The material originates from the Talbragar Fossil Fish Beds, situated north of Gulgong in central New South Wales, Australia. A recent analysis of zircon crystals from this strata dates it to the very late Oxfordian–Tithonian (Late Jurassic), ca 157.3–145.0 Ma (Turner et al. 2009). The deposit represents a shallow-water-lake palaeoenvironment at the northern end of the deposit, grading to a shoreline palaeoenvironment towards the southern end of the deposit. It was part of a productive lake supporting a large population of fish and a diverse aquatic/shoreline and terrestrial woodland fringe palaeoecosystem, which has produced a significant number and variety of insect fossils (Beattie and Avery 2012). Several new insect taxa have recently been described, comprising members of the orders Odonata (Beattie and Nel 2012), Hemiptera (Chen et al. 2019; Li et al. 2022), Coleoptera (Fikáček et al. 2014; Ashman et al. 2015; Oberprieler et al. 2016), Hymenoptera (Oberprieler et al. 2012), and Diptera (Oberprieler and Yeates 2012, 2014; Oberprieler et al. 2015).

The material was examined dry and under a film of ethyl alcohol using Olympus SZX7 and Leica M205 C stereomicroscopes. The photographs were taken using a Canon EOS 550D and a Canon EOS 1200D digital camera equipped with MP-E 65 mm and EF-S 60 mm macro-lenses or attached to the Leica M205 C stereomicroscope. Original photographs were processed by stacking and editing using Adobe Photoshop™ version CS6 (Adobe Systems Incorporated, San Jose, U.S.A.). The measurements of individual body parts were inferred from the photographs taken with a calibration scale.

Abbreviations for wing veins used throughout the text follow Béthoux (2005): C, costa; ScP, subcosta posterior; R, radius; RA, radius anterior; RP, radius posterior; M, media; MA, media anterior; MP, media posterior; Cu, cubitus; CuA, cubitus anterior; CuP, cubitus posterior.

All the material is deposited in the Australian Museum, Sydney, Australia under the accession numbers as specified below.

Although the wing venation is not completely preserved in the specimens of Talbragaria australisgen. et sp. nov., the anterior part of the wings is unambiguously discernible (Fig. 1A, B, F–H). It features the so-called nemourid X, consisting of ScP that fuses with RA and again diverges from it more distally (Cui et al. 2018), the basal part of RP bifurcation and a crossvein between RP and M. According to Zwick (2000), this pattern of veins is generally present in the families Nemouridae and Notonemouridae (Arctoperlaria, Euholognatha). Although the nemourid X indicates a close relationship between T. australissp. nov. and the Nemouridae + Notonemouridae clade, a similar pattern of veins can also be observed in the likewise euholognathan families Taeniopterygidae and Capniidae (Zwick 2000). However, the possible attribution of T. australissp. nov. to these families can be excluded using other characters. Capniidae are characterised by an apomorphic reduction of m-cua crossveins (at most two) and cua-cup crossveins (at most one) (Zwick 2000). In our fossils, the number of respective crossveins is obviously higher, despite the fragmentary preservation, but there are no crossveins in the apical portion of the wings (Fig. 1F). Taeniopterygidae exhibit an elongated second tarsomere, which is at least 4/5 of the length of the third tarsomere and represents an apomorphy of this family (Nelson 2009). The second tarsomere in T. australissp. nov. is shorter, subequal to 2/3 of the length of the third tarsomere (Fig. 1C). Since the first tarsomere is missing in available specimens of T. australissp. nov., there is a possibility that the second tarsomere is broken basally and incomplete. However, since the first tarsomere is not preserved on the counterpart, we assume that the second and third tarsomeres are the only ones present in the fossil and the fragmentation of the tarsus happened not during the splitting of the matrix, but early in the fossilization process. Thus, a disarticulation between tarsomeres is more probable than a crack within a basal part of tarsomere. Consequently, we consider the second tarsomere to be likely complete. Therefore, all available morphological evidence testifies for the placement of T. australissp. nov. into either Notonemouridae or Nemouridae. Zwick (2000) questioned the monophyly of notonemourids based on the presence of an ovipositor in some genera only. Based on morphology, he considered Notonemouridae as possibly representing a paraphyletic assembly (grade) of early nemourid lines surviving on fragmented Gondwanaland. However, the families Notonemouridae and Nemouridae were recovered as sister groups in the molecular phylogeny of Ding et al. (2019) and the monophyly of Notonemouridae is strongly supported by McCulloch et al. (2016). One more purely fossil family with a close relationship to the Nemouridae + Notonemouridae clade is represented by Pronemouridae, hypothesised by Liu et al. (2011) to represent the stem group of this clade.

