Long before the description of the taxa by Poeppig (1832); Schomburgk (1837), and d’Orbigny (1840), Victoria was well-known to the Indigenous Peoples of South America, featuring in indigenous narratives (Prance and Arias, 1975) and with only partially documented cultural usages.

Local names for V. amazonica that have been recorded include ‘auapé-yaponna,’ after auapé (Jacana jacana), a small bird often seen running on its leaves (Schultes, 1985; Box et al., 2022). Victoria cruziana has been called ‘yrupé,’ ‘yacare yrupé,’ or ‘naanók lapotó’ (‘poncho del yacaré’) (Crovetto, 2012; Scarpa and Rosso, 2014; Mereles et al., 2020).

In 1832, after having been noted by several botanical explorers (”Haenke in 1801 and Bonpland in 1819,” and d’Orbigny in 1827; d’Orbigny, 1835, 1840), the German explorer and naturalist Eduard Friedrich Poeppig described a giant Water Lily, Euryale amazonica, that he had encountered in the environs of the Solimões river in Brazil that same year. Despite news of this discovery spreading quickly through Germany, it apparently reached neither Paris nor London and five years later the same species was described again, almost simultaneously, by the German botanist, Robert Schomburgk (Schomburgk, 1837) and the British botanist John Lindley (Lindley, 1837), under two different species epithets (regina, regia) and as a new genus (Victoria). Lindley’s species epithet prevailed in usage, possibly because of the great pomp with which it was delivered and its use to lobby a new monarch:

“It is therefore not less my duty …. in distinguishing your Majesty’s illustrious name, by far the most majestic species in the family of the Nymphs – one of the most noble productions of the Vegetable Kingdom – found in your Majesty’s South American dominions by a gentleman [Schomburgk] traveling under the auspices of your Majesty’s Government…” [Lindley, 1837].

The naming of the species was of great significance for the British scientific establishment as it came at a strategically important time, in the first months of the reign of Queen Victoria and at a time when several institutions were lobbying for royal patronage. Lindley’s opportunistic description of Victoria regia not only helped the Royal Geographical Society and the Horticultural Society of London (now Royal Horticultural Society) obtain patronage from Queen Victoria (Opitz, 2013), but also contributed to the decision not to close the Royal Botanic Gardens, Kew (Hooker, 1847).

According to Opitz (2013), it was also the first signal of the prominent status that science would come to play during her reign. The description of this species therefore had a much broader impact on botanical science in the British Victorian era, at a time when it was arguably at its most influential.

Given the fashion for greenhouses and cultivation of exotic plants, it is not surprising that the cultivation of Victoria in Europe and North America became a symbol of social status and horticultural achievement (Holway, 2013; Aniśko, 2014), with several Botanic Gardens even constructing dedicated greenhouses for the purpose. It could also be argued that it became a symbol of the British Empire, Paxton incorporating the structure of the leaf into the architecture of Crystal Palace (Aniśko, 2014).

For most of the 19th and 20th centuries, the giant water lily from Amazonia was incorrectly known by the binomial Victoria regia. This was despite two earlier binomials for the taxon having priority, Victoria R. H. Schomb. for the genus name, and amazonica Poepp. for the species epithet.

Lindley had described Victoria regia [Victoria Regia: 3 (1837)] to accommodate material collected by Robert Hermann Schomburgk in Guyana. He did so amid some controversy (d’Orbigny, 1840; Opitz, 2013; Aniśko, 2014). Not only had Lindley prepared his manuscript in secrecy and against Schomburgk’s instruction, but he did so a month after Schomburgk had unwittingly published his in the Athenaeum (September 9 vs. October 16), thanks to a presentation on Schomburgk’s behalf by John Edward Gray to the Botanical Society (Gray, 1837). In all likelihood, Lindley’s secrecy reflected his appreciation of the role it could play in securing royal patronage and he wanted his name to have priority over John Edward Gray’s publication, whose preparation he was aware of. Lindley was, however, likely unaware of Poeppig’s earlier description [Froriep’s Not. Natur- Heilk. 25: 131 (1832)] but was soon made aware of it by a German correspondent, Weissenborn, who drew attention to the fact in the Magazine of Natural History (Weissenborn, 1837), a widely read British publication. On hearing of this earlier name, the epithet of which would have had priority over ‘regia,’ he did not either respond or correct the nomenclature and it was only 10 years later that the German botanist Johann Friedrich Klotzsch published the corrected combination, V. amazonica (1847). Lindley’s epithet, however, remained that most commonly applied to material of V. amazonica for over a century until Ghillean Prance’s clarification of the nomenclature of the species (Prance, 1974), after which time it was slowly replaced by V. amazonica (Poepp.) Klotzsch.

Victoria cruziana Orb. was described as a second species by the French botanist Charles Henry Dessalines d’Orbigny [Ann. Sci. Nat., Bot., sér. 2, 13: 57 (1840)] based on material that he had collected in Corrientes, Argentina (considered Bolivia by d’Orbigny) in 1832. Since then, several additional names have been published, Euryale bonplandia Rojas Acosta [Cat. Hist. Nat. Corrientes: 151 (1897)], Euryale policantha Rojas Acosta [Cat. Hist. Nat. Corrientes: 65 (1897)], Victoria trickeri (Malme) Mutzek [based on V. cruziana f. trickeri Malme, Gartenwelt 29: 616 (1925)], all of which have subsequently been considered as synonyms of either V. amazonica or V. cruziana. In addition, two forms of V. cruziana were described by the Swedish botanist Gustaf Oskar Andersson Malme in 1907 [Acta Horti Bergiani 4(5): 12. 1907], Victoria cruziana f. trickeri Malme and Victoria cruziana f. mattogrossensis Malme.

Poeppig (1832) and Lindley (1837) disagreed over which genus the giant South American water lily should be assigned to, Poeppig assigning it to Euryale Salisb., a monotypic genus of large- leaved spiny water lilies from southern and eastern Asia, Lindley and Schomburgk considered it a distinct genus restricted to the neotropics. Both opinions are supported by molecular analyses which consistently recover species of Victoria and Euryale as sister to each other, either within (Löhne et al., 2007; Borsch et al., 2008), or sister to (Les et al., 1999; Borsch et al., 2008; Pellicer et al., 2013; Zhang et al., 2020) Nymphaea.

The large seeds of V. cruziana are consumed as a substitute for maize (Arbo et al., 2002; Mereles et al., 2020) and the rhizomes also have recorded usage as food (Scarpa, 2009). Similarly, amongst communities inhabiting the Paraguay river in the Pantanal region of Brazil, the seeds of V. cruziana f. mattogrossensis are ground with a pestle into a starch (Bortolotto et al., 2015). The petioles also have recorded usage as food (Kinnup and Lorenzi, 2014) and a juice obtained from the roots of V. amazonica is a source of natural black dye, used locally to color hair (Rosa-Osman et al., 2011). Medicinal uses of V. amazonica include wound treatment (Schultes, 1990), and V. cruziana has been recorded as an anti-inflammatory and a means for combating respiratory illnesses (Hurrell et al., 2016). The total cultural, spiritual and ethnobotanical knowledge of Victoria discovered by Indigenous Peoples is certainly more extensive, but poorly documented in the literature.

Whilst the floral morphology of Victoria (Figures 2, 3, 4–6 and Table 1) has been well described, there has been no consensus in the terminology applied. The flowers have an inferior ovary, comprising 25–40 radially arranged syncarpous carpels (Figure 2A). The upper part of the ovary is a concave, papillose stigmatic surface, which is divided into raised segments (Figure 2B). Each segment corresponds to the roof of the locule below and bears a dorsal longitudinal slit through which pollen reaches the locules. Each locule contains 8–28 ovules, attached parietally to both sides of the locule’s walls (Figure 2C). At the center of the stigmatic surface is a column of residual stelar tissue, also known as the floral apex (Schneider, 1976; Figure 2D). There has been some inconsistency in the description of the perianth segments. Following Warner et al. (2008), we are referring to all perianth segments as tepals. From the apex of the ovary’s external rim, four rigid fleshy outer tepals arise (Figure 2E; Warner et al., 2008). From the tepals, helically arranged series of petaloid inner tepals follow (Figure 2F). Adjacent to these, and moving inwards, rigid, thick, outer staminodia arise, which are presented in a whorl-like arrangement of one or two whorls (Figure 2G). In bud and on the first night opening, the outer staminodia (along with the stamens and inner staminodia which follow) are sigmoid, and arch strongly over the stigmatic surface (Figures 3B,E,H). After these are sigmoid stamens (Figure 2H), borne in two or three whorls, pressed tightly against the outer staminodia in bud and during the first night of opening. The stamens are followed by sigmoid inner staminodia in one or two whorls (Figure 2I). The inner staminodia (previously classed as paracarpels: Prance and Arias, 1975) are partially adnate to the upper portion of the carpellary appendages, which lie beneath them. These L- shaped carpellary appendages (Figure 2J) are aligned with the locules below, and correspond with them in number. The lower portions of the carpellary appendages are abaxially fused to the rim-like extension of the stigmatic surface which is also fused with the basal tissue of the inner tepal bases.

The cavity enclosed by these parts is referred to as the stigmatic chamber (Figure 2B). In bud and during the first night of flowering, the apices of the outer staminodia, stamens and inner staminodia are pressed tightly into each other to form an entrance tunnel (Figures 3B,E,H). This tunnel remains intact in bud and during the first night of opening, providing entry to the beetle pollinator, but is absent on the second night of opening, when all the outer staminodia, and most of the stamens, reflex to varying degrees (Figure 2). The inner staminodia, however, do not reflex, but instead fall further downwards, thus blocking the entrance to the stigmatic chamber. Flower size, the number of locules and the number of all parts (apart from the outer tepals, which always number four) are variable both between individual plants and on different flowers produced by the same plant.

Victoria flower buds develop underwater and emerge above the surface when ready to bloom. Each flower opens over two consecutive nights, changing form and color dramatically in- between. These form and color changes reflect their role in pollination, which is to trap pollinating beetles (Schomburgk, 1837; Prance and Arias, 1975) of the Cyclocephalini tribe (Scarabaeidae) (Prance and Arias, 1975). Floral morphology of the fossil Microvictoria, suggests that this mode of pollination was already established in the Cretaceous (Gandolfo et al., 2004). The carpellary appendages are believed to produce scent attractants and nutritional rewards (Prance and Arias, 1975; Zini et al., 2019) and with the warming of the flower (thermogenesis) at night (Planchon, 1850, 1851; Knoch, 1899; Decker, 1936; Prance and Arias, 1975; Lamprecht et al., 2002) serve as both an attractant and stimulant for the pollinators (Seymour and Mathews, 2006). The stigmatic surface remains receptive for the two nights during which the flower blooms (Archangeli, 1908). Despite the above investment in cross-pollination, Prance and Arias (1975) demonstrated that self-pollinated flowers were still capable of setting seed. Furthermore, seed produced in cultivation as a result of selfing is viable (Magdalena, personal observation). There have been few published pollination studies of V. cruziana and only one record of the putative new species from the Mamoré river basin in Bolivia (Magdalena, personal observation). After pollination, the fruit forms below the water surface (Prance and Arias, 1975). The seeds are covered by a mucilaginous tissue that has been proposed to represent an aril, are buoyant for a few days and released as the fruit decomposes (Prance and Arias, 1975). Prance and Arias (1975) suggest that seeds of V. amazonica, produced at the end of the wet season, are dispersed over long distances because of the annual flooding of much of its habitat and this may also be the case for V. cruziana. Although there is no evidence of endochory in the genus, this should not be excluded as the dispersal biology of the species remains very poorly studied.

Victoria occurs in white-water and occasionally black-water (Byrne, 2008) leas and igapos of the Amazonian and Paraná river basins (Rosa-Osman et al., 2011) at depths of up to 5.25 m (V. amazonica, Prance and Arias, 1975). Victoria spp. are commonly classified as annuals but Prance and Arias (1975) propose that, in the case of V. amazonica at least, this life cycle likely reflects a constraint posed by the dramatic changes in water level associated with flooding and drought, which characterize the wet and dry seasons across the range of this species. In a stable horticultural environment, V. amazonica and Victoria ‘Longwood’ hybrids can thrive for several years. They should, therefore, be considered to be short-lived perennial species. The seeds of V. amazonica are desiccation intolerant (Rosa-Osman et al., 2011). Seedlings develop very quickly in the river mud forming mature plants in three to five months. Development is quicker in Victoria cruziana compared to V. amazonica (Kit Knotts et al., personal observation). This is possibly a reflection of the shorter and more predictable growing season in the temperate biome of V. cruziana. Senescence is triggered by the detachment of the rhizome from the riverbed, or desiccation as the river level drops (Aniśko, 2014).

Taxonomic treatment of Victoria has been hampered by several factors. Foremost is the fact that the type collections of the two currently recognized species (World Checklist of Vascular Plants, 2022) have been lost or destroyed, making it challenging to unambiguously name material and thus delimit species. Poeppig’s collection(s) at Naturhistorisches Museum Wien and University of Leipzig were likely destroyed during WW2 and d’Orbigny’s s spirit collection disappeared from the Paris museum for reasons unknown. In addition, d’Orbigny diagnosed V. cruziana against material of Victoria morphotype ‘boliviana’ that he had mistakenly considered to be V. amazonica and in doing so established misleading species-limits at a time when very few collections were available for study. The above issues were compounded by the fact that Victoria is notoriously difficult to make herbarium specimens from, being big, fleshy, covered in prickles and prone to rotting in the dryer. This likely explains why there are relatively few herbarium specimens of wild collected material: only 97 of V. amazonica and 18 of V. cruziana (The Global Biodiversity Information Facility [GBIF], 2022), despite the species’ broad distributions.

These reasons could explain why the genus attracted relatively little taxonomic attention in the 20th Century, during which the most notable contributions were made by Malme (1907) and Prance and Arias (1975). In the 21st Century, renewed interest was, associated with research for the Flora of Brazil. Notably Pellegrini’s treatment for the Flora of Brazil (2020) that recognizes the two species, and de Lima et al. (2021), which recognize a single variable species (V. amazonica).

Significantly, much of our knowledge regarding the natural history of Victoria has been obtained from material growing in cultivation and compiled by horticulturalists in the additional literature or worldwide web¹ and it was observations by C. Magdalena which first suggested the existence of additional species and the need for a re-evaluation of the genus.

The aims of this study were to (1) revisit species delimitation in Victoria and to do so through an iterative process, using morphological observations to establish species hypotheses and suites of observations to test these, and in doing so (2) reveal the principle evolutionary lineages, their age, permeability and biogeography, (3) identify diagnostic morphological and DNA sequence characters for the species, and use the above to (4) revise the nomenclature, descriptions, distribution, and conservation status of the species.

We scored morphological character states from 58 herbarium collection specimens and 175 iNaturalist observations, giving a total of 233 individuals (see Supplementary Tables 4, 5). The congruence of morphological character states to our four morpho species was evaluated by eye, resulting in diagnostic characters being selected (see Table 2 and Figures 4, 5, 7, Key to the species): leaf rim morphology (shape in cross-section, height, color), flower bud apex shape, size and shape of the stigmatic chamber, outer tepal prickle morphology, number and distribution, inner tepal color in bud, carpellary appendage morphology and attachment, seed shape, size and presence / absence of a raphe. These were represented by 25 characters, 13 of which are novel.

We recorded 209 geographical observations assignable to morphospecies (Figure 3 and Supplementary Table 2), 175 of these correspond to iNaturalist records (indicated by * in Supplementary Table 2), and 44 to biological collections in herbaria. In several cases, iNaturalist records extended the distribution of Victoria delimited by herbarium collections (Figure 3). For example, in the case of the cruziana morphotype, herbarium collections indicate a southernmost limit of the –30.37°S latitude, whilst iNaturalist increased this to –32.92°S. Social media posts provided additional probable distributions for Victoria not documented in herbaria or iNaturalist. For example, images of V. amazonica from Irinida in Colombia, close to the Irinida and Guaviare rivers, both of which drain into the Orinoco river; and images of V. cruziana from the Esteras del Ibera, Argentina.

Our review of herbarium and iNaturalist records (Supplementary Table 2 and Figure 8) and of social media images suggest that Victoria is absent from a number of river systems that form part of the Amazon, Essequibo and Paraná river basins (Figure 8 and Supplementary Table 3). The most notable of these being its absence from central eastern and north western Amazonian Brazil and, despite being documented in Colombian tributaries of the Orinoco river, its apparent absence from Venezuela and Ecuador.

Using the geographical observations (Supplementary Table 1) we mapped our four morphospecies across South America (Figure 8). This shows strong congruence between morphospecies and geographical location. Based on our dataset, amazonica and ‘boliviana’ morphotypes are restricted to mutually exclusive portions of the Amazon river basin, the ‘mattogrossensis taxon incertum’ morphotype is restricted to the Pantanal (Paraná river basins), and cruziana is restricted to the lower portion of the Paraguay river and Paraná. Despite the presence of both cruziana and the ‘mattogrossensis taxon incertum’ morphotype on the Paraguay river, occurrence records suggest a large geographical separation between both.

Victoria is largely restricted to temperate and tropical Southern Hemisphere South America, not occurring further than 4.2°N and 32.9°S. Whilst it has an extensive temperate distribution in the southern hemisphere, it does not occur in the temperate Northern Hemisphere outside of cultivation.

V. amazonica is known from Colombia, Brazil, Guyana, Peru, and Bolivia where it is restricted to the river basins of the Amazon, Guaviare river (a tributary of the Orinoco) and Essequibo rivers. In the Amazon river basin V. amazonica occurs in most of the major tributaries except for the Xingu and Madeira rivers. V. cruziana is the only temperate species and is restricted to the Paraná River and its tributaries. The V. ‘boliviana’ morphotype appears to be endemic to the Llanos de Moxos in Bolivia with all the records concentrated in the Mamoré river basin. A photographic record from Rurrenabaque, however, suggests that it may also be present in the neighboring Beni river.

Through the application of a heuristic multidisciplinary approach, we provide revised species delimitations and diagnoses for Victoria. This has resulted in the recognition of a species new to science: V. boliviana Magdalena and L. T. Smith and highlighted the morphological distinctness of V. cruziana forma mattogrossensis taxon incertum. Species delimitation and nomenclatural stability in Victoria have until now been hampered by the loss of the original type material that served to fix the species name as well as a paucity of biological collections. This resulted in disagreement over the number of species recognized (Pellegrini, 2020; de Lima et al., 2021), the application of an incorrect name for Victoria amazonica for most of the 19th and 20th centuries (Prance, 1974) and a failure to recognize taxa in this iconic genus. Nevertheless, the recent selection of a neotype for V. amazonica and a lectotype for V. cruziana (de Lima et al., 2021) has helped to underpin nomenclatural stability and anchor species delimitation with respect to morphology. Here, we sought to delimit Victoria species, through the development of a heuristic and iterative approach – one which integrates field, horticultural, morphological and genomic observations and analyses. In the primary iteration of this investigation, we used geographical observations to integrate morphological observations from biological (herbarium) collections with those from field observations (iNaturalist, horticultural observations). In doing so, we were able to recognize discrete morphological units within a heuristic species concept that focuses on cohesion rather than divergence. These formed prima facie null hypotheses that we tested using genomic observations. Based on this approach, we recognize three discrete units at the rank of species, and provide strong justification for further research into a fourth taxon of unknown rank. Our morphological species delimitation differs from that of de Lima et al. (2021) who propose considering all populations of Victoria as a single species, V. amazonica. Their conclusion was based on the interpretation of the resulting large variation in the morphological characters observed in their single species as being the product of its broad spatial distribution and aquatic habit.

Past studies (e.g., de Lima et al., 2021) have been limited by the small number of herbarium collections available to study and their state of preservation, both of which are likely a product of the large size and fleshy nature of the plants. We overcame this through the use of high-resolution specimen scans available online, supplemented by “research quality” geo-referenced iNaturalist field images and observations of horticultural material. Using iNaturalist and digitized herbarium specimens we were able to incorporate 233 collections of Victoria, a significant increase on previous studies based on herbarium specimens alone (de Lima et al., 2021). iNaturalist images and observations allowed for characters usually lost in herbarium such as leaf rim morphology, the color of leaf blades, flower bud apex shape, the distribution and shape of prickles on the outer tepals, and the color and shape of tepals (Table 1).

The geo-location of Victoria lily pads enabled us to use these records to undertake assessments of extinction threat and so greatly increase the accuracy of those assessments. Whilst we did not use social media accounts as a source of georeferenced localities, we did use them to identify potential gaps in our knowledge of Victoria’s distribution. Instagram, which is image-based, was particularly useful in indicating the presence of Victoria cruziana in the Esteras del Ibera wetlands (Argentina), suggesting that it occupies a broader swathe of south eastern South America. Instagram also suggested the presence of V. amazonica in the Orinoco, and the cultivation of V. cruziana f. mattogrossensis taxon incertum under the names V. amazonica or V. regia in Brazil.

We identified the morphology of the leaf rim, flower prickles, the stigmatic chamber, carpellary appendage shape and size, and the seeds, as phylogenetically informative diagnostic characters in Victoria. Of these, we are the first to propose stigmatic chamber, carpellary appendage and seed morphology. This was surprising as these characters are all readily observable and prominent features that would be obvious to an experienced observer. A possible explanation may be that there was confusion over species delimitation stemming from d’Orbigny’s (1840) mistaken diagnosis of V. cruziana against V. boliviana sp. nov., and not V. amazonica, as he and subsequent authors supposed. In the absence of type material or an independent class of observations this would have been difficult to resolve.

The prickles on the abaxial surface of the outer tepals have not been proposed as diagnostic in Victoria since Malme (1907). Based on our observations, Victoria amazonica always has prickles which taper smoothly to a point at their apex (Figure 4) and are distributed relatively evenly over the entire tepal surface. By contrast, in V. cruziana (Figure 6) the prickles are absent or relatively sparsely distributed, but, where present, they taper abruptly to a point and are distributed over only the basal third of the tepal, whilst in V. boliviana sp. nov. (Figure 5) prickles are usually absent but, if present, may be found anywhere on the tepal surface and taper abruptly to a point. We speculate that prickles play a defensive role and protect the relatively nutrient rich contents of the bud and the protein-rich beetles trapped within at anthesis but are unable to account for the variation between species in the genus given the incomplete and small number of studies focusing on Victoria biology or autecology.

We also discovered that the size and shape of the stigmatic chamber, and of the carpellary appendages which surround and heat it, differs between species (Table 2). As above this suggests a link to pollination but again, given that the pollination-biology of Victoria cruziana and V. boliviana sp. nov. is very poorly known, it is only possible to speculate that the size and shape of the stigmatic chamber may be responding to differences in pollinator type, size or number, whilst differences in the size and disposition of the carpellary appendages may be products of the need to accommodate a different size of stigmatic chamber, or to produce greater or lesser amounts of heat in relation to ambient temperatures or pollinator preferences.

We found observable differences between the seeds of all three species with respect to their shape and size, and of the prominence of the raphe. Victoria amazonica has ellipsoid seeds compared to the globose seeds of V. cruziana and V. boliviana sp. nov. V. boliviana sp. nov. has relatively large seeds with a prominent raphe, compared to V. cruziana and V. amazonica. The significance of seed shape is unclear but may be related to dispersal through the gut of an unknown disperser, an ellipsoid seed being easier to pass than a globose one. Whilst no evidence of endochory has been found, given the lack of research it should not be excluded. Seed size has also been associated with the establishment depth of Nymphaea (Jacobs and Hellquist, 2011) suggesting that smaller-seeded species establish in shallower water. This would concur with the observations of previous authors (Prance and Arias, 1975; Rosa-Osman et al., 2011, Magdalena, personal observation) and suggest that V. boliviana sp. nov. establishes itself at greater water depths than V. amazonica and V. cruziana. Finally, similarly to V. cruziana, V. boliviana sp. nov. develops more rapidly than V. amazonica (Magdalena, personal observation.). As above, these aspects remain to be further studied.

We explored complementary concepts of genomic distinctness (in the form of genetic clusters) and divergence of evolutionary lineages in order to test whether the identified morphotypes were corroborated by molecular evidence. At this shallow taxonomic scale, a PCA has more power to highlight distinctness in nuclear genomic data than phylogenomic methods; the latter can be confounded by ILS (Lissambou et al., 2019). Even with our genome skimming approach, given the size of the Victoria nuclear genome [1C of 4.66 (V. amazonica), 4.10 (V. cruziana) and 4.24 (V. boliviana sp. nov.)], which is at least double the size of most Nymphaeales species profiled using flow cytometry (Pellicer et al., 2013), the capacity to retrieve appropriate nuclear genes at high coverage for phylogenomic inference was limited. Our response was to retrieve genotypes by mapping to curated transcriptomic data and to summarize this genetic variation using a dimensionality-reduction method. The tight respective clusters of V. cruziana and V. boliviana sp. nov. on the PCA and their degree of separation along the 2nd axis of variation serves to demonstrate their genetic distinctness. Importantly, this distinctness could be due to geographical isolation alone, which is why such results must always be assessed against other lines of evidence, such as morphological differences. Our PCA additionally demonstrates a greater degree of genetic structuring and variation within V. amazonica compared to V. boliviana sp. nov. or V. cruziana. This is concordant with the broader geographical spread of V. amazonica in northern South America, though we also note the larger sample size of V. amazonica available for this analysis. Furthermore, we would expect to see more continuity of the V. amazonica cluster on the 2nd PC, linking the two samples from Guyana (NPNY6 and NPNY14) and the samples from Brazil, had we been able to genotype a more broadly geographically sampled set of accessions. Implementation of a population genetic framework using high throughput sequencing (HTS) datasets to support molecular-based species delimitation is not a widespread practice, but is gaining some traction in studies of plants. For example, where they occur in sympatry (Ikabanga et al., 2017) or where taxonomic incongruity is prevalent within the genus (Rodríguez-Rodríguez et al., 2018; Rutherford et al., 2018). Our study demonstrates a novel way to apply this approach – in the absence of a reference genome, but utilizing an available transcriptome.

The dataset of mapped full plastid genomes suggests a similar conclusion in terms of delimiting separate evolutionary units of Victoria. This is presented as a strongly supported monophyly of respective clades containing species amazonica, boliviana sp. nov. and cruziana. The longer branch lengths of samples within the amazonica clade also suggests ancient differentiation. Our molecular dating suggests that plastid populations in Victoria diverged ∼5 Ma, with the time of divergence of V. boliviana sp. nov. and V. cruziana set to have occurred as recently as 1.1 Ma.

An unresolved question is the degree to which hybridisation and introgression might have been involved in the evolution of V. boliviana sp. nov. By revealing that this species has a chromosome count identical to that of V. cruziana, a hybrid origin cannot be easily supported. Even though the genome size of V. boliviana sp. nov is intermediate between that of V. cruziana and V. amazonica, additional molecular processes such as repeat amplification and chromosome rearrangements have almost certainly been involved its genome evolution, especially given its divergence time from V. cruziana.

Finally, we supplement these lines of evidence derived from different cellular compartments with genomic features identified in the assembled chloroplasts that we propose to be diagnostic to the new species. The absence of large-scale genomic rearrangements in our plastid genomes was not surprising given that gene order in the chloroplasts of Nymphaeales has been found to be conserved (Gruenstaeudl et al., 2017). The three indels unique to V. boliviana sp. nov., could be used to support the molecular identification of Victoria specimens. The longest indel was found in the ycf1 gene (105 bp in V. amazonica and 42 bp in V. boliviana sp. nov.). Ycf1 is a large housekeeping gene (Drescher et al., 2000) involved in photosystem biogenesis (Yang et al., 2016). Due to its high variability (Dong et al., 2015), ycf1 has been highly utilized in phylogenetics (Neubig et al., 2009; Dastpak et al., 2018). Here, we shed light on the utility of this gene as a potential tool for DNA barcoding at shallow phylogenetic scales. One application could be in genome skimming studies, where retrieval of full chloroplast genomes is routine. A low-cost alternative is a simple PCR of this genomic region, where, due to length differences, gel electrophoresis could be used for species identification. Diagnoses of new taxa that incorporate DNA-based characters are not common, but can be more useful than lineage-based diagnoses, especially when applied to known or type specimens. They are also an unbiased means of species delimitation (Renner, 2016). A small number of previous studies have used diagnostic molecular characters e.g., nucleotides at certain positions of matK and trnL-trnF regions in Buxus spp. (Buxaceae; Gutiérrez et al., 2011) and of nhdF and the ITS region in Brunfelsia (Solanaceae; Filipowicz and Renner, 2012). Length variation associated with indels has been previously applied as a form of diagnostic genetic variation – for example to the study of Abies (Xiang et al., 2018). Our approach has extended this to examination of the whole plastome.

Ensuring that species of Victoria remain for future generations requires that risks of extinction can be accurately evaluated and monitored. In the case of Victoria this requires greater knowledge of the species’ natural history and autecology, specifically their pollination and dispersal biology – against which potential threats can be evaluated – and the extent and fluctuation of population sizes.

This is important as it enables threats to the viability of populations to be evaluated. Knowledge of both is at best superficial and based on a small number of field observations of V. amazonica and V. cruziana (Schomburgk, 1837; Archangeli, 1908; Prance and Arias, 1975; Hanagarth, 1993). The dispersal biology of Victoria is poorly known and based largely on speculation rather than observation of tested hypotheses. There is also no literature on the size, fluctuation and connectivity of populations of the three species. A viable approach for doing so would be to use publicly available time-stamped remote-sensed data, such as Google Earth image layers to monitor populations given that the lily pads can be seen in higher resolution images.

Our small molecular sample set recovered genetic structure within V. amazonica (Figure 8) and we would argue for greater sampling of the genus, especially for the edges of its Amazonian range (Colombia, Guyana, Peru, Venezuela) and in the vicinity of the Pantanal. Additionally, since morphological observations suggest a fourth species (V. c. f. mattogrossensis taxon incertum), strategic sampling for genomic work may result in the molecular support for it at the rank of species.

Establishing the status of V. c. f. mattogrossensis taxon incertum should be seen as a high priority. Should it be evaluated as a distinct species, it would be one of the most vulnerable to extinction, having the smallest range and occupying a region that has been impacted by extreme drought during the last decade (Marengo et al., 2021).

Finally, a larger sample set would allow for investigation of the barriers to dispersal within the genus and extent of gene flow between the different populations of Victoria species.

Understanding the latter would also require the generation of more extensive nuclear molecular datasets and a more contiguous genome of reference. Construction of a nuclear phylogenomic framework would enable the computation of introgression tests based on patterns of allele sharing between taxa (e.g., D statistics; Durand et al., 2011) and permit a more nuanced investigation of the evolutionary history of these aquatic plant species, including investigation of forces driving speciation of V. boliviana sp. nov.

The above research would require improved sampling of the species’ distributions, underpinned by verifiable biological (herbarium) collections. Combined with the need for increased natural history, ecological and genetic observations we would propose that the genus be the focus of a dedicated field campaign.

Victoria also has great potential to serve as a valuable model for exploring the biogeography of continental South America as it is an aquatic species mostly restricted to large river systems; its seeds are desiccation intolerant and thus unable to escape flooding plains of their water catchment. The ephemeral nature of such flooding plains (Cowgill and Prance, 1989) is thought to have driven gigantism, as a mode of outcompeting other aquatic plants (Box et al., 2022). Victoria could feasibly be part of a monophyletic clade comprising Microvictoria-Euryale-Victoria, as is suggested by morphological observations (Gandolfo et al., 2004), the age of which predates the complete break up of Gondwana ca 83 Ma (Seton et al., 2012). The current distribution of genus Victoria spans a vast area of river systems, representing ca 44% of the South American drainage basins (ca 7.8 × 10⁶ km²) in a region where orogenesis and changing climates have wrought major changes in the last 12 Ma (Antonelli et al., 2009; Figueiredo et al., 2009; Hoorn et al., 2010, 2022).

Victoria R. H. Schomb., Athenaeum (London) 1837 (No. 515): 661 (September 9 1837).

Victoria Lindl., Monog. 3 (October 16 1837).

Victoria J. E. Gray, Mag. Zool. Bot. 2(10): 373 (December 1 1837).

Aquatic perennial herb, rhizome erect, tuberous, elongate to cylindrical, roots adventitious. Leaves floating, orbicular, peltate, perforated by stomatodes, adaxial surface of lamina glabrous, lacking prickles, green; abaxial surface of lamina with prominent radial and reticulate ribs, juvenile leaves sagittate; leaf margins flat or upturned; prickles covering petiole and ribs. Inflorescences uniflorate, bracteate. Flowers axillary, solitary, multiple buds per plant; pedicel with 4 primary air chambers, 8 minor chambers, covered in prickles; flowers opening one at a time, projecting above water surface shortly before anthesis, projecting above or resting on water surface at anthesis, each flower opening over two nights and partially closing in between, protogynous. Epigynous, ovary globose, covered in prickles externally, ovules parietal, attached by short funiculi, globose. Outer tepals 4, triangular, apex acute to rounded. Inner tepals 40–c.100, arranged in spiral series, creating a torus (attachment point of tepals forming a ring of tissue), tepals gradually reducing in size towards the center and changing shape from apically rounded to acute from outer to innermost; outer staminodia in 1 or 2 whorls, thick, rigid, apiculate; stamens > 100, borne in c. 3 series, subulate, introrse; anthers linear-elongate; inner staminodia, >50, sigmoid, subulate, partially adnate to carpellary appendages, detaching at second-night anthesis; carpellary appendages L-shaped, arising from extension of stigmatic surface, lower parts adnate to tissue extending from tepal base attachment, corresponding in position and number with stigmatic surface ridges and locules. Fruit ripening just below surface of water, 10–15 cm in diameter (excluding prickles) at maturity, fleshy, oblate, topped by a shallow cylinder-shaped mass of dark reddish to maroon ring of persistent hard tissue formed by the remnant bases of tepals; inner staminodia persistent and curved over concave stigmatic surface while ripening; outer layers of pericarp disintegrating to release seeds. Seeds smooth, surrounded by a mucilaginous aril.

Three, possibly four species, tropical and temperate South America.

Victoria amazonica (Poepp.) Klotzsch, Bot. Zeitung (Berlin) 5: 245 (1847). Euryale amazonica Poepp. Froriep’s Not. Natur- Heilk. 35: 131 (1832). Type: Poeppig s.n. (holotype W -presumed destroyed in WWII); Brazil, Amazonas, Careiro da Várzea [Teresina], Ilha de Careiro, 25 Sept. 1974, G.T. Prance 22745 (neotype (selected by de Lima et al., 2021): INPA (INPA46745); isoneotypes: K (K000837777!), NY (NY2269910, NY2269911, NY2269928), MO (MO3414212), US (US01341606)). Vernacular names: Forno de Jaçanã, Auapé yapóna, Victoria regia, Giant Amazonian Waterlily. Figures 1A, 2, 3A–C, 4, 9.

Victoria regina R.H.Schomb. Athenaeum (London) 515: 661 (September 9, 1837).

Victoria regia Lindl., Monograph: 3 (October 16, 1837). Type: Victoria Regia: 3, Plate 1 (October 16, 1837). nom. superfl.

Victoria regina J. E. Gray, Mag. Zool. and Bot. 2(11): 440 (December 1, 1837). nom. superfl. Victoria reginae Hook. Hooker’s J. Bot. Kew Gard. Misc. 2: 314 (1850). orth. var.

Leaves up to 2.3 m broad, adaxial surface of lamina green, occasionally tinged bronze in younger leaves; abaxial surface of lamina maroon or green, radial and reticulate ribs maroon, yellow or green; leaf margins form a low to moderate rim c. 4–7% of the lamina length (higher in crowded habitats), rim curved at its base then ± perpendicular to adaxial surface, abaxial surface of rim maroon or green; hairs 0.3–12 mm, simple, multicellular, 3–12 segmented. Flowers up to 28 cm in diameter at second-night anthesis. Ovary 8–12 cm diameter, outer surface covered in prickles 1– 18 mm (dried), prickles gradually tapering to a sharp point, hairs absent or present, where present simple, 0.1–0.4 mm, inner surface of ovary with deeply concave stigmatic surface, rounded to triangular in longitudinal profile, ridged with lines corresponding with 25–36 radially arranged locules, each containing 25–28 ovules, 1–1.5 mm diameter (fresh). Outer tepals 4.9–12 × 4–8 cm when fresh; abaxial surface predominantly brown/maroon, bearing 55–330 prickles per tepal, prickles tapering gradually to a sharp point, ranging from 1–14 mm (dried), spaced regularly, irregularly, or clustering more densely toward the base over entire surface, hairs absent or present on abaxial surface, where present 0.1–0.2 mm. Inner tepals 7–15 × 2–6 cm (fresh), innermost deep maroon in bud; all others remaining white or turning pink to dark pink at second-night anthesis; outer staminodia > 25, 5–6 × 1–1.5 cm thick, rigid, apiculate; stamens 2–4 × 0.5–1 cm; inner staminodia, 4–6 × 0.5–1 cm; base of lower parts of carpellary appendage auriculate/rounded in shape and hanging free from extension of stigmatic surface, length of upper parts not exceeding that of lower parts. Flower at first night of anthesis, inner tepals white, with innermost tepals dark maroon, outer staminodia tipped pink; second night anthesis, innermost tepals dark maroon, inner tepals remaining white or pink to dark pink or red, darkest at base, outer staminodia remaining white or dark pink for basal two thirds of their length, tipped pink, inner staminodia pink at base. Seeds 600–1000 per fruit, 7–8 × 9–10 mm, ellipsoid, green to brown, raphe faintly visible.

Distribution and Conservation Status —Victoria amazonica is restricted to the Amazon river basin, from Northern Brazil, Bolivia, Colombia, Guyana and Peru. Its EOO is estimated to be 2,640,795 km², exceeding the threshold for an IUCN threat category under criterion B, whilst its AOO is estimated as 476 km², falling into the Endangered category. We believe that our calculation of the AOO is likely an underestimate, resulting from difficulties in observing the species in the field and under-representation of the genus in biological collections.

There are more than 10 locations for which threats have been assessed, but there is not the information on population fragmentation or fluctuation available in order to be able to assess the ‘severely fragmented’ and ‘extreme fluctuations’ subcriteria. A continuing decline in habitat quality is inferred due to the presence of hydroelectric dams, mining, and deforestation of the river systems from where V. amazonica is documented. For example, in Peru, localities along the Marañón river to the headwaters of the Amazon and the Ucayali river have been heavily deforested by gold mining (i-Terra, 2021). Gold mining is associated with profound mercury contamination of rivers and aquatic species (USGS Environmental Health Program, 2019) and increases in Peruvian mercury imports suggest that contamination must be increasing (Swenson et al., 2011). In Brazil, as in Bolivia there have also been reports of mining activities in indigenous lands adjacent to several Victoria populations (Hutukara Associação Yanomami and Associação Wanasseduume Ye’kwana, 2020; INPE, 2021; MapBiomas, 2021; Mercado, 2021).

Victoria amazonica is here assessed as Least Concern (LC) considering that its range stretches across Amazonia and currently exceeds the parameters for a threatened category under criterion B. We note, however, a moderate number of locations where populations are under threat and there is a continuing decline in habitat quality. Further investigation and surveys are needed to better understand trends in population size, fragmentation, distribution and the impact of climate change.

Notes. Victoria amazonica is the only species whose leaves do not always form an upturned rim, and when they do, it is usually low and vertical in profile rather than recurving over the flat part of the lamina. Its flowers are distinguished from V. cruziana and V. boliviana both in bud and on first-night opening, as the innermost tepals are dark maroon rather than white. Carpellary appendages are curved at the base of the lower part and hang freely away from the attachment point. The prickles covering both outer tepal abaxial surface and outer ovary are uniquely gradually tapering to a sharp point (not abruptly tapering as in other species), and always cover the entire abaxial surface of the outer tepals. Its seeds are ellipsoid rather than globose.

Material Examined— BOLIVIA. Pando: Manuripi: pond north of Rio Madre de Dios., –66.126667, –10.903611, 09/07/1997, Ritter, N., Crow, G. & Crow, C. 4170 (LPB, MO). BRAZIL. ‘North Brasil’:

1898, Vaughan, G. 61 (K). Acre: Río Moa, margem esquerda; lugar chamado Humaita, –72.895556, – 7.614444, 01/10/1984, Ferreira, C. A. 5123 (NY). Amapá: Itaituba, Igarape no Rio Tapajos, – 55.961111, –4.229167, 16/12/2017, Brogim, R. 4 (UPCB). Amazonas: Ipixuna, Margem do Rio Croa, –72.556667, –7.745278, 15/02/2009, Quinet, A., Saraiva, B., Firmeza, T.1582 (K, SPF); Teresina, Ilha de Careiro, –59.81667, –3.1, 25/09/1974, Prance, G. T. 22745 (K, NY, US); Basin of Rio Purus area. Lago Preto, 3 km north of Labrea, –64.813056, –7.229772, 29/10/1968, Prance, G. T., Ramos, J. F. and Farias, L. G. 8016 (NY); Rio Solimões, south bank near Carreiro, –59.806667, –3.168889, 05/02/1974, Steward, W. C. And Ramos, J. F. P20211 (K, NY, US); Rio Amazonas, from Manaus to 100 km lower reaches, –59.141944, –3.216111, 08/08/1987, Tsugaru, S. and Yotaro Sano B-769 (MO, NY); Ilha do Cantagalo, –61.503889, –1.570833, 04/07/1995, Adalardo-Oliveira, A. 2645 (NY, SPF); Riverside and small islets of Rio Solimões within 100 km upper-stream from Manaus, –60.728056, 3.260278, 15/08/1987, Tsugaru, S. and Yotaro Sano B-1069 (NY); Pará: Santarem, Igarape, Ilha Grande de Santarum, –54.706840, –2.450070, –/10/1849 and –/11/1849, Spruce, R. 441 and Spruce s.n., s.d., (K, M, P), – /04/1850(NY) 1849 (P); Oriximina, Lago Uraria, SW of Orixima, across Rio Trombetas, – 55.9025, –1.812778, 11/06/1980, Davidson, C. and Martinelli, G. 10241 (MO, NY, RB, US); Rio Cupari, Lago Curuca, 01/01/1948, Black, G. A. 48-2223 (IAN); Rio Cupari, Lago Curuca, 02/01/1948, Black, G. A. 48-2253 (IAN); Rio Cupari, Lago de Curuca, 02/01/1948, Black, G. A. 48-2254 (IAN, NY, US); Oriximina, Rio Trombetas, Lago Ururia, 6 km SW de Oriximina, –55.910278, –1.803889, 08/06/1980, Martinelli, G. 6945 (RB); Pacoval, Rio Curua, –55.083333, –1.833333, 6–8/08/1981, Jangoux, J., and Riberio, B. G. S. 1647 (NY); Santarem, 25/12/1938, Markgraf 3873 (RB); Monte Alegre, Rio Gubatuba, proximo a vila Pare Sol, –54.043333, –2.011944, 17/07/2011, Lima, C. T. 503 (HUEFS); Pacoval, Rio Curua, –55.083333, –1.833333, 6–8/08/1981. Roraima: Rorainopolis, Rio Branco, Lago do Pirarucu, 25 km antes da boca com o Rio Negro, –61.8525, –1.158333, 28/03/2012, Martinelli, G., Moraes, M. A., Benevides, P., Forzza, R. C., Nadruz, M., Gallucci, S., Costa, D. 17700 (RB). COLOMBIA. Amazonas:Leticia, below Quebrada de Arara, –70.065278, –4.05944, 28/01 – 07/02/1969, Plowman, T., Lockwood, T., Kennedy, H., Schultes, R. E. 2313 (K). GUYANA. Berbice: Berbice, –58.2778, 4.394033, 1837, Schomburgk s.n. (K). Upper Takutu-Upper Essequibo: Karanambo, Rupununi River, –59.3, –3.75, 27/09/1988, Maas, P. J. M., Koek-N, J., Lall, H., ter Welle, B. J. H., Westra, L. Y. 7727 (K). PERU. Maynas: East of Puerto Alegria, –70.0625, –4.103611, 15/03/1977, Gentry, A. and Daly, D. 18351 (MO); Isla Padre (Cocha Paster), –76.166667, –3.75, 21/12/1982, Vasquez, R.., Grandez, C. and N. Jaramillo, N. 3684 (MO); Padre Isla in Río Amazonas, and in the cato below Iquitos., –73.163611, –3.651944, 22/05/1978, Gentry, A., Jarmillo N. 22133 (MO). See Supplementary Data.

Victoria boliviana Magdalena and L. T. Sm., sp. nov. Type: Bolivia, Beni Department, Provincia Ballivían, subiendo el Río Yacuma desde Puerto Espíritu, laguna en conexíon al Río Yacuma, unos 20 m al N, 29 Mar. 1988, S. G. Beck 15173 (holotype: LPB; isotype: K (K000798309). Vernacular names: Reina Victoria, Victoria regia. Figures 1B, 3D–F, 5, 10.

Most similar to V. cruziana Orb., from which it can be distinguished by the lower rim of the floating leaf, convex apex of the flower bud, length of the upper part of the carpellary appendages exceeding that of the lower part and the larger seeds. The V. boliviana plastid genome differs from that of other Victoria species by a 14 bp insertion between plastid genes ndhC and trnV in the large single copy region (LSC), a 5 bp deletion between trnK and rps16, a 7 bp deletion adjacent to trnC in the LSC and a 42b p deletion in the CDS of gene ycf1, within the SSC. Finally, a 4 bp transversion unique to V. boliviana sp. nov. was found in the LSC.

Leaves up to 3.2 m broad, adaxial surface of lamina green; abaxial surface of lamina dark green, maroon or dark-blue, radial and reticulate ribs yellow or green; leaf margins upturned to form a moderate rim c. 4–7% of leaf length, rim recurving strongly over blade surface at base and curving inwards or flared outwards at top, abaxial surface of rim deep maroon or very pale green/white in color, glabrous or with hairs, where present 1.2–3 mm, simple, multicellular, 6–15 segmented. Flower bud broadly ovoid, convex at apex, up to 36 cm in diameter at second-night anthesis. Ovary 8–10 cm diameter, outer surface covered in prickles, 1–10 mm (dried) glabrous; prickles abruptly tapering from c. half of length to sharp apex; inner surface of ovary with shallowly concave stigmatic surface, oblong in longitudinal profile, ridged with lines corresponding with 25–36 radially arranged locules, each containing 8–14 ovules, 2–2.5 mm diameter (fresh). Outer tepals 4, 10–15 × 8–10 cm when fresh, abaxial surface predominantly green, or tinged maroon, prickles absent or present, where present up to 10 per tepal, prickles tapering abruptly at their midpoint to a sharp point, 1–10 (dried), distributed irregularly over entire surface, glabrous. Inner tepals 6–15 × 1.5–9 mm (fresh), innermost remaining white or turning pale pink at their base at the second-night anthesis; outer staminodia > 50,3–4 × 0.5– cm, thick, rigid, apiculate; stamens, 4–5 × 0.5–1 cm; inner staminodia 4– 5 × 0.5–0.7 base of lower parts of carpellary appendage angular in shape and arising at 45 degree angle from stigmatic surface, length of upper parts exceeding that of lower parts. Flower at first night of anthesis, inner tepals white, outer staminodia tipped blue-violet; second night anthesis, inner tepal adaxial surface pink, inner tepals pale pink at base, white or pink towards the apex, outer staminodia dark pink for basal two-thirds of their length, white then violet towards the apex, inner staminodia pink at base. Seeds c. 300 per fruit, 12–13 × 16–17 mm, globose with a prominent raphe (especially when dry), dark brown to black, surrounded by a mucilaginous aril.

Distribution and Conservation status Victoria boliviana Magdalena and L. T. Sm. is restricted to Bolivia and the flood plains of the Llanos de Moxos, Mamoré watershed, identified as a Centre of Plant Diversity and Endemism (Site SA24) (Beck and Moraes, 1997). These area is considered by Langstroth Plotkin (2012). Moxos is surrounded by the forests of the Upper Madeira basin and is an area of largely open vegetation – herbaceous wetlands, grasslands, savannas, and woodlands (Beck, 1983, 1984; Langstroth Plotkin, 2012). Images (not included in the minimum calculation of the EOO or AOO) suggest that V. boliviana’s range extends further west (natural or cultivated) to Rurrenabaque. We estimated a minimum and maximum EOO and AOO. The maximum range was based on the potential habitat across the Llanos de Moxos region (including geographical information from public unverified images) and verified herbarium collections and iNaturalist images. The minimum range was based on the coordinates of herbarium collections and iNaturalist images alone. We estimated that the EOO of V. boliviana ranges between 8,006 km²(minimum) and 33,151 km² (maximum), falling close to the thresholds of the Vulnerable category under criterion B. We estimated that the AOO of V. boliviana ranges from 32 km² (minimum) to 2,000 km² (maximum), falling between the Endangered and Vulnerable categories. There are less than five known locations for V. boliviana. We could find no information about population fragmentation but believe that V. boliviana may be vulnerable to fluctuations in flooding and drought throughout the year. For example, Beni Department has been recently affected by seasonal floods, fires and droughts due to El Niño and La Niña climate events (Vásquez, 2015). A recent increase in agriculture-lead deforestation has been documented along the Trinidad-Santa Cruz highway (Langstroth Plotkin, 2012), to the south of known V. boliviana populations and satellite images from Google Earth Pro and i-Terra suggest extensive deforestation along the edges of roads (i-Terra, 2021; NASA, 2021; GoogleEarth, 2022) which we use to infer an active decline in habitat quality.

Taking a precautionary approach and based on the small EOO and AOO, small number of locations (5), and continuing decline in habitat, we assess V. boliviana as Vulnerable (VU), according to criteria B1ab(iii)+B2ab(iii).

Notes. Victoria boliviana sp. nov. has the largest observed leaves of the three species, with laminae > 3 m in length having been observed. The abaxial surface of the upturned rim surface varies between individual plants in the same locality from dark maroon to very pale, almost “white” green, a characteristic not found in other species. Prickles on abaxial outer tepal surfaces are absent or very few, and if present are not confined to the lower portion of the outer tepals unlike V. cruziana where the smaller number of prickles are confined to the lower one-third of the abaxial tepal surface.

Victoria boliviana is the only species of Victoria whose carpellary appendages have upper portions that are longer than the lower portions. In addition, V. boliviana’s stigmatic chambers are the shallowest of the three Victoria species. Haenke saw Victoria plants in Yacuma in 1801, during the Malaspina expedition (Gickelhorn, 1966, p. 105; Ibañez Montoya, 1984), but no identifiable description is available and no voucher has been found. d’Orbigny (1840, p. 57) reported seeing this species on the banks of the Mamoré river in 1832 and mistakenly assigned this species to V. amazonica when publishing his description of V. cruziana.

Further investigations and surveys are required to better understand the species’ current range, population fluctuations and habitat and thus better predict the impacts of the threats identified.

Field observations of the flowers from a single population in the Llanos de Moxos suggest that whilst pollinated by beetles, V. boliviana sp. nov. flowers may host fewer individuals of pollinators than V. amazonica, only 4–10 individuals being observed in the flowers of the former, compared to > 20 in the flowers of the latter. This could be due to a lower density of pollinators in their area of occupation.

Additional Material – BOLIVIA. Beni: Santa Ana del Yacuma, –65.4236, –13.74, –/6–7/1845, Bridges, T. s. n. (K); Cercado: Laguna Suarez 5 km sur de la ciudad de Trinidad, –64.864167, – 14.872222, 08/05/2019, Magdalena, C. Melgar, D. G., Salazar, C. D., Alvarez, C., Gutierrez, G., Arias, J. 154 (German Coimbra Sanz Herbarium, Jardin Botanico Municipal); Moxos, pasando el Río Mamoré, cerca al puente del Río Tijamuchi a lado del camino, –65.145278, –14.851111, 09/05/2019, Magdalena, C. Melgar, D. G., Salazar, C. D., Alvarez, C., Gutierrez, G., Arias, J.155 (Herbario German Coimbra Sanz Jardín Botanico Municipal).

Victoria cruziana Orb., Ann. Sci. Nat., Bot., sér. 2, 13: 57 (January 1840). Type: Bolivia [Argentina], Corrientes, banks of the Paraná river, Arroyo de San José, beginning of 1827, d’Orbigny s.n. (lectotype: P (P02048598*) (designated by de Lima et al., 2021); isolectotypes: P (P02048599).

Vernacular names: Irupé (yrupé), yacare yrupé, naanók lapotó (poncho del otany), maíz de agua, Santa Cruz Waterlily, Victoria regia. Figures 1C, 3G–I, 6, 11.

Victoria regia var. cruziana (Orb.) G. Lawson, Proc. & Trans. Roy. Soc. Canada 6(4): 109 (1889)

Euryale brasiliana Steud., Nomencl. Bot. [Steudel], ed. 2. 1: 617 (November 1840).

Euryale policantha Rojas Acosta, Cat. Hist. Nat. Corrientes 65 (1897).

Euryale bonplandia Rojas Acosta, Cat. Hist. Nat. Corrientes 151 (1897).

Victoria cruziana f. trickeri Henkel ex Malme, Acta Horti Berg. 4(5): 12. 1907 (“Trickeri”), nom. nud.

Leaves up 2.4 m broad, adaxial surface of lamina green, abaxial surface of lamina green or dark blue-green, radial and reticulate ribs yellow or green; leaf margins form a high rim 8–10% of leaf length, rim ± perpendicular to or slightly recurved over adaxial surface at base, flared outwards at top (sigmoid in profile), abaxial surface of rim green or tinged maroon, hairs 1–3 mm, simple, multicellular, 10–15 segmented. Flower bud broadly ovoid, concave just before apex, up to 30 cm diameter at second-night anthesis. Ovary 7–10 cm diameter, outer surface covered in prickles 1–22 mm (dried), prickles abruptly tapering from c. half their length to sharp apex; hairs absent or present, where present 0.1–12 mm; inner surface of ovary with moderately concave stigmatic surface, rounded to triangular in longitudinal profile, ridged with lines corresponding with 25–38 radially arranged locules, each containing 20–25 ovules 1.5–1.8 mm (fresh). Outer tepals 4, 10–13 × 4–9 cm when fresh, abaxial surface green and/or tinged maroon abaxially prickles absent or present, where present up to 100 per tepal, prickles tapering abruptly at their midpoint to a sharp point, 1–10 mm (dried), distributed up to lower one third of surface, hairs absent or present, where present 0.1–1 mm. Inner tepals 7–10 × 1.5–9 cm (fresh), innermost all white both in bud and during first-night anthesis, crinkled in appearance, turning pale to dark pink on second-night; outer staminodia, 6–7 × 1–1.5 cm, thick, rigid, apiculate; stamens 4–6 × 0.5–1 cm inner staminodia > 50, 4–6 × 0.5 cm; base of lower parts of carpellary appendage flat, arising from stigmatic surface at 45 degree angle, cuneate, length of upper parts not exceeding that of lower parts. Flower at first night of anthesis: all inner tepals white, outer staminodia tipped pink; at second night anthesis, outer tepal adaxial surface pink, inner tepals pale or dark pink at base, white or pink towards apex, outer staminodia dark pink for basal two-thirds of their length, white then pink towards apex, inner staminodia pink at base. Seeds, c. up to 1000 per fruit, 7–9 × 8–10 mm, globose, raphe faintly visible, brown to black, surrounded by a mucilaginous aril.

Victoria cruziana f. mattogrossensis taxon incertum has hitherto been included within Victoria cruziana until this study. Because our genomic analyses are limited with respect to the rank of this taxon and of its relationship to the other species, we have not considered forma mattogrossensis taxon incertum as conspecific with V. cruziana for the purposes of an extinction risk assessment.

Victoria cruziana is restricted to the Paraná river basin and tributaries, from Paraguay to Argentina, and possibly Bolivia. Based on the maximum potential habitat of wetlands (including Esteros de Ibera National Park Wetlands) and a combination of verified herbarium collections and iNaturalist images we estimate the EOO of V. cruziana to be between 46,563 and 132,945 km². This exceeds the threshold for a threatened category under criterion B (IUCN Standards and Petitions Committee, 2019). We calculate the AOO to be 120 km², although an upper estimate based on the extent of the river may exceed 2,000 km² (but not more than 3,000 km²). This would assess V. cruziana as between Endangered and Vulnerable categories. There are more than 10 locations but no information about population fragmentation. We infer a continuing decline in habitat quality due to the increasing frequency of droughts, the abstraction of water, deforestation (Caivano and Calatrava, 2021; Comisión Nacional de Actividades Espaciales, 2021) and big hydroelectric dams. For example, Itaipu is one of the largest dams in the world (Stevaux et al., 2009) and lies within the V. cruziana range. Whilst the AOO upper estimate is close to the threshold for VU, there are more than sufficient criteria for a threatened category under criterion B, and we therefore assess V. cruziana as LC.

We recommend further documentation of the distribution and size of V cruziana populations and investigations into their fluctuation through time as we believe that it may be vulnerable to an increase in the frequency and severity of droughts associated with climate change and increased sedimentation caused by the construction of large dams within its habitat (Stevaux et al., 2009).

Notes. Victoria cruziana forms the proportionately highest leaf rims of all the species, and these are always slightly recurved over the flat part of the lamina, flaring out at the top. The concavity of the outer tepals before their apex gives the bud a pinched-in appearance. Prickles are absent or occur on the outer tepal abaxial surface, but only up to one-third of their length from the base. In this species, hairs which are sometimes present on the lower outer tepal abaxial surface and ovary are the only ones large enough to see without magnification. At first night anthesis, inner tepals have a crinkled appearance.

Victoria cruziana f. mattogrossensis taxon incertum was described by Malme based on material present in the spirit collection of the Swedish Museum of Natural History comprising two jars of the same gathering. S07-84 comprises a longitudinal section through the ovary and perianth showing clearly the distribution of prickles and their form on both the ovary and outer tepals, also the morphology of the carpellary appendages; S07-85 comprises sections of the petiole apex and either the petiole or pedicel.

Both are preserved in excellent condition. S07-84 was selected as lectotype as it displayed a greater number of diagnostic morphological features.

Victoria cruziana f. mattogrossensis taxon incertum was described from, and material corresponding to it has only been observed from the Pantanal, in Bolivia, Brazil and Paraguay in the Uruguay river basin.

Phylogenomic data were unable to confirm whether the material sampled represents