant density, and life forms present in the forest (Cuevas-Reyes et al., 2004). SR had an elevated number of galling DTs in A. conferta and R. schenkii, the most abundant taxa at the locality, which indicate the prevalence of galler arthropods attacking both plant hosts. Additionally, for CCP and SAP, both representing xeromorphic environments, galling was their second most abundant FFG. Concerning the damage at SR, DT34 was the most common gall observed. DT34 damage also was observed at SAP, CCP, and WD. The galls DT85 and DT121 were reported for the CBB plant assemblage (Labandeira and Allen, 2007) and at SR. We observed DT85 on the SR callipterid R. schenkii and lyginopterid S. germanicum, the latter also occurring at the CBB plant assemblage along A. conferta. DT121 was present in the conifer Walchia piniformis at the CBB plant assemblage and also occurs on the calamitalean A. equisetiformis at SR. At SR, a new type of gall on the woody axes of the voltzialean conifer Walchia sp. 1 is DT415 ( Supplementary Material ). The Taint plant assemblage lacked the galling functional feeding group, in contrast to other Paleozoic plant assemblages in Texas ( Figure 11 ), consistent with its occupation near a stream drainage system.

The piercing-and-sucking functional feeding group expresses the same degree of richness as that observed for surface feeding at SR. We observed that most pteridosperms, including Autunia, Rhachiphyllum, cf. Lodevia sp., Odontopteris, Neurodontopteris, and Sphenopteridium, collectively had 13 DTs, representing 65% of all piercing-and-sucking DTs at SR. Xu et al. (2018) reported the expansion of the piercing-and-sucking functional feeding group during the latest Pennsylvanian corresponds to 24.33% of all DT occurrences at WD. In addition, WD exhibited nine DTs, including distinctive DT286 and DT287 that are scale insect impression marks. At SR, we noticed that DT286 occurred only on pteridosperms, whereas at WD, there is a high incidence of DT286 on sphenophytes. The piercing-and-sucking data at SR is consistent with an expansion of a particular type of herbivory on a variety of taxa during the Pennsylvanian–Permian transition. DTs 46, 47, and 48 are the most common in the Paleozoic plant assemblages and are considered of intermediate damage-type functional breadth. These three DTs are present in several major plant lineages, including lycophytes, sphenophytes, marattialean ferns, gigantopterids, pteridosperms, and coniferaleans. Notably, the Taint and CBB plant assemblages lack this functional feeding group ( Figure 11 ).

The oviposition record is well documented for Paleozoic floras, even though the insertion of eggs into plant tissues is a type of resource use different from feeding. Paleozoic ovipositional damage from Angara, Cathaysia, Euramerica, and Gondwana have been broadly reported (Pinheiro et al., 2012; Vassilenko and Karasev, 2020; Donovan and Lucas, 2021; Ma et al., 2023), and plant assemblages such as CCP and WD are noted for their high level of oviposition, consisting of 33.74 and 32.17%, respectively, of all DT occurrences (Schachat et al., 2014; Xu et al., 2018). The SR plant assemblage has a proportional representation of ovipositional damage comparable to SAP, with somewhat <10% of all DT occurrences ( Table 2 ). Sanzenbacher shares seven oviposition DTs with the other Texan plant assemblages. DT101 is the most common in these Paleozoic floras and is present in almost all the sites, followed by DT76 and DT100. One DT that is exclusive to SR is DT294, which occurs in a Walchia sp. 1 stem.

Seed predation occurs in a few Paleozoic plant assemblages from the Early Pennsylvanian to middle Permian (Sharov, 1973; Jennings, 1974; Scott and Taylor, 1983; Shcherbakov, 2008; Labandeira et al., 2013; Schachat et al., 2014, 2015; Labandeira et al., 2016; Dos Santos et al., 2020). A few studies found a sizeable number of interactions in mostly platyspermic but also cylindrospermic seeds (Schachat et al., 2014, 2015; Dos Santos et al., 2020). DT73 and DT74 has been reported for the MCF and CCP plant assemblages that targeted two and three seed morphotypes, respectively. Additionally, the presence of DT124 was reported for the MCF plant assemblage (Schachat et al., 2015). For SR, predation was noticed on five seed morphotypes. Seed sp. 9 harbored four DTs, the most prominent of which is DT74 with two DT occurrences, seed sp. 1 had DT74 and DT124, and seed sp. 2 exhibited DT73 and DT74. Two other seeds, seed sp. 5 and seed sp. 6, displayed DT73. Although seed predation showed multiple DT occurrences at SR, such interactions have not been recorded at the WD, CBB, Taint, and SAP plant assemblages.