The only one of these families known from the Southern Hemisphere is Notonemouridae (Zwick 2000); therefore its presence in the Upper Jurassic of Australia is more probable than in the case of Nemouridae or Pronemouridae. Cui et al. (2018) introduced a new apomorphic wing venation character to define Notonemouridae, namely CuP in the forewing closely approaching the posterior wing margin (just distal of the end of AA1), then diverging from it, until it reaches it more distally. This character is only partially observable in our fossils, since the posterior portion of wings is not completely preserved. In the specimen F.136 857, a short visible portion of CuP is distally converging with CuA (Fig. 1G, H). It indicates the course of CuP as characteristic for Notonemouridae. Nevertheless, other evidence for the attribution of T. australissp. nov. to Notonemouridae is represented by the structure of the female genitalia that very closely resembles some Recent notonemourids. It exhibits a pronounced corrugated subgenital plate posteriorly on sternite VIII enabling interlocking with the male epiproct or other copulatory structures, and posteriorly narrowed subanal plates on sternite IX (compare Figs 1D, E, 2D, E and fig. 3.5 in Stevens 2008). Furthermore, T. australissp. nov. differs from the families Pronemouridae and Nemouridae in the structure of the tarsi. The second tarsomere in both above-mentioned families is distinctly short, reaching to approximately one-half of the third tarsomere (Nelson 2009; Liu et al. 2011). The length of the second tarsomere in T. australissp. nov. is slightly longer, subequal to two thirds of the length of the third tarsomere, which represents a value reported for several representatives of Recent Notonemouridae (Nelson 2009). Pronemouridae are further characterised by two plesiomorphic characters, CuA ending with two branches and multisegmented cerci (Liu et al. 2011). Cerci in T. australissp. nov. are one-segmented (Fig. 1D, E), an apomorphic character shared with Notonemouridae and Nemouridae (Cui et al. 2018). The course of CuA in T. australissp. nov. is not observable distally of the M-CuA crossvein, however, no bifurcation is visible in the preserved portion of CuA (Fig. 1F). Also taking into account the biogeographic pattern of the discussed individual stonefly taxa, we consider the attribution of T. australissp. nov. to Notonemouridae as the most probable.

Within Notonemouridae, Cui et al. (2018) established a subfamily Paranotonemourinae to accommodate two fossil species of the genus Paranotonemoura Cui and Béthoux, 2018. The species Talbragaria australissp. nov. is not attributable to Paranotonemourinae since the mp-cua crossvein is not more than twice as long as the longest crossvein in the area between CuA and CuP (Fig. 1F), which does not comply with the diagnosis of Paranotonemourinae (Cui et al. 2018). Therefore, we attribute T. australissp. nov. to the subfamily Notonemourinae within Notonemouridae.

There are 23 extant genera of Notonemouridae (DeWalt et al. 2023). It is difficult and in some cases impossible to compare T. australissp. nov. with these genera, since some are defined based on characters not visible in T. australissp. nov. (notably all characteristics of male genitalia). In any case, given the estimated ages of Plecoptera clades (Cui et al. 2016, 2018), it is not reasonable to expect a Recent genus being already present in the Jurassic. We would like to refrain from implying closer relationship to any particular Recent genus, which is not really grounded because of the lack of reliable characters.

Our finding from the Southern Hemisphere complements the journey of Notonemouridae by confirming their presence in the territory of Gondwana already in the Late Jurassic, since Talbragar Fish Beds are dated as Oxfordian–Tithonian (ca 157.3–145.0 Ma) and Notonemouridae must have already been established in the area of current Australia at that time. Their dispersal to the Southern Hemisphere thus predates the rifting of Pangaea, exactly as presumed by Liu et al. (2011) and Cui et al. (2018). Therefore, the hypothesis of McCulloch et al. (2016) and Ding et al. (2019) about the more recent colonisation of Gondwana by Notonemouridae by long-distance dispersal postdating the Gondwana break-up can be rejected. Recent representatives of this family occur only in the fragments of Gondwana, suggesting that the southward dispersal was followed by the extinction of Notonemouridae in the Northern Hemisphere. This extinction possibly occurred in the mid-Cretaceous, when a massive increase of the extinction rate was reported for Plecoptera (Jouault et al. 2022).

The fact that Notonemouridae were present in Australia in the Late Jurassic and at about the same time also occurred in China and Mongolia (Cui et al. 2018) further indicates that the lineage must have been very widespread, considering that the dispersal route by land between these sites spanned several of today’s continents (Fig. 3). Importantly, it also provides evidence that the lineages within stonefly suborders up to the level of some families were already diversified in the Late Jurassic. Given the fact that Notonemouridae is not a basal lineage in the Plecoptera phylogeny (Zwick 2000; Ding et al. 2019), this indicates that the split between the suborders Antarctoperlaria and Arctoperlaria occurred much earlier than the Late Jurassic. Thus, some published age assessments for this split are certainly underestimated. McCulloch et al. (2016) dated the split between suborders as Early Cretaceous (121 Ma, with a confidence interval 143–109 Ma) and Ding et al. (2019) estimated the split of Antarctoperlaria vs Arctoperlaria as Early Jurassic (181.45 Ma, although with a wide confidence interval spanning most of Mesozoic, 106–250 Ma). Probably a more realistic estimate for the major divergences within Plecoptera were given by Cui et al. (2016, 2018) based only on the study of the fossil species. They suggested that the Antarctoperlaria vs Arctoperlaria split occurred in the early Mesozoic or the late Palaeozoic (approximately 260–230 Ma). Interestingly, this timing would rule out the possibility that the split between stonefly suborders was a vicariant event resulting from the continental break-up between Gondwana and Laurasia 215 to 175 Ma, as traditionally hypothesised (Zwick 2000; Ding et al. 2019). Since some Recent families were already diversified in the Late Jurassic, the split must have occurred much earlier, predating the break-up.