Proteaceae

Proteaceae View in CoL in the early Paleogene of South America

We establish the presence of a new Proteaceae leaf species in the latest Paleocene of southern South America and highlight that the affinities are with extant Australasian taxa. The species is also the oldest convincing leaf record of the family in the Americas, with the previous oldest being several early Eocene taxa from Laguna del Hunco, Patagonian Argentina ( Gonzalez et al. 2007). Other southern South American early Paleogene (late Paleocene or early Eocene) floras, including some that have been collected and researched extensively, lack records of Proteaceae macrofossils. These floras are from strata exposed at Caleta Cocholgüe ( Gayó et al. 2005) and Quinamávida ( Troncoso 1992) in Chile, the Ligorio Márquez Formation that crops out on the Chile – Argentina border ( Troncoso et al. 2002; Carpenter et al. 2018), and, in Patagonian Argentina, the Danian Salamanca and Peñas Coloradas formations ( Iglesias et al. 2021) and the lower member (Eocene) of the Rio Turbio Formation ( Panti 2018). Further south, in Antarctica, several late Paleocene leaf impressions from the Cross Valley Flora of Seymour Island were referred to Proteaceae by Dusén (1908) and two taxa were accepted by Tosolini et al. (2013) at family level. However, as noted by Doktor et al. (1996) with regard to the Eocene La Meseta Formation and also accepted more recently by Tosolini et al. (2023), these and other fossils from Seymour Island are too poorly preserved for unequivocal assignment to the family. Similarly, fossil woods unequivocally attributable to Proteaceae have not been documented from the Antarctic Peninsula region ( Oh et al. 2020; Tilley 2024).

Although there is little pre-Eocene macrofossil evidence of Proteaceae in South America, several pollen morphospecies have been recognised in the south of the continent from the Campanian onwards ( Askin and Baldoni 1998) and proteaceous pollen is common in Eocene strata of Seymour Island ( Askin 1991). The chronology of records and typically much lower diversity than in the Australian region has usually indicated eastwards spread from a southern Australian diversification centre via Antarctica ( Askin and Baldoni 1998; Dettmann and Jarzen 1998; Cantrill 2018). Similarly, the general scarcity of proteaceous pollen and absence of grevilleoids in Maastrichtian–Eocene strata of the Salta Basin in the northwestern corner of Argentina ( Quattrocchio and Volkheimer 1990; Llorens et al. 2022) is consistent with a southern origin, although Lamont et al. (2024) proposed an arrival from northern South America instead. Whatever the true history of the family in the region, we consider that the pre-Eocene paucity of macrofossils is likely an indication that Proteaceae plants were not common in the South American vegetation at that time or had low fossilisation potential due to one or a combination of plants being of small stature, not sclerophyllous or growing in habitats remote from depositional sites.

Regarding the Arauco coals, Proteaceae pollen was evidently neither common nor diverse, and neither of the two Proteacidites Cookson ex R.A.Couper spp. recorded by Palma-Heldt (1980) and Collao et al. (1987) are usefully informative for affiliation to extant taxa.

Does the Lota–Coronel flora have a Neotropical signature?

Romero (1986), following Engelhardt (1891, 1905) and Berry (1922), placed the Lota–Coronel flora in the ‘northern, purely American Province’ (p. 457) of his Neotropical Paleoflora, consisting of genera (‘or their ancestors’) currently occurring exclusively within the Neotropical Phytogeographic Region of South America defined by Cabrera and Willink (1973). Gayó et al. (2005) continued to regard the Lota–Coronel flora as part of the Neotropical Paleoflora but with a revised list of>80% Neotropical and pantropical and <20% Australasian generic composition (citing Villagrán and Hinojosa 1997). The same composition was also used by Hinojosa (2003, 2005; see also Hinojosa et al. 2006; Quattrocchio et al. 2013), a co-author with Villagrán on the Gayó et al. (2005) paper, who reclassified the Lota–Coronel flora as part of his Gondwanic Paleoflora. Regardless of these potentially confusing alternative names, the acceptance of wholly or dominantly Neotropical floristic affinities in these studies is surprising, especially given that>80 years ago Florin (1940, p. 24) not only found that the Lota–Coronel conifers were incorrectly identified but also questioned the reliability of the angiosperm identifications, these being based only on external morphology and with many of the specimens being fragmentary.