The Paleozoic record of borings is scarce, although there are rare exceptions (Naugolnykh and Ponomarenko, 2010; Feng et al., 2017). Borings have been observed previously at two plant assemblages from the late Paleozoic interval of north-central Texas. Two DTs of borings were documented on axes from the latest Pennsylvanian (Gzhelian) WD plant assemblage (Xu et al., 2018), in which DT284 was present on Calamites sp. 1 and DT160 was present on the conifer Walchia sp.1. We also recorded DT284 on Calamites sp. 1 at SR. Moreover, Schachat et al. (2014) observed that DT243 borings from the early Permian (Kungurian) CCP plant assemblage were present on an unaffiliated plant. This study expands the knowledge of the borings functional feeding group for the early Permian.

Pathogens were not assessed for all late Paleozoic plant assemblages in north-central Texas but were recorded for the WD, CCP, and SR plant assemblages. The oldest locality, WD, established the presence of DT58, and DT75 and DT97 that preferentially occurred on medullosan foliage (Xu et al., 2018). At CCP, DT58 and DT97 were present (Schachat et al., 2014). All three pathogen DTs were present at SR. Donovan and Lucas (2021) recorded DT75 and DT97 pathogen damage at an earlier Late Pennsylvanian assemblage at the Kinney Brick Quarry in New Mexico. It appears that DT75 and DT97 was a persistent pathogen present on seed fern, mostly medullosan, foliage, throughout the Middle Pennsylvanian (Labandeira and Beall, 1990) to earliest Permian.

Comparisons of DTs among these plant assemblages were analyzed using an ordination that was represented by non-metric multidimensional scaling (NMDS) ( Figures 11 , 12 ). The results show that, in comparisons of functional feeding groups, the SR plant assemblage is most closely associated with the SAP and MCF plant assemblages. Gigantopterids and callipterids from these three plant assemblages exhibited a strong association with the FFG external foliage feeding and with galling. The high prevalence of galling in these three plant assemblages and their association with habitats where moisture is limiting ( Table 3 ) supports the hypothesis proposed by Fernandes and Price (1988) that harsh environments are occupied by gallers. Subsequent studies by these and other authors (Fernandes and Price, 1992; Price et al., 1998) provide additional support from the modern record that gall richness increases in dry sites compared to riparian sites.

The only fossil insects known from the Sanzenbacher Ranch locality consists of five wings of cockroaches (Blattodea) in the NPL 1161 collection and one in the USGS 9999 collection, the latter probably representing species of Phyloblatta. In addition, a caterpillar-like holometabolous larva was found that currently remains undescribed.

Insects with mandibulate mouthparts are principally responsible for causing external foliage feeding in plants. Plant damage caused by the functional feeding groups of hole feeders (DT01, DT02, and DT08), margin feeders (DT12, DT14, DT15, DT162, and DT348), and surface feeders (DT25, DT27, DT29, and DT30) could have been inflicted by four major groups of Permian insects. They are (i) orthopteroid insects such as grasshopper-like Oedischiidae of the Orthoptera (Sharov, 1971); (ii) the elongate Protophasmatidae and spinose Geraridae of the “Eoblattida” (Rasnitsyn, 2002); (iii) caterpillar-like, larval holometabolans Metabolarva and Srokalarva (Nel et al., 2013; Haug et al., 2015); and (iv) possibly early beetle lineages such as Tshekardocoleidae and Permocupedidae (Ponomarenko, 1969). The first three of these groups possessed downwardly directed (hypognathous) mouthparts typically associated with herbivory, unlike the forwardly directed (prognathous) mouthparts of early Permian predatory insects (Labandeira, 2019). More direct evidence of a possible culprit is the three-dimensionally preserved caterpillar-like larva, a likely herbivore, found in the NMNH Sanzenbacher Ranch collection. The dietary spectrum of early beetle lineages is less clear; virtually, all Permian lineages have been associated with wood (Zhao et al., 2021), a conclusion buttressed by distinctive galleries present in late Permian conifer wood (Feng et al., 2017). However, although the associations of beetles with wood during the Permian often have been noted (Zhao et al., 2021), there is evidence that some beetle taxa consumed live and dead plants (Dmitriev and Ponomarenko, 2002).

The Pennsylvanian and Permian record of endophytic oviposition is extensive, and several insect taxa have been suggested as culprits. Based on external ovipositor structure and the size and shape of oviposition lesions on plant surfaces (Lin et al., 2019), several groups of early Permian ovipositing insects are implicated (Lin et al., 2019). Evidence indicates that endophytic oviposition was inflicted by insects of medium to large size whose females possessed large, piercing or slicing ovipositors (Bechly, 2001), often blade-like and with sawtooth edges. Damage of larger, lenticular lesions (DT72, DT76) are consistent with slicing ovipositors. The distinctive damage pattern of DT100 is similar to the gymnosperm leaf Pursongia and a related species described from the Permian of Russia and attributed to kennedyine dragonflies (Meganeurina: Kennedyidae), ecological analogs of certain extant damselflies (Odonata: Zygopteran) (Vassilenko, 2011; Aristov et al., 2013). The structure of DT108, representing coalesced, large, circular to polylobate lesions, likely represents a damage from the slicing ovipositor of a medium sized orthopteran insect such as the Oedischiidae (Sharov, 1971). The small, ovoidal lesions of DT228 and DT245 are best attributed to Archescytinidae, an early Permian hemipteran lineage that possessed small, piercing, ovipositors (Shcherbakov and Popov, 2002).

The piercing-and-sucking functional feeding group expanded during the Late Pennsylvanian (Xu et al., 2018). This expansion included newly emergent lineages such as Archescytinidae, Prosbolidae, Prosbolopseidae, Coleoscytidae, and Ingruidae, insects with stylate mouthparts and body lengths of generally 4–10 mm (Shcherbakov and Popov, 2002). These five lineages were capable of producing the single punctures of DT46, DT47, and DT48, or the clustered pattern of DT281. By contrast, the fabricators of the scale insect impression scars of DT211 and DT286 remain a mystery, as they do for similar impression scars for the three-million-year earlier, latest Pennsylvanian, WD plant assemblage (Xu et al., 2018). DT211 occurs on foliar midveins, is ellipsoidal, and shows substantial height; by contrast, DT286 occurs on a leaf blade, is circular, and flat. Both DTs are morphologically similar to various modern soft scales (Hemiptera: Coccidae), although the earliest occurrence of the more inclusive clade of scale insects, Coccomorpha, has an earliest body-fossil occurrence in the Late Triassic (Shcherbakov and Popov, 2002).

Galls on plants have been recorded since the Middle Devonian (Labandeira and Prevec, 2014), and additional gall DTs have been reported in several late Paleozoic plant assemblages (Labandeira, 2021), The nine gall DTs at Sanzenbacher Ranch is one of the most diverse for the Paleozoic and exhibits the general shift of galls being hosted on plant axes during the Pennsylvanian to being hosted on plant foliage during the Permian (Labandeira, 2021). Sanzenbacher galls can be divided into axial galls (DT121 and DT415), generalized, featureless galls (DT34, DT80, and DT125), midrib galls (DT85 and DT304), and galls with distinctive features such as ornamentation (DT197 and DT247). The gall with the greatest resolution as to a culprit is DT121 (Labandeira and Allen, 2007), a conifer gall attributable to a pine aphid (Hemiptera: Adelgidae) or precursor lineage similar to the extant Cypress Twig Gall Midge (Lynch, 2017). Most of the generalized, featureless galls—DT34, DT80, and the compound gall DT125—likely were made by small, galling hemipterans such as Protopsyllidiidae, or by thripids such as Lophioneuridae that had the requisite small bodies important for galling during the Permian (Schachat and Labandeira, 2015).

Seed predation damage has been reported in Gondwanan and Euramerican floras during the Pennsylvanian and Permian (Zherikhin, 2002; Labandeira, 2006b; Dos Santos et al., 2020). Suggested culprits for DT73 and DT74 are hemipterans such as Archescytinidae and Protopsyllidiidae, the thripid Lophioneuridae, and the permothemistid Diathemidae, a Paleodictyopteroidea lineage of small body size that evolved stylate mouthparts and penetrating ovipositors distinct from Hemipteroidea (Labandeira, 2006b; Schachat and Labandeira, 2015; also see Sharov, 1973). The culprit for DT124 possibly is an early beetle lineage (Coleoptera: Tshekardocoleidae), based on the damage that indicates a chewing insect with robust, prognathous, mandibulate mouthparts (Ponomarenko, 1969) primarily may have had habits of boring into wood (xylophagy) (Zhao et al., 2021).

Other than small oribatid mites, insects producing borings into early Permian wood of cordaites and conifers, larger borings almost always would involve larger archostematan beetles (Zherikhin, 2002; Zhao et al., 2021). DT160, consisting of linear tunnels approximately circular in cross section and packed with frass, and DT284 consisting of more irregular galleries with adjacent tunnels bordered by partitions, likely involved culprits such as Tshekardocoleidae and Permocupedidae (Zhao et al., 2021). The consistency of wood tunneled by beetles would range from punky and in the initial stages of decomposition (Naugolnykh and Ponomarenko, 2010), to highly indurated secondary xylem (Feng et al., 2017).

Pathogens are poorly represented in the Permian, and little modern evidence has been provided for their taxonomic identifications. Most evidence for pathogenic damage comes from glossopterid plant assemblages of the late Permian (Labandeira and Prevec, 2014). At SR, DT66 and DT174 are distinctive as they house fructifications. The circular pattern of DT66 is similar to the ascomycotan Tar Spot Disease (Rhytismatales: Rhytismataceae) (Conrad, 2009), although this pattern of three to four concentric rings of miniscule fructifications is not well developed. The distinctive ovoidal blotch of DT174 that contains a central ring of small fructifications is reminiscent of a basidiomycotan rust fungus (Uredinales: Phragmidiaceae) that occurs on a variety of modern angiosperm hosts (Coombs, 2018). The remaining pathogens—DT58, DT75, and DT97—have minimal identifying features and can be attributed to various unidentifiable leaf necroses such as leaf spot and leaf blight.

This study recorded 48 DTs for the SR plant assemblage. Compared to other late Paleozoic plant assemblages from north-central Texas, we noticed a high DT richness at SR. Such elevated richness also was observed for the CCP and WD plant assemblages (Schachat et al., 2014; Xu et al., 2018), where the presence of 52 and 47 DTs, respectively, was reported. Nonetheless, by comparison, a considerably lower number of DTs were recorded for the remaining plant assemblages, ranging between 8 and 22 DTs (Beck and Labandeira, 1998; Labandeira and Allen, 2007; Schachat et al., 2015; MacCracken and Labandeira, 2020). Late Pennsylvanian to early middle Permian plant assemblages presented higher levels of DT occurrences, as observed at WD with 892 and at CCP with 1,390 DT occurrences (Schachat et al., 2014; Xu et al., 2018). By contrast, SR had 205 DT occurrences, and CBB, of similar age, reported 137 DT occurrences. The SAP and MCF plant assemblages showed 41 and 58 DT occurrences, respectively. Studies conducted in modern forests show that environmental conditions and plant community structure could influence damage composition, referred to as herbivorous diversity (Cuevas-Reyes et al., 2004; Espírito-Santo et al., 2012). A study published by Leal et al. (2016) showed that, in modern dry tropical forests, the vegetation structure can largely define the herbivore composition in terms of richness and abundance, which is based on tree richness.

There are three discrete patterns defined by the herbivory index (HI) for the seven plant assemblages ( Table 3 ). First, plant assemblages with an HI of 0.2%–0.3% are at SR, CBB, and MCF and have a taeniopterid or a callipterid as the most herbivorized plant host. Second, plant assemblages with an HI of 0.8%–0.9% are WD and SAP and have a gigantopterid or medullosan as the most herbivorized plant host. Third, plant assemblages with an HI of 1.5%–2.6% are Taint and CCP and have a gigantopterid or a peltasperm as the most herbivorized plant host. In all instances, these data indicate that seed plants are the most herbivorized host, and overall environmental conditions, especially access to water sources, may be the prime determinant of the HI.

The seven plant assemblages from the latest Pennsylvanian to early middle Permian of north-central Texas exhibit similarities and differences regarding the most herbivorized taxa ( Table 3 ). For the early Permian plant assemblages of SR and CBB, a callipterid and A. conferta are the most herbivorized taxa. At CBB A. conferta accounted for 335 specimens, 85 of which express at least one instance of herbivory, representing 91.4% of the total herbivorized specimens of the plant assemblage (Labandeira and Allen, 2007). For SR, we observed 174 specimens, 42 of which show herbivory, representing 27.63% of the plant specimens. When comparing herbivory on Autunia, the values are similar between the two plant assemblages: CBB shows 25.37% of the specimens were herbivorized, and SR shows 24.13% of the specimens have been herbivorized. Labandeira and Allen (2007) documented at CBB the functional feeding groups of hole feeding, margin feeding, skeletonization, and galling for A. conferta, with an herbivorized area of 10.32 cm² compared to a total foliar area of 2420.34 cm². In contrast, at the SR plant assemblage, A. conferta exhibited the functional feeding groups of hole feeding, margin feeding, surface feeding, piercing and sucking, oviposition, and galling, with an herbivorized area of 6.11 cm² out of a total surface area of 5,212.77 cm². The difference in total herbivorized area between these plant assemblages could be attributed to a wide variation in the DT composition of damage. Whereas CBB showed a higher presence of functional feeding groups such as margin feeding and hole feeding, the SR plant assemblage had a small number of DTs attributable to piercing and sucking and oviposition ( Supplementary Table S4 ). Additionally, the second most herbivorized taxa in our study was the callipterid R. schenkii, representing 15.79% of the total herbivorized specimens. The functional feeding groups observed on R. schenkii were hole feeding, margin feeding, surface feeding, piercing and sucking, oviposition, galling, and pathogens ( Supplementary Table S4 ). Callipterids, such as Autunia, can have fronds up to 80 cm long (Kerp, 1988) and a thick cuticle in the adaxial portion that thins out considerably toward the abaxial region with the added presence of epidermal papillae (Kerp, 1988; Kerp and Barthel, 1993). The presence of thick cuticle and papillae could be protection from herbivory by arthropods and pathogens (Glover, 2000; Peeters, 2002).

Medullosans were extensively herbivorized at the latest Pennsylvanian WD plant assemblage. This plant assemblage featured Macroneuropteris scheuchzeri as the most herbivorized plant host, comprising 52.44% of all interactions (Xu et al., 2018). The authors reported for M. scheuchzeri a diversity of functional feeding groups that encompassed all four subgroups of external foliage feeding, and piercing and sucking, oviposition, galling, and pathogens. Pathogens were developed on M. scheuchzeri that included especially an elevated presence of DT75 and DT97. In our examination of SR, Neurodontopteris auriculata was the most herbivorized medullosan. As observed previously (Xu et al., 2018), DT75 and DT97 were also documented on this plant host ( Supplementary Table S6 ). Whereas we assigned these DTs to pathogen damage based on the latest version of the Damage Guide (Labandeira et al., in prep), Xu et al. (2018) mentioned in their study that both DTs were probably caused by a combination of factors and were the result of an opportunistic pathogen.

The MCF and CCP plant assemblages show that the abundantly represented Taeniopteris spp. as the most herbivorized group of plants (Schachat et al., 2014, 2015). SR had five taeniopterid specimens, two of which showed at least one instance of herbivory. We observed one specimen with DT02 hole feeding and the other specimen with DT162 margin feeding. DT02 was recorded at the CPP and MCF plant assemblages, and DT162 is the first instance of this DT reported for a Paleozoic assemblage.

The late Artinskian and the early Guadalupian plant assemblages of Taint and SAP present gigantopterids as the most herbivorized taxa ( Figure 1 , Table 3 ). The Taint plant assemblage shows that Zeilleropteris wattii (Beck and Labandeira, 1998) and the SAP plant assemblage shows that Euparyphoselis gibsonii (MacCracken and Labandeira, 2020) were extensively herbivorized. In contrast, the SR plant assemblage lacked gigantopterids, although a similar level of overall herbivory was observed, similar to the latest Pennsylvanian WD and the late Sakmarian CBB plant assemblages (Labandeira and Allen, 2007; Xu et al., 2018) ( Figure 1 ). Apparently, gigantopterids were not abundant during the early Permian in Euramerica but became more abundant in its southwestern region during the uppermost Sakmarian to the earliest Guadalupian (Mamay, 1988; Mamay et al., 1988; DiMichele et al., 2005; DiMichele et al., 2018b).

Herbivore preference for targeting and preferring specific taxa could be associated with the presence of particular plant groups in certain environments. Callipterids and medullosans were abundant at the end of the Pennsylvanian and beginning of the Permian, as observed in this study and in other plant assemblages (Labandeira and Allen, 2007; DiMichele et al., 2013, 2015; Xu et al., 2018; DiMichele et al., 2018a; Luthardt et al., 2021). From the Sakmarian Stage onwards, there is a noticeable shift in both the plant assemblages and in the environments they inhabited ( Table 3 ). This transition shifted from humid to seasonally dry environments (DiMichele et al., 2004; MacCracken and Labandeira, 2020), a shift that increased the diversity and abundance of gigantopterids and accordingly their arthropod herbivores.