Introduction

Anuran amphibians are characterised by a biphasic ontogeny, consisting of radically different larval (“tadpole”) and adult (“frog”) morphs. Although the fossil record for tadpoles is more limited by taphonomical reasons (e.g. Maus and Wuttke 2002; McNamara et al. 2009; Gardner 2016), compared to the record for postmetamorphic frogs, the former is more extensive and informative in terms of ecological and evolutionary developments than generally assumed (Roček 2003; Roček et al. 2006; Gardner 2016; Bardua et al. 2021). Today, tadpoles may reach high abundances (Gibbons et al. 2006; Pinero-Rodríguez et al. 2020) and have an important ecological role in lakes, contributing to nutrient cycling, altering the trophic web, and even causing bioturbation within nutritious sediments (Costa and Vonesh 2013, Cortés-Gómez et al. 2015; Kloh et al. 2019; Pinero-Rodríguez et al. 2020). Tadpoles are often primary consumers or even omnivorous, with a diet mainly consisting of bacteria, algae, crustaceans, insects, and periphyton. They also are zoophagic (Escoriza et al. 2016), prey on anuran embryos (Tejedo 1991), even actively hunt larval insects (Montaña et al. 2019), or even cannibal conspecifics (Jefferson et al. 2014; Comas and Escoriza 2015). In general, tadpoles can be divided into three ecomorphotypes adapted to feed on the surface (neustonic), the water column (nektonic), and the bottom (benthic). Tadpoles show considerable behavioural plasticity. They can shift their diet in response to the availability of resources or as a result of competition and predation risk (Altig et al. 2007; Richter-Boix et al. 2007; Jefferson et al. 2014; Arribas et al. 2018; Montaña et al. 2019).

One of the few fossil sites worldwide with a rich record of tadpoles of different ontogenetic stages is the upper Oligocene Enspel fossil site in the Westerwald Mountains (Germany; Maus and Wuttke 2002, 2004; Roček and Wuttke 2010). The fossiliferous oil shale of the Fossillagerstätte Enspel was deposited in a deep crater lake (Schindler and Wuttke 2015) and yielded an exceptionally preserved biota (see below). The interval of deposition of the fossiliferous black pelites is well-constrained by both biostratigraphical data and radioisotopic dating of basalts below and above the fossil-bearing sediments. Accordingly, the age is constrained to mammal Paleogene zone MP28 (Storch et al. 1996) and an absolute age of 24.79–24.56 million years (Mertz et al. 2007). The Fossillagerstätte Enspel has become an exemplar of the palaeoenvironment during the late Oligocene in Central Europe since the locality was rediscovered approximately 30 years ago (Müller 1997) after the first find of fossils in the late 19th century. The fossil-bearing sediments have yielded a plethora of different plant and animal remains (e.g. Poschmann et al. 2010; Wuttke et al. 2015). These include benthic cyanobacteria, dinoflagellates, other algae, and thecamoebians (Köhler and Clausing 2000; Herrmann 2010; Schiller and Wuttke 2015), spores and pollen from terrestrial plants (Herrmann et al. 2009, 2010), as well as leaves, fruits, seeds, cones, and flowers from various gymnosperms and angiosperms, and remains of bryophytes and ferns (Köhler and Uhl 2014; Uhl 2015; Uhl and Poschmann 2018). Fossil animals are abundant at this site, including both invertebrate and vertebrate remains. Insects are incredibly diverse (incl. beetles, flies, midges, ants, bees, wasps, dragonflies, caddisflies, etc.) and are among the most commonly found fossils at Enspel (Wedmann 2000; Wedmann et al. 2010; Fanti and Poschmann 2019; Brockhaus et al. 2020; Legalov and Poschmann 2020; Nel et al. 2020; Skartveit and Wedmann 2021; Poschmann and Nel, 2022; Vršanský et al. 2022). The vertebrate diversity is relatively high, with records of fish (Böhme 1996), amphibians (Maus and Wuttke 2002, 2004; Roček and Wuttke 2010; Schoch et al. 2015), reptiles (Frey and Monninger 2010; Karl and Wettlaufer 2011), birds (Mayr 2007, 2015; Mayr et al. 2006; Mayr and Poschmann 2009), and especially mammals (e.g. Storch et al. 1996; Engesser and Storch 1999; Mörs and Königswald 2000; Mörs and Kalthoff 2010; Schwermann and Martin 2012; Engler and Martin 2015; Mähler et al. 2015; Smith and Wuttke 2015; Schwermann 2023). The most abundant fossil vertebrates at Enspel are tadpoles belonging to the family Pelobatidae, assigned to Pelobates cf. decheni (Maus and Wuttke 2002, 2004; Roček and Wuttke 2010). Those tadpoles number in the thousands, but only a few specimens show details of soft parts preserved, such as jaw sheaths, notochord, “skin shadow”, and eyespots (these soft tissue remnants consist of sulfurized melanosomes; McNamara et al. 2009). Some specimens are preserved with their gut contents still recognisable. Depending on finds from different layers within the fossiliferous sequence (see sediment types in Felder et al. 1998; Schindler and Wuttke 2015, fig. 3; Uhl and Poschmann 2018, fig. 2), the coiled intestines are predominantly filled with diatoms or tuffaceous silt to sand. Former intestine content is rather commonly found as isolated, short cord segments, which sometimes densely cover entire layer surfaces. Sometimes “sand pebbles” (isolated spherical sandy nodules, < 2cm) are found in the sediment, probably due to taphonomic processes. The identification of “sand pebbles” as tadpole cololites (e.g. Seilacher et al. 2001) may be debatable. However, both their similarity in form and size to identifiable intracorporal confluent tadpole gut contents and recognition of transitional stages of decaying specimens with “skin” residues and only a few bones associated with “sand pebbles” strongly suggest their tadpole origin (Wuttke et al. in prep.).

Fossilised gut/stomach contents are rare (Hunt and Lucas 2021), especially in the terrestrial vertebrate record. Even rarer are vertebrate fossils containing in situ pollen preserved in their digestive system. Among others, these include a single bird from the Eocene of Messel, Germany (Mayr and Wilde 2014), a tapir from the Miocene of the Gray Fossil Site, Tennessee, USA (McConnell and Zavada 2013), and a few Pleistocene mammoth, horse, and woolly rhinoceros fossils from the permafrost of Russia (e.g. Heintz 1958; van Geel et al. 2008; Boeskorov et al. 2011a, b; Kosintsev et al. 2012). Of these, only the bird from Messel is believed to have actively sought out and extracted pollen from flowers (Mayr and Wilde 2014). Until now, there was no fossil record of pollen feeding among tadpoles, but research on living hylid tadpoles has shown pollen to be part of their diet (Wagner 1986). The only documented active pollen feeding by tadpoles from water surfaces in the wild indicates that tadpoles of different species of the family Hylidae actively feed on pollen. Sabagh et al. (2012) reported pollen-feeding in hylid tadpoles inhabiting the phytothelms of Alcantarea glaziouana (Bromeliaceae) but only accessorily. Britson and Kissell (1996) reported that in Tennessee, the breeding season of the upland chorus frog (Pseudacris triseriata feriarum, family Hylidae) coincides with the pollination of shortleaf pine (Pinus echinata) and southern red oak (Quercus falcata). An experiment by these authors on this species that compared pine pollen and algae feeding versus a diet of fish food found that feeding exclusively, either purposely or accidentally, on pollen was detrimental to normal development and all tadpoles stayed smaller. However, Kloh et al. (2021b) found that the majority of the diet of the stream-dwelling and surface-feeding Phasmahyla jandaia (Hylidae) tadpoles was composed of pollen. All developmental stages showed a positive, high selectivity for this food item, meaning a remarkable preference for pollen. It seems that Hyla regilla also can alter their submerged feeding behaviour in the presence of dense pollen scums and that they can discriminate between fresh and leached pollen, suggesting an advanced interaction with pollen as a food source (Wagner 1986). Kloh et al. (2018, 2019, 2021a) reported on several tropical species at the montane meadows of southeastern Brazil feeding on pollen. Additionally, Scinax machadoi (Hylidae) tadpoles were reported to vertically swim through the water column to actively feed on pollen even when presented with other more easily accessible food items (Kloh et al. 2021b). Also, Kloh et al. (2024) showed that tadpoles of Scinax machadoi (Hylidae), in initial developmental stages 25 and 30, grew more when fed with mixed pollen, compared to artificial food and a mixture of both diets. Kloh et al. (2023) concluded that neustonic tadpoles generally consumed more pollen and broke them down (mechanically, osmotic shock, enzymatically, etc.) faster in their gut, compared to other ecomorphotypes.

Amphibian larvae make up a large proportion of the biomass in near-shore zones of ponds and lakes and play a significant role in the nutrient cycle and the associated energy flows (Montaña et al. 2019). Interactions with predators (see above) and competitors influence not only their abundance but also their feeding behavior and niche structure (Pavignano 1990; Caut et al. 2013). Despite the large number of tadpoles found in Lake Enspel, there is only one specimen with a predominant filling of the intestinal tract with pollen. Therefore, more detailed conclusions about the above-mentioned plastic changes in Enspel pelobatid larval trophic ecology cannot be drawn until more analyses of the food spectrum from the tadpoles with diatom or sand grain fillings are available. Tadpoles in general are mostly regarded as opportunistic omnivores (e.g. Hoff et al. 1999; Altig et al. 2007), as classic studies, based on stomach gut contents, have displayed. The tadpoles proved to be microphagous, suspension feeders, herbivores, and detritivores or are more carnivorous than previously thought (Montaña et al. 2019). This also applies to the larvae of the pelobatid family, about which, as in most other species, only incomplete data is available. Therefore, Montaña et al. (2019) called “for increased research on the subject to better quantify their ecological roles as consumers”. Stomach content analyses of the recent Pelobates fuscus, probably the closest living relative of P. cf. decheni, show that it feeds on cyanobacteria, green algae, and phanerogams (Pavignano 1990; Dgebuadze et al. 2017). Another relative, P. cultripes feeds on detritus, algae, and macrophytes (Busack and Zug, 1976; Caut et al. 2013; Montaña et al. 2019; Pinero-Rodríguez et al. 2021), and as first evidence, provided by Escoriza et al. (2016) for P. varialdii, and P. cultripes, the consumption of aquatic species of “Insecta (Coleoptera, Diptera, Ephemeroptera, and Odonata), Collembola (Sminthuridae and Poduridae), large Branchiopoda (Anostraca, Notostraca, and Spinicaudata), and Gastropoda (Physidae and Planorbidae)”. The authors hypothesised that grazing–rasping tadpoles, like the above mentioned species, have an omnivorous role in aquatic trophic webs. This hypothesis is supported by the observation of cannibalism by Comas and Escoriza (2015). Only Busack and Zug (1976) reported the intake of pollen (and spores) in P. cultripes; this was probably an accessory intake (in this case), given the small amount within the gut content. The ability of Pelobates tadpoles to surface feeding is only described by Belkin and Gans (1968) to our best knowledge but is well-known by field herpetologists. Gary Nafis (CaliforniaHerps Project, https://californiaherps.com/info/contact.html) published corresponding movies on the World Wide Web, displaying surface feeding of Spea sp.: http://www.californiaherps.com/movies/shammondiitads310.mov; and cf. Spea bombifrons http://www.californiaherps.com/movies/shammondiitads411.mov. Similar observations of European Pelobates fuscus tadpoles were reported to us in writing by Ana Ćurić (Zagreb) and Thomas Bobbe (Griesheim).

To provide an evolutionary perspective here of pollen feeding in pelobatids, we report on the intestinal contents of a giant fossil tadpole from the Oligocene of Enspel, Germany, which consists entirely of pollen. The tadpole, its gut/stomach contents, the pollen grains, are all described. The discovered pollen types and the parent plants they represent are compared to previous plant fossil records (dispersed pollen, leaves, etc.) from the same site. Conclusions are drawn about the assumed season of pollen dispersal and pollen ingestion of the tadpole during the late Oligocene at Enspel. These assumptions are based on the fossil pollen taxa and the pollination period of potential modern analogues. Finally, hypotheses about specialised surface feeding in pelobatid tadpoles are presented.

Geographical and geological setting

The Enspel Fossillagerstätte is positioned beneath the former Stöffel hill southeast of the village of Enspel in the Westerwald Mountains, Germany (UTM 32U E: 421561.486; N: 5607702.328). It is part of the Paleogene volcanic field of the High Westerwald (Fig. 1a–c; Schindler and Wuttke 2010, 2015; Schäfer et al. 2011). The bedrock of the Stöffel and its surroundings comprises Lower Devonian sedimentary rocks (Meyer and Stets 1980). The Devonian bedrock is overlain by various Cenozoic sedimentary rocks, both siliciclastic and volcaniclastic, as well as basalts (Felder et al. 1998). The Enspel sediments accumulated within a volcanic crater/maar lake (Schindler and Wuttke 2010, 2015).

Fig. 1
figure 1

Geographic and geological maps of the study area. a Geographic position of Enspel in central west Germany. b Paleogene volcanic field of the High Westerwald. c The Enspel Fossillagerstätte SE of the village of Enspel. d Stratigraphic column of the sediment fill in the Enspel crater based on the 1996 Enspel drilling. Modified after Schindler and Wuttke (2010, 2015)

A closed lake basin developed with an original depth of 240 m, and a diameter of 1.3– 1.7 km (Pirrung 1998; Pirrung et al. 2001), which existed for up to 200–220 Kyr (Mertz et al. 2007; Herrmann 2010). Afterwards, a vast basaltic inflow from the south stopped clastic sedimentation, filling the remaining lake basin by more than 140 meters. The sedimentary infill of the crater lake, mainly debrites and turbidites and minor bituminous and diatomitic black pelites (“oil shale”), is defined as the Enspel Formation (Schäfer et al. 2011). This kind of sedimentation lasted throughout its entire existence (Schindler and Wuttke 2015; Wuttke et al. 2015). The crater fill has been divided into three major units. The c. 114 m thick informal basal unit (Fig. 1d) is composed of mixed pyroclastics (referred to as Lithozone A comprising syneruptive basement breccias and Lithozone B comprising syneruptive breccias and tephras by Schindler and Wuttke 2015). These are succeeded by the c. 140 m thick Enspel Formation (Fig. 1d), which comprises organic-rich pelites interspersed with tuffs/tuffites (referred to as Lithozone C/D comprising debrites with intercalated black pelites by Schindler and Wuttke (2015)). The sediments are then topped by a vast basaltic inflow into the lake basin (Fig. 1; Schindler and Wuttke 2010, 2015; Schäfer et al. 2011). Radiometric dating of volcanics below and above the lake sediments suggest an age between 24.79±0.05 to 24.56±0.04 Ma for the fossiliferous Enspel Formation (Mertz et al. 2007). Fossil collection has been restricted to the top layers of the Enspel Formation (Fig. 1d) that are positioned a few decimetres below the basalt due to alteration and down to a depth of c 1.5 m due to limited access at outcrops (Gunkel and Wappler 2015; Felder et al. 1998). As the fossil tadpole reported herein, was found close to the upper basalt flow, its age is close to 24.56 ± 0.04 Ma years. For more detailed information on the geology, stratigraphy, and sedimentology of the Enspel Fossillagerstätte please consult Storch et al. (1996), Pirrung (1998), Felder et al. (1998), Engesser and Storch (1999), Pirrung et al. (2001), Hilder (2003), Mertz et al. (2007), Schindler and Wuttke (2010, 2015), and Schäfer et al. (2011).

Material and methods

Material under study

The previously undescribed fossil tadpole specimen, NHMMZ PW 2011/5509-LS, was collected in 2010 during the excavation of the oil shales at Enspel (Fig. 1), Germany, by the then Landesamt für Denkmalpflege Rheinland-Pfalz / Referat Erdgeschichte, now GDKE RLP, Direktion Landesarchäologie/Erdgeschichtliche Denkmalpflege. The specimen was recovered at excavation site G25 from oil shale layer S16 (Fig. 1d; sensu Felder et al. 1998). The fossil is housed in the collections of the Naturhistorisches Museum Mainz/Landessammlung für Naturkunde Rheinland-Pfalz, Mainz (NHMMZ, Germany) and can be accessed by contacting the collection manager.

Preparation of tadpole specimen

Both part and counterpart of the tadpole fossil were prepared under a stereo microscope using fine needles and are now stored in plastic boxes filled with glycerine.

Preparation of palynological material

The fossil tadpole was examined and sampled according to the method described in Geier et al. (2023). A stereomicroscope equipped with epifluorescence illumination was used to examine and photograph the fossil. The specimen was photographed with both white- and fluorescence light. Pollen grains were extracted from the gut of the fossil tadpole and processed for combined light microscopy (LM) and scanning electron microscopy (SEM) analysis by bleaching and acetolysis following Geier et al. (2023) and Halbritter et al. (2018), respectively. The pollen grains were investigated using the “Single-grain method” by Zetter (1989).

Climate data harvesting and analyses

Occurrence data of modern Picea (https://doi.org/10.15468/dl.xv4kzr), Pinus (https://doi.org/10.15468/dl.xyn3u6), Fagus (https://doi.org/10.15468/dl.gpg6na), and Ulmus (https://doi.org/10.15468/dl.6xgkzz) species were obtained from the Global Biodiversity Information Facility (GBIF, 2023, https://www.gbif.org). This raw data set was then modified (multiple occurrences with the same location as well as non-native occurrences were removed, Supplementary Material S1). Native distribution of species was defined by using chorological literature (Ohwi 1965; Browics and Zieliński 1982; Nixon 1997; Thompson et al. 1999a, b, 2001, 2006; eFloras 2008; Fang et al. 2009; Caudullo et al. 2017), the native distribution data from ‘Plants of the World Online’ (POWO 2023), and the ‘USDA PLANTS Database’ (USDA 2023). This modified data set is the basis for extant distribution maps calculated in QGIS 3.32.3-Lima (QGIS Development Team 2023), which were finalized in Adobe Illustrator 28.0. Potential modern analogue climatic niches were produced by plotting the modified datasets onto 1 km2 grid Köppen-Geiger maps (1979–2013 data; Cui et al. 2021) in QGIS 3.32.3-Lima (QGIS Development Team 2023). To extract the Köppen-Geiger climate types for each occurrence, the datasets were processed using the ‘Sample Raster Values’ Toolbox in QGIS. The Köppen climate types for each individual species’ georeferenced occurrences were then calculated (number of occurrences in each Köppen climate type, divided by all occurrences), resulting in their Köppen profile (Denk et al. 2023a, b). A genus‘ (Fagus, Picea, Ulmus) or subgenus’ (Pinus subgen. Pinus) Köppen profile is the summed Köppen profiles of all available species belonging to this genus or subgenus and divided by their number.

Results

Systematic palaeontology

The fossil tadpole from Enspel is described first followed by the different pollen types discovered in its gut. The pollen taxa are arranged according to Christenhusz et al. (2011, gymnosperm pollen) and APG IV (2016, angiosperm pollen). The pollen terminology follows Punt et al. (2007, LM) and Halbritter et al. (2018, LM and SEM).

Kingdom Animalia Haeckel, 1866

Tetrapoda Jaekel, 1909

Class Lissamphibia Haeckel, 1866

Order Anura Fischer von Waldheim, 1813

Family Pelobatidae Bonaparte, 1850

Genus Pelobates Wagler, 1830

Pelobates cf. decheni Troschel, 1861

(Fig. 2)

Fig. 2
figure 2

The fossil giant tadpole and its intestine content. a Photograph showing the more complete part of the specimen. b Schematic drawing of the tadpole pinpointing most diagnostic and noteworthy features preserved. c Close up of the anterior region showing skull bones, axial skeleton, and part of intestine. d, e Close-up of intestine/gut filled with yellowish pollen mass in white light (d) and under fluorescence (e), pollen strongly illuminates green. f Close-up of the intestine/gut showing densely packed pollen grains. Scale bars 10 mm (a–e), 1 mm (f)

Sampled specimen: NHMMZ PW 2011/5509-LS a, b (part and counterpart).

Description: Pelobatid tadpoles from Enspel were thoroughly described by Maus and Wuttke (2002, 2004). Thus, we restrict our description to briefly characterise the present fossil. The fossil represents a large premetamorphic tadpole with a body length of more than 12.5 cm, judging from the incompletely preserved skin shadow. Limbs had not developed yet, but the frontoparietals, parasphenoid, and otic capsules are already ossified. The total ossified part of the skeleton is about 40 mm long. The frontoparietal complex is clearly divided into two halves or pars frontalis; a posteriorly situated median element, the pars medialis, is either not ossified or not preserved. The parasphenoid shows a reversed T-shape, and on the medial process, the presence of a ventral keel is indicated (although seen in dorsal aspect in the part). Parts of the otic capsules (exoccipitalia and prootica) are present but poorly preserved. The axial skeleton comprises well-ossified presacral vertebrae I–VIII with the lateral processes of the anterior three vertebrae clearly visible. Sacral and caudal vertebrae IX and X can be recognised but are less well ossified. Furthermore, the fossil shows soft-part preservation in the form of a probable eyespot and an incompletely preserved skin shadow (Fig. 2a–c). The intestine with pollen content is clearly defined within a sac-like structure and displaced to the right of the skeleton in dorsal view.

Remarks: The configuration and morphology of skull bones described above, including the ventrally keeled parasphenoid, are diagnostic for Pelobatidae, and furthermore only found in Pelobates decheni and Eopelobates (Roček and Wuttke 2010). Assignment to Eopelobates is rendered unlikely by the complete lack of adults of this genus at Enspel (Roček and Wuttke 2010). According to the stage of skeletal ossification, the fossil tadpole is assignable to developmental stage 38/39 after Gosner (1960) (see also Maus and Wuttke 2004). Tadpoles of Gosner stage 39 are the most common developmental stage encountered at Enspel. Later stages are rare probably because the tadpoles then show an increasing adaptation to air-breathing and moved from open water to shore vegetation, thereby lowering their fossilisation potential (Maus and Wuttke 2004).

All skeletal elements of the tadpole are still in their original anatomical context, indicating the tadpole sank to the bottom of the lake in an anaerobic environment shortly after its death (cause cannot be reconstructed), and before putrefactive gases could have formed to an extent that would have prevented it from sinking (for related taphonomic regularities please see Wuttke and Reisdorf 2012; Reisdorf and Wuttke 2012; Smith and Wuttke 2015, and literature therein). A great depth of water or a ,stick’ effect from fluids released from the carcass while resting on the fluid-soaked surface of the sediment (Orr et al. 2016) or adhesion due to the electrochemical properties of the clayey materials the carcass rested on, prevented resurfacing, regardless of the development of putrefactive gases. The onset of microbial decomposition of the soft tissue, which originated in the gut, led to the outflow of the pollen-containing gut contents into the abdominal cavity. After the decomposition of the skin, the pollen could also spread outside the body. Of the organic tissues, only cellular melanosomes escaped microbial degradation, as anaerobic bacteria cannot degrade them, followed by a diagenetically induced sulfurization to be preserved over geological times (McNamara et al. 2016).

Kingdom Plantae Haeckel, 1866

Gymnospermae

Order Pinales Gorozh., 1904

Family Pinaceae Spreng. ex F. Rudolphi, 1830

Genus Pinus L., 1753

Subgenus Pinus L., 1753 (Diploxylon)

Pinus subgenus Pinus sp.

(Fig. 3a–c)

Fig. 3
figure 3

Light microscopy (a, d, g, j) and scanning electron microscopy (b, c, e, f, h, i, k, l) micrographs of pollen extracted from the intestine of the fossil giant tadpole from Enspel.  ac Pinus subgenus Pinus sp., close-up of corpus (c).  df Picea sp., close up of corpus (f).  gi Ulmus sp., close if porus (i).  jl Fagus sp., close up of colpus and polar area (l). Scale bars 10 µm (a, b, d, e, g, h, j, k) and 1 µm (c, f, i, l)

Previous reports: 2009 Pinuspollenites sp. – Herrmann et al., p. 63, pl. 5, figs 2, 3 (Enspel).

Description: Pollen, monad, bisaccate, P/E ratio oblate, corpus elliptic in polar view (PV); sacci spherical, narrowly attached to corpus, sacci attachment area diameter 25–30 μm (LM), sacci width 40–45 µm (LM, SEM), sacci height 30–35 µm (LM); pollen diameter including sacci 70–80 µm (LM), 60–70 μm (SEM), corpus diameter 60–60 µm (LM), 45–55 μm (SEM), corpus width 45–55 μm (LM), 40–50 µm (SEM); leptoma; corpus exine 1.8–2.7 μm thick; sacci with indistinct alveolate structuring (LM); exine sculpture scabrate, psilate in leptoma area (LM), sculpture in cappa area nanoverrucate, perforate (SEM), sacci psilate, perforate (SEM).

Remarks: Pinus subgenus Pinus (Diploxylon) pollen is characterised by near spherical sacci with narrow attachment to the corpus (Halbritter et al. 2018). Extant Pinus pollen has been investigated by several authors using LM and/or SEM (e.g. Erdtman 1965; Klaus 1977, 1978; Jacobs 1985). Pinus pollen grains are easily distinguished from similar bisaccate pollen of Abies, Picea and Pseudolarix (Bagnell 1975; Weir and Thurston 1977; Zanni and Ravazzi 2007). Pollen morphology within a single extant Pinus species can be variable (size, sacci shape, wall structure, conspicuousness of nodula, etc.), and many species show overlapping pollen characteristics. Pinus pollen is rare in the tadpole gut contents (2.33 %).

Fossil records: The earliest fossil records of Pinus have been reported from Cretaceous sediments (e.g. Singh et al. 2018). In the Northern Hemisphere, Pinus is a common and frequent component of various vegetation units from the Eocene onwards (Ferguson 1967; Axelrod 1986; Mai 1986; Millar 1998; Eckert and Hall 2006; Erwin and Schorn 2005). Pinus pollen, belonging to both subgenera Pinus and Strobus, is present in many mid-Paleogene palynofloras of Asia (Zheng et al. 1999; Wang 2006), North America (Graham 1999) and are regularly encountered in Oligocene floras of Europe (e.g. Krutzsch 1971). Pinus seeds have also been reported from the Enspel sediments (Köhler and Uhl 2014). The form-genus Pinuspollenites (e.g. Krutzsch 1971; Stuchlik et al. 2002) is commonly used for fossil Pinus pollen.

Genus Picea A.Dietr., 1824

Picea sp.

(Fig. 3d–f)

Previous reports: 2009 Piceapollis tobolicus (Panova 1966) Krutzsch 1971 – Herrmann et al., p. 62, pl. 4, fig. 7 (Enspel); 2009 Piceapollis planoides Krutzsch 1971 ex Hochuli 1978 – Herrmann et al., p. 62, pl. 5, fig. 4 (Enspel).

Fig. 4
figure 4

Frequency of pollen types in the gut of the fossil tadpole

Description: Pollen, monad, bisaccate, P/E ratio oblate, pollen outline elliptic to circular in PV and EV; sacci half spherical, small and originating well above the equator, broadly attached to corpus, sacci attachment area diameter 70–90 μm (LM), 65–75 µm (SEM), angle between corpus and sacci wide in EV, 145–160 wide; pollen diameter including sacci 115–140 μm, 90–110 µm (SEM), corpus diameter 90–110 μm (LM), 80–100 µm (SEM), corpus height 70–90 μm (LM); leptoma; exine in cappa area 2–2.5 μm thick (LM), sacci with alveolate structuring, alveolate pattern 4–5 μm thick (LM); corpus exine sculpture scabrate (LM), microverrucate, perforate, fossulate (SEM), sacci granulate, perforate (SEM).

Remarks: On the figured specimen’s saccus the impression of a diatome alga can be seen, this was regularly observed on pollen grains in the tadpole gut contents. Extant Picea pollen comparable to the fossil pollen grains have been investigated by several authors using LM and/or SEM (e.g. Erdtman 1965; Huang 1972; Birks 1978; Lindbladh et al. 2002; Fujiki et al. 2005; Li et al. 2011; Miyoshi et al. 2011).

Fossil record: Picea macrofossils and pollen grains are frequent in the fossil record. The earliest remains are ‘Picea’-like wood fragments from the Lower Cretaceous of Arctic Canada (Bannan and Fry 1957), the Lower to Upper Cretaceous of China (He 1995; Ding 2000), and the Upper Cretaceous of Kamchatka, Russia (Blokhina and Afonin 2009) and Japan (Nishida and Nishida 1995; Nishida et al. 1995). Additionally, several cones, cone scales, and seeds of Picea have been reported from the Lower and Upper Cretaceous of the Russian Far East (Volynets 2005; Bugdaeva et al. 2006; Blokhina and Afonin 2009). In Paleocene sediments, only a few macrofossil records have been reported, but in the Eocene, Picea fossils have been reported from many palaeofloras in North America, Europe, and in Asia (Ferguson 1967; Mai 1995; Graham 1999; LePage 2001; Blokhina and Afonin 2009). The fossil record indicates that Picea was widespread and a frequent component of post-Paleocene Northern Hemispheric floras until the Pliocene–Pleistocene. Still, no Picea macrofossil have been discovered in the Enspel sediments up to date. The form-genera Piceapollenites (Potonié, 1958) or Piceapollis (Krutzsch, 1971) are commonly used for fossil Picea pollen.

Angiospermae

Order Rosales Bercht. and J.Presl, 1820

Family Ulmaceae Mirb., 1815

Genus Ulmus L., 1753

Ulmus sp.

(Fig. 3g–i)

Previous reports: 2010 Polyporopollenites undulosus (Wolff 1934) Thomson et Pflug 1953 – Herrmann et al., p. 45, pl. 10, fig. 21 (Enspel).

Description: Pollen, monad, P/E ratio oblate, outline circular to elliptic in PV; equatorial diameter 32–38 μm wide (LM), 28–34 μm (SEM); tetraporate, pori lalongate; exine 1.2–1.8 μm thick, exine slightly thickened around pori; sculpture rugulate (LM, SEM), nanoechinate, perforate (SEM), rugulae width 1.0–1.2 μm (SEM).

Remarks: Extant Ulmus pollen has been investigated by several authors using LM and/or SEM and sometimes TEM (e.g. Huang 1972; Lieux 1980; Zavada 1983; Zavada and Dilcher 1986; Takahashi 1989; Xin et al. 1993; Jones et al. 1995; Stafford 1995; Wang et al.1995; Morita et al. 1998; Beug 2004; Li et al. 2011; Miyoshi et al. 2011).

Fossil record: Denk and Dillhoff (2005) summarised the early Ulmus-like fossil (Paleocene-Eocene) records. From the early Eocene McAbee and contemporaneous floras of British Columbia, western Canada, associated and attached fruits and leaves of U. okanaganensis Denk et Dillhoff have been reported (Denk and Dillhoff 2005). Manchester (1989) summarised the Eocene to Oligocene fossil records of Ulmus and suggested that unambiguous Ulmus remains were not found in Eurasia before the Oligocene. Also, fossil Ulmus leaves have been described from the Enspel sediments (Köhler and Uhl 2014). The form-genera Ulmipollenites or Polyporopollenites (e.g. Wolff 1934; Stuchlik et al. 2009) are commonly used for fossil Ulmus pollen.

Order Fagales Engl., 1892

Family Fagaceae Dumort., 1822

Genus Fagus sp. L., 1753

Fagus sp.

(Fig. 3j–l)

Previous reports: 2010 Faguspollenites verus Raatz ex R.Potonié 1960 – Herrmann et al., p. 25, pl. 8, fig. 23 (Enspel); 2010 Faguspollenites gemmatus Nagy 1969 – Herrmann et al., p. 25, pl. 8, fig. 19 (Enspel); 2010 Faguspollenites minor Nagy 1969 – Herrmann et al., p. 26, pl. 8, fig. 20 (Enspel); 2010 Faguspollenites subtilis Nagy 1969 – Herrmann et al., p. 26, pl. 8, figs 24, 25 (Enspel)

Description: Pollen, monad, outline circular to elliptic in EV; polar axis 32–42 μm long (LM), 28–38 μm (SEM), equatorial diameter 32–42 μm wide (LM), 28–38 μm wide (SEM); tricolporate, colpi narrow, colpus length 1/2 to 2/3 of polar axis; exine 1–1.5 μm thick (LM); sculpture scabrate (LM), rugulate, fossulate (SEM), rugulae often diverging and protruding (SEM).

Remarks: Extant Fagus pollen has been investigated by several authors using LM and/or SEM and sometimes TEM (e.g. Crepet and Daghlian 1980; Miyoshi 1982; Saito 1992; Denk 2003; Wang and Pu 2004; Li et al. 2011; Miyoshi et al. 2011). Pollen from members of subgenus Engleriana (incl. F. engleriana Seemen, F. japonica Maxim., F. okamotoi C.F. Shen) differ slightly from those of subgenus Fagus (all other species). Fagus subgenus Engleriana pollen have long, narrow colpi that extend nearly to the poles and the pollen grains are generally smaller than those of subgenus Fagus (e.g. Praglowski 1982; Denk 2003).

Fossil record: Extending back to the Eocene, Fagus is well represented in the Cenozoic fossil record (e.g. Tralau 1962; Zetter 1984; Kvaček and Walther 1991; Denk and Meller 2001; Denk 2004; Manchester and Dillhoff 2004; Grímsson and Denk 2005). Denk and Grimm (2009) summarised the fossil record and biogeographic history of Fagus. Cupules, fruits, foliage, and pollen of Fagus langevinii Manchester et Dillhoff (Manchester and Dillhoff 2004) from the late early Eocene of British Columbia, western Canada and Washington, north-western United States are currently the earliest unequivocal fossils of this genus (for age determination see Denk and Dillhoff (2005). Further, Fagus macrofossils and pollen have been reported from the middle Eocene of Vancouver Island (western Canada), Axel Heiberg Island (arctic Canada) and western Greenland (e.g. McIntyre 1991; Richter and LePage 2005; Mindell et al. 2009; Grímsson et al. 2015). In Asia, the oldest Fagus macrofossils to date are of middle to late Eocene age and were reported from the Russian Far East and north-eastern China (e.g. Denk and Grimm 2009). The fossil record indicates that Fagus dispersed from East Asia into Europe following the closure of the Turgai Seaway during the early Oligocene. Dispersed fossil Fagus pollen studied using combined LM and SEM has been reported from the late early Eocene of British Columbia (Manchester and Dillhoff 2004) and the Russian Far East (Narishkina and Evstigneeva 2020), the middle Eocene of western Greenland (Grímsson et al. 2015), the early Oligocene of Germany (Denk et al. 2012), and the Late Miocene of Japan (Saito 1992). Fossil Fagus cupules and leaves are also known from Enspel (Köhler and Uhl 2014). The form-genus Faguspollenites (e.g. Raatz 1937; Stuchlik et al. 2014) is commonly used for fossil Fagus pollen.

Pollen counts

For a statistical evaluation of the pollen content discovered in the gut of the tadpole we counted 300 grains to assess the frequency of each pollen type (Fig. 4). Based on this, Picea pollen is the most dominant type, representing about 96.3 % of all grains counted. The other three genera are much rarer, with Pinus pollen at 2.3 % and pollen from both Fagus and Ulmus with less than 1 % occurrences.

Discussion

Macro/microflora from Enspel versus gut contents of the tadpole

The macro- (leaves, fruits, seeds, etc.) and microflora (spores, pollen) of bryophytes, pteridophytes, gymnosperms, and angiosperms from Enspel have been studied in detail. Herrmann et al. (2009, 2010) identified 41 different spore morphotypes, 20 types of gymnosperm pollen, including, among others, bisaccate pollen of Abies, Cathaya, Cedrus, Keteleeria, Picea and Pinus, and 80 different angiosperm pollen taxa, representing both various monocots and dicots (incl. both Fagus and Ulmus), from this locality (Table 1). The macroflora reported by Köhler and Uhl (2014) only depicted remains from three different spore producing plants, specifically a single bryophyte and two pteridophytes. Gymnosperms producing bisaccate pollen are represented only by Pinus remains; other gymnospermous fossils were assigned to Taxaceae, Cupressaceae, or Cephalotaxaceae. The angiosperm component is the most diverse among the macroremains (Table 2), suggesting the presence of about 30 families of both monocots and dicots (incl. Fagaceae, Fagus; Ulmaceae, Ulmus). In short, out of the four genera we discovered in the gut of the fossil tadpole, three of them, Pinus, Fagus, and Ulmus, are known from both the fossil dispersed palynoflora (Herrmann et al. 2009, 2010) and the macroflora (Köhler and Uhl 2014), but Picea has only been recorded from the dispersed palynoflora (Herrmann et al. 2009, 2010).

Table 1. Palynomorphs from Enspel investigated with LM
Table 2. Plant macrofossils from Enspel

Today, Picea occurs all over the Northern Hemisphere (Fig. 5), and the different species show variable ecological amplitudes (Farjon 1990, 2010). Some Picea thrive in boreal conifer forests, others in oceanic climates, and some occur only inland where a dry continental climate prevails (Farjon 1990, 2010). Still, there are numerous reports of Eocene to Miocene Picea growing in lowland environments (swamps, hummocks, etc.) close to waterbodies, suggesting they were minor components of conifer swamp forests during most of the European Cenozoic (e.g. Figueiral et al. 1999; Dolezych and Schneider 2006). Based on the amount of Picea pollen found in the gut of the tadpole, and despite the lack of macrofossils, we assume a similar phenomenon and that Picea was growing close to the waterbody where the sediments accumulated. As with Picea, the genus Pinus currently has a wide Northern Hemispheric distribution (Fig. 5) and remarkably diverse ecological tolerance (Farjon 2005, 2010). The habitats in which Pinus can be encountered extend from tropical to high-latitude lowlands and mountains, both in coastal regions as well as inland (Farjon 2005, 2010). It is uncertain if the pines producing the pollen were growing close to the accumulation site. Theoretically, the parent plant could have been growing “far away” and the few pollen discovered in the gut of the fossil tadpole might have been transported some distances (see, e.g. fig. 87 in Klaus 1987). Considering the reported macrofossils (Köhler and Uhl 2014), Pinus was likely growing close to the palaeo-lake. If so, this contradicts a possible link between the rare occurrence of the cockroach Latiblattella basaltica at Enspel to the supposed scarcity of conifers in the immediate vicinity of the palaeo-lake (Vršanský et al. 2022); instead, taphonomic filters, such as limited transport of preferentially ground-dwelling insects into the depositional environment, are more likely explanations.

Fig. 5
figure 5

Distribution and climate preferences of extant genera. Note: Geographic distribution of Fagus, Picea, Pinus, and Ulmus, based on GBIF occurrence data sets modified using chorological literature (Ohwi 1965; Browics and Zieliński 1982; Nixon 1997; Thompson et al. 1999a, b, 2001, 2006; eFloras 2008; Fang et al. 2009; Caudullo et al. 2017). Köppen climate map modified after (Cui et al. 2021)

As with Picea, several authors have shown that Pinus was an important component of the European Eocene to Miocene lowland wetland vegetation (e.g. Mai 1986, 1995; Figueiral et al. 1999; Philippe et al. 2002; Teodoridis and Sakala 2008). Ulmus and Fagus are Northern Hemispheric genera (Fig. 5), but Ulmus has a much wider and more continuous distribution than Fagus. Both genera occur in typical mixed broad-leaved evergreen-deciduous forests, but Fagus grows on well-drained soils compared to Ulmus, which is also prominent along rivers/streams (Zhou and Li 1994; Peters 1997; Fu et al. 2003; Fang et al. 2009). Since both genera are also represented in the Enspel macrofossil record (Köhler and Uhl 2014), we suggest that Ulmus and Fagus were growing just outside the rim of the ancient Enspel crater, with Fagus on well-drained slopes and Ulmus along nearby streams. Previous work on the Enspel palaeoflora by Herrmann et al. (2009, 2010) and Köhler and Uhl (2014) suggests a mixed forest vegetation corresponding in its composition to, among others, present-day East Asian forests. The forest vegetation was divided into various units, its foothills probably reached the crater rim of the Enspel Lake, which was only somewhat flatter in the area of slope foot sediments (Schindler and Wuttke 2015). In the latter area, an azonal flora was present, i.e. aquatic plants, a riparian plant community and a riparian forest (Herrmann et al. 2009, 2010; Köhler and Uhl 2014). This fits well within our visioned scenario Overall, we assume that the region where the Enspel sediments accumulated was characterised by mixed conifer (evergreen) and broadleaved (deciduous and evergreen) forest vegetation. The deciduous broadleaved component was dominant, with evergreen broadleaved shrubs and small trees as part of the understory or occurring in sheltered areas or/within microclimates (ravines, gullies, etc., e.g. as proposed by Smith and Wuttke (2015)).

Pollen dispersal and pollination interval

All four plant genera (Picea, Pinus, Fagus, Ulmus) discovered in the gut of the fossil tadpole are currently confined to the Northern Hemisphere, with a primarily mid-latitude distribution (Fig. 5). Both Picea and Pinus stretch slightly into high latitudes at the rim of the arctic circle, and Pinus extends southwards into low-latitudes in both Central America and southeast Asia. The wide distribution of both Picea and Pinus is reflected in their broad climate tolerances (Fig. 5, 6; Table 3). Both genera grow under varied climates, including arid- (B), warm temperate- (C), and snow- (D) climates (for climate category definitions, see Kottek et al. 2006). Ulmus, and especially Fagus show a more selective climate range (Fig. 5, 6; Table 3). While Fagus is restricted to prevailing warm temperate- (C) and snow- (D) climates, a few Ulmus species extend this genus’ distribution into equatorial- (A) climates. At present, the four encountered genera co-occur in three geographic regions: 1) Central East North America, 2) Western Eurasia (Europe/Caucasus), and 3) Southeast/East Asia (Fig. 5). In North America the plants thrive under a fully humid warm temperate- (Cf) or snow-climate (Df) with hot (Cfa, Dfa) to warm summer (Cfb, Dfb). The climate variant depends on latitude and elevation (lowland versus mountains) where the plants are growing. In Europe, the genera thrive together mostly in fully humid warm temperate- (Cf) or snow-climate (Df) with warm summer (Cfb, Dfb). Again, in relation to latitudinal distribution and elevation. In Far East Asia (Japan), the scenario is quite similar, with plants thriving in lowland regions under fully humid warm temperate-climate with hot summer (Cfa) and extending into snow-climate variants with warm summer (Dfb) at higher elevations. In South East Asia (China and neighbouring countries), the taxa also co-occur in fully humid warm temperate-climate variants (Cfa, Cfb) but equally well in warm temperate regions enduring winter dryness (Cwa, Cwb) (Fig. 5, 6). All regions and climate types hosting the plant taxa encountered in the fossil tadpole are characterised by seasonal changes (winter, spring, summer, autumn) with distinct differences in summer versus winter temperatures and deciduous broadleaf components in the forest vegetation (e.g. Ohwi 1965; Thompson et al. 1999a, b, 2001, 2006; eFloras 2008; Fang et al. 2009). Also, considering previously reported angiosperm plant macrofossil records from Enspel (Table 2) a deciduous broadleaved forest vegetation is further supported by the presence of taxa such as Acer, Betula, Carpinus, Carya, Cercidiphyllum, Craigia, Crataegus, Cornus, Fagus, Juglans, Liriodendron, Platanus, Quercus, Tilia, Ulmus, and Zelkova. The dispersed palynoflora indicates a similar signal (Table 1). Previous palaeoclimatic studies on Cainozoic floras of Central Europe, e.g. Mosbrugger et al. (2005), show a distinct drop in temperatures from the Eocene onwards and an overall increase in seasonality until the present. These authors also suggested that seasonal temperature variations during the late Oligocene of Central Europe might have had a 20°C temperature difference between the coldest winter day/month/period and the hottest summer day/month/period (Mosbrugger et al. 2005). Also, based on the Enspel palaeoflora, Uhl and Herrmann (2010) estimated the mean annual temperature to have been around 15-17°C, the warmest month mean around 25°C, and the coldest month mean about 5-7°C. Regardless of the actual temperature differences, the Enspel flora (Herrmann et al. 2009, 2010; Uhl and Herrmann 2010; Köhler and Uhl 2014) and other Central European Oligocene floras (e.g. Mosbrugger et al. 2005; Uhl and Jasper 2018; Uhl et al. 2022) distinctly suggest that seasonality already occurred at that time. This implies that the late Oligocene Enspel region/flora endured seasonal changes and flowered/bloomed/pollinated during particular intervals of the year. In regions where potential modern analogues of the plants discovered in the gut of the fossil tadpole currently co-occur, their flowering/pollination period overlaps from April to May and partly into June, with all taxa completing their pollination between February and July (Table 4). As noted in the pollen count, Picea pollen was by far the most frequent type, with the rare occurrence of Pinus, Fagus, and Ulmus. This suggest that the tadpole was feeding from the water surface during the ‘main’ pollination interval of Picea (May and June), at a time when Pinus/Fagus/Ulmus had already released their pollen (only remnants remained floating on the water surface). Therefore, we assume that the tadpole was feeding on the pollen encountered during late spring / earliest summer, equivalent to present-day May or June.

Fig. 6
figure 6

Köppen profiles of Picea, Pinus subgen. Pinus, Fagus, Ulmus, and combined. Note: Köppen profiles are based on the accumulated GBIF occurrence data for available species, only values over 5% are included in the barcharts. Combined Köppen profiles comprise all four genera under study, (A) showing all Köppen climate types, (B) only including values over 5%

Table 3. Köppen climate types of encountered taxa
Table 4. Main pollination periods in co-occurrence regions of the encountered taxa

How pollen organic matter enters freshwater food webs

Deciduous tree and conifer pollen deposition is a visually striking spring event in lakes, displayed by a yellowish scum at the air-water interface; there, they become neustonic particles. The pollen grains represent a potentially major allochthonous input of limiting nutrients (Graham et al. 2006; Masclaux et al. 2013). Pollen grains of most deciduous tree species became waterlogged after a short time and began to sink (Graham et al. 2006), probably forced by rupture due to osmotic pressure (Roulston and Cane 2000). However, air bladders make the saccate pollen grains of coniferous trees buoyant (Hopkins 1950; Leslie 2010; Masclaux et al. 2013). Conifer pollen amendments to the neuston significantly increased total phytoplankton, especially filamentous green algae and diatoms, and herbivorous zooplankton biomass (Turner et al. 2006). Graham et al. (2006) reported that the positive effects of pollen on zooplankton were likely indirect via its nutritive effect on edible phytoplankton and other smaller food items (e.g. protozoans, bacteria) rather than direct via consumption. It can therefore be assumed that tadpoles are attracted not only by the pollen, which releases nutrients but also by the rapidly reproducing zooplankton.

Olfaction across the water-air interface in tadpoles

Tadpoles appear to be nearsighted, and as such, visual cues appear to play little role detecting prey from great distances or with much precision (Hoff et al. 1999). Sensory perception of chemical cues plays a key role in the detection of food and foraging behaviour in tadpoles (Veeranagoudar et al. 2004; Gazzola et al. 2022). Hydrophilic molecules, ranging from small organic molecules to large proteins, easily dissolve in water, and thus make them accessible to aquatic olfactory organs of tadpoles (Weiss et al. 2021). After the release of the above-mentioned odorants of the pollen scum, water-borne odor molecules can successfully interact with olfactory receptors on receptor neurons of the tadpoles and direct them via pollen scum.

Nourishment content of pollen – was it worth a bite?

Pollen grains contain all components necessary for germination and pollen tube formation, including proteins, lipids, carbohydrates, minerals, vitamins, and other plant metabolites (Roulston and Cane 2000; Roulston et al. 2000). Proteins, polypeptides, and amino acids make up between 2.5 and 61 % of the dry weight of pollen (Roulston et al. 2000), and 0.5% phosphorus by weight (Turner et al. 2006). The lipid portion of pollen represents the fats and oils in sticky pollen coatings (pollenkitt) found on the pollen surface, the pollen protoplast (containing vegetative and generative nuclei/cells), and the intine. Lipids in pollen come in various forms, such as fats, waxes, sterols, free fatty acids, and pigments (Roulston and Cane 2000; Stanley and Linskens 1974). The carbohydrate content in pollen is mainly stored as starch in amyloplasts, ranging from 0 to 22 % (Roulston and Buchmann 2000). Minerals contribute between 1 and 7 % of the dry weight of pollen, while vitamins are only present in low quantities (Lunden 1954; Solberg and Remedios 1980). Therefore, pollen can be a valuable food source for various animals, especially insects and small vertebrates (Fægri and van der Pijl 1979). Pollen grains from entomophilous plants tend to have a higher protein content than those from strictly anemophilous plants (Ruedenauer et al. 2019). The parent plants producing the pollen grains we detected in the intestine of the fossil tadpole (i.e. Pinus, Picea, Ulmus, and Fagus) are all wind pollinated and therefore lack the sticky pollen coatings. However, they were probably still full of various nutrition components and represented a valuable food source for the Oligocene tadpoles as reflected by their present caloric values: Ulmus americana (5830 cal/g), Pinus ponderosa (5870 cal/g), Pinus edulis (6080 cal/g) (Collin and Jones 1980). Also, McLellan (1977) calculated ca. 11 kJ/g of pollen available to animal consumers, excluding the non-digestible pollen wall. Pollen of Picea, the most abundant pollen type in the stomach of the tadpole, contains 5.5 % ash (minerals), 56 milliequivalent (mEq) of potassium, 239 mEq nitrogen, 21 mEq phosphorous, and 20 mEq sulphur (Knight et al. 1972). Mature pollen of Picea glauca contains lipid globules, proteins, and carbohydrates in the form of small starch grains inside plastids (Dawkins and Owens 1993). Pinus halapensis pollen contains about 2.5 % dry-weight (dw) lipids, such as phospholipids, fatty acids, waxes and glycolipids (Andrikopoulos et al. 1985). In addition, Doskey and Ugoagwu (1992) analysed nutrients in pollen of Pinus strobus and P. resinosa and discovered that it contained ca. 500 mg/g carbon, 20 mg/g nitrogen (=11.2% protein; conversion factor 5.6 after Mariotti et al. 2008), 2–2.8 mg/g sulphur, and about 3 mg/g phosphorous. Pinus edulis and Pinus pondersosa pollen contain around 5.46 % and 4.00 % starch by dry weight, respectively (Roulston and Buchmann 2000). Based on a polysaccharide test with iodine staining Pinus mugo contains large amounts of starch (Halbritter et al. 2023). Fagus sylvatica pollen contains, as percentage dry-matter, 11.4 % ether extractable matter (roughly corresponding to the lipid fraction), 1.95 % ash (minerals), 2.58 % nitrogen (ca. 14.5 % protein; Mariotti et al. 2008) and 20.4 % “lignin” (McLellan 1977). From ultrastructural studies on Ulmus (U. laevis and U. minor) both species contain lipids and starch as pollen reserves (Halbritter et al. 2016, 2020). Based on all the components present in pollen of potential modern analogues of the fossil taxa and their nutrition values, it can be concluded that the tadpole gained considerable amounts of protein, carbohydrates, and lipids from ingesting fresh Picea, Pinus, Ulmus, and Fagus pollen floating on the surface of the Enspel palaeo-lake. In short, it was surely worth a bite.

Gut contents of fossil/extant tadpoles and their feeding behaviour

Identification of fossilised stomach or gut content in anurans has rarely been reported. Špinar (1972) for example, examined several hundred skeletons of adult palaeobatrachid frogs from the late Oligocene site of Bechlejovice but did not report any identifiable gut contents. Only Xing et al. (2019), Keller and Wuttke (1997), and Wuttke and Poschmann (2010) reported on vertebrate prey in a pipanuran, an eopelobatid, and a palaeobatrachid frog from the Early Cretaceous of China, the Eocene Messel maar lake, and Lake Enspel, the last two in Germany, respectively. Fossil tadpoles from the Early Miocene of Güvem, Turkey, also show intestine filling of uncertain origin (Dubois et al. 2010). McNamara et al. (2009, 2012) examined semiaquatic frogs (Rana pueyoi) from the Miocene of Lake Libros in Spain. They found stomach contents in almost two-thirds of their adult specimens consisting mainly of remains of aquatic planorbid and bithynid gastropods (McNamara, in Wuttke and Poschmann 2010). Furthermore, in single specimens, the water plant Ruppia, seeds, sponge spicules, tadpole bones, arthropod cuticles and phosphatized fibres of a possible ingested vertebrate were identified (McNamara et al. 2009). In slightly more than one-quarter of the tadpoles of the same taxon, McNamara et al. (2010) found contents of the intestine with a granular texture comprising fragmented diatoms and siliceous sponge spicules. Based on the high abundance of diatoms in the gut contents, these authors concluded that rasping of algal communities associated with submerged vegetation was a vital feeding strategy in the Libros tadpoles. Gut content consisting of siliceous microfossils, possibly diatoms, has also been found in the Enspel tadpoles but requires further examination. Wagner (1986) reported extant Hyla tadpoles feeding on wind-driven pollen accumulations at the surface of a pond and observed that the frog larvae changed their feeding behaviour to exploit this resource and that the tadpoles were also able to sense differences in the nutritional value of pollen and adjusted their feeding behaviour accordingly. Kloh et al. (2021a, b) experimentally proved that Ololygon machadoi tadpoles feed on pollen floating on the surface even when another, easier accessible food source is present. They hypothesise that pollen’s high caloric and nutritional value is worth exploiting even though the tadpoles must change their behaviour from a benthic scavenging lifestyle to a more “acrobatic” one. Feeding on lipoprotein surface films may be a rich source of nutriments, such as proteins (Goldacre 1949), sustaining tadpoles, especially in their premetamorphic phase of ontogeny (Nathan and James 1972). The silty/sandy intestinal fill recorded in the Enspel tadpoles possibly also stimulated peristalsis and thus the breakdown of plant cells by additionally acting as some kind of grinding medium (Nathan and James 1972) or feeding on bacteria and algae, adhering to the grains or on the mesopsammon (Wuttke et al. in prep.).

Conclusions

The Enspel Fossil-Lagerstätte has yielded several cases of “fossilised behaviour” concerning the feeding ecology of organisms within this late Oligocene ecosystem. Hitherto scientists have recorded plant damage by insects (Gunkel and Wappler 2015), evidence of cyprinid fish preying on weevils (Legalov and Poschmann 2020), a palaeobatrachid frog that fed on cyprinid fish (Wuttke and Poschmann 2010), a phasianid bird with a gastric mill of ingested stones and associated plant remains (Mayr et al. 2006), a mousebird which swallowed Oleaceae seeds (Mayr 2013), pellets produced by birds (possibly owl birds) that mainly preyed on rodents but also on other vertebrates (Smith and Wuttke 2015), and a possible case of crocodile feeding on a gaviid waterbird (Mayr and Poschmann 2009). Adding to the list pollen feeding by a giant pelobatid tadpole shows that fossils from this locality continue to provide vital information on food webs, animal behaviour, their interactions with the environment, and their role in this ancient ecosystem. The pollination period of potential modern analogues of the plant taxa encountered in the gut of the giant tadpole suggests that it fed on this pollen during late spring or earliest summer, providing a fragile time-related framework for the extension of its larvae stage prior to metamorphosis into a frog. The amount and purity of pollen discovered in the gut of the tadpole suggest intentional and selective feeding of pollen floating on the water’s surface. Still, in tadpoles of the genus Pelobates (Fig. 7), as at Enspel, neustophagia has been previously described. This inertial feeding mechanism for ingesting fine particles from water surfaces is not confined to pollen. Therefore, it remains open whether the giant Oligocene tadpole selectively fed only on pollen or whether pollen was just one course from a zooneuston-rich menu during opportunistic surface feeding.

Fig. 7
figure 7

Surface feeding of a Pelobates fuscus tadpole. Note: Courtesy of Ana Ćurić (Zagreb)