We strongly agree with Florin, and emphasise that the only Lota–Coronel foliar fossils that have been rigorously identified to genus, i.e. Lygodium Sw. , Dacrycarpus (Endl.) de Laub. , Retrophyllum C.N.Page and Coronelia Florin (none of which were referred to in the phytogeographical studies above) do not have exclusively Neotropical affinities. Thus, Lygodium is currently globally widespread, mostly in the tropics, with L. skottsbergii evidently belonging to a complex of species that was distributed across a vegetated Antarctica in the Paleogene ( Rozefelds et al. 2017). Dacrycarpus is currently found only in Australasia, Malesia and Southeast Asia with D. chilensis closely matching D. dacrydioides (A.Rich.) de Laub. of New Zealand ( Florin 1940; Wilf 2012). Retrophyllum represents an interesting case of a Gondwanan genus with disjunct species in Australasia and northern South America, where the latter occurrence most likely resulted from a rare, deep northwards migration of an austral taxon ( Wilf et al. 2017). Coronelia is of uncertain affinity but we agree with Florin (1940) that this belonged to Podocarpaceae . The known distribution is purely austral, with well-preserved foliage being found also in the Ligorio Márquez Formation, Argentina ( Carpenter et al. 2018), Australian Eocene floras in Tasmania ( Townrow 1965) and Queensland (R. J. Carpenter, unpubl. data), and the Miocene of New Zealand (R. J. Carpenter, unpubl. data) to date. Although more taxonomic work is required, the presence of Caldcluvia paniculata (Cav.) D.Don-like leaves at Buen Retiro is further evidence of austral affinities because the extant species is endemic to Patagonian rainforests and the closest relatives are restricted to Australasia. Species of Parablechnum with foliage similar to the Buen Retiro fossils do occur in the Neotropics but these species occur mostly in montane areas and overall, the genus is typical of southern temperate zones ( Testo et al. 2022).

Among the Lota–Coronel taxa that have been unjustifiably linked to Neotropical genera, the proposed Lauraceae taxa are exemplary: at least 14 species from 9 genera were originally named ( Engelhardt 1891; Berry 1922) but numerous authors have since noted the impossibility of confidently assigning lauraceous leaf fossils to extant genera, even when cuticle is well preserved (e.g. Hill 1986; Bannister et al. 2012; Carpenter et al. 2018). Therefore, the abundant latest Cretaceous to early Paleogene records of the family from Lota–Coronel and other Patagonian and Antarctic Peninsula floras could alternatively represent lineages currently found remotely from the Neotropics, such as in Australasia.

Our brief assessment of the early studies and existing collections from Lota–Coronel sites other than Buen Retiro shows Sabal ochseniusi ( Arecaceae ) to be of interest as a possible Neotropical element. Although Sabal Adans. is currently found only in this region, many authors (e.g. Horn et al. 2009; Matsunaga and Smith 2021) have concluded that certain inference of closest affinity with Sabal is not possible because similar palmate or costapalmate foliage occurs widely in the large, mostly pantropical Coryphoideae, and palmate foliage also occurs elsewhere in the family. Similarly, the specimen assigned to the Neotropical cycad Zamia L. by Engelhardt (1891) was not verified by any cuticular evidence and may represent another cycad or monocot ( Wilf et al. 2014).

Similarly to the macrofossils, the biogeographic significance of the Lota–Coronel pollen flora has been overlooked or evidently misinterpreted in previous studies. Although many of the>40 spore and pollen taxa (e.g. Cyathidites Couper spp. , Gleicheniidites M.E.Dettmann , Myrtaceidites Cookson & K.M.Pike , the tricolpates) listed by Doubinger (1972), Doubinger and Chotin (1975), Palma-Heldt (1980) and Collao et al. (1987) have imprecise or widely distributed modern affinities, there is scant evidence of taxa with definite and exclusively Neotropical relatives, whereas there is abundant evidence of those with austral relatives ( Doubinger and Chotin 1975). These taxa include? Araucariacites Cookson sp. (? Araucariaceae ), Podocarpidites Cookson spp. ( Podocarpaceae ) and Proteacidites spp. ( Proteaceae ) and of particular note, Haloragacidites harrisii . This last pollen species is widely accepted as belonging to the exclusively Australasian family Casuarinaceae (e.g. Macphail 2007), not Myricaceae , as treated by all the abovementioned authors. Importantly, Casuarinaceae macrofossils ( Gymnostoma L.A.S.Johnson ), including foliage and staminate inflorescences with in situ H. harrisii pollen, also occur in the early Eocene Laguna del Hunco deposits of Argentina, and dispersed pollen grains have been found in numerous early Paleocene to Eocene palynoassemblages from other Patagonian localities ( Zamaloa et al. 2006). The long history of Nypa Steck-like pollen was recently reviewed by Pocknall et al. (2022), although these authors did not include the records of Spinizonocolpites from southern South America. Nypa abundance and diversity evidently reached a global peak during the warm Paleocene and Eocene intervals, including across northern South America, but became extinct in the Americas since that time, and was reduced to the single species N. fruticans elsewhere. We consider that the presence of Nypa in the Arauco coals was likely a consequence of southward range expansion, but the genus is clearly no longer a Neotropical genus.

Angiosperm leaf fossils from other Paleogene assemblages in central Chile and the Patagonian region, such as Quinamávida ( Troncoso 1992), Ligorio Márquez ( Troncoso et al. 2002) and the Río Turbio Formation ( Panti 2016, 2018; Vento and Prámparo 2018) have been assigned to taxa described by Engelhardt (1891, 1905) and Berry (1922) from Lota–Coronel, although in many cases, with severe reservations because of poor preservation, including lack of cuticular evidence (e.g. Troncoso et al. 2002). Although of floristic and physiognomic interest, these studies contribute little to the question of whether or not Neotropical elements were present. By contrast, detailed reviews of the wider South American palaeobotanical (especially palynological) evidence by Jaramillo and Cardenas (2013) and Jaramillo (2023) show that early Paleogene global warming events did not result in expansion of Neotropical forests into higher southern latitudes, nor broad migration into the tropics from temperate regions. These authors proposed that although thermal regimes may have been suitable, and some individual lineages could extend or change ranges, marked differences in both the variability and total amounts of solar insolation were likely limiting at the biome level.

Our work on the Lota–Coronel flora adds to other evidence that the strongest early Paleogene southern South American floristic connections were with the Australasian region, not the Neotropics, via an ice-free and vegetated Antarctica (e.g. Zamaloa et al. 2006; Wilf et al. 2009, 2014, 2016, 2017; Gandolfo et al. 1988, 2011; Hermsen et al. 2012; Wilf 2012; Carvalho et al. 2013; Knight and Wilf 2013; Macphail et al. 2013; Carpenter et al. 2014 b; Gandolfo and Hermsen 2017; Escapa et al. 2018; Jud et al. 2018; Rossetto-Harris et al. 2020) that was connected until final separation at the end of the Eocene ( McLoughlin 2001). Many of the pan-Antarctic lineages persist only in subtropical rainforest associations of Australasia, including several genera within the World Heritage Gondwana Rainforests of eastern Australia and others in Malesia ( Kooyman et al. 2014).

The Lota–Coronel palaeoenvironment

The nearest inferred extant relatives of Proteaceaefolia araucoensis are rainforest trees confined to warm, wet and humid climates in Australasia, consistent with physiognomic evidence for such climates in the early Paleogene of southern South America based on leaf fossils from both the Lota–Coronel flora and floras of similar age and geographic location (Ligorio Márquez and Caleta Cocholgüe) ( Romero 1986, 1993; Hinojosa and Villagrán 1997; Villagrán and Hinojosa 1997; Gayó et al. 2005). None of the other nearest living relatives of the Lota–Coronel fossils that has reasonably well-justified identifications contradict this conclusion, and some of the angiosperms in the assemblage have extended apices likely representing drip tips for shedding water in strongly humid settings ( Fig. 5 View Fig ). Together with the noted high abundance of fungal bodies and pteridophyte spores in the Arauco coal ( Collao et al. 1987), Haloragacidites harrisii ( Casuarinaceae ; presumably Gymnostoma ) pollen is of particular importance as an indicator of both palaeohabitat and palaeoclimate. This species represents up to ~43% of the total pollen composition in the samples examined by Collao et al. (1987), levels that likely indicate the type of dominance attained by extant Gymnostoma species of warm, lowland peat forest habitats of Borneo on poorly drained soils ( Brunig 1990). In terms of the other most reliably identified nearest living relatives, the presence of Lygodium , Dacrycarpus , Retrophyllum , Lauraceae and likely Cunoniaceae and coryphoid palms among the macrofossils, and Nypa and Gymnostoma among the microfossils, do not contradict the inference of a warm, mesic Arauco palaeoclimate, as all these taxa have overwhelmingly modern tropical to subtropical distributions in generally high rainfall and humid habitats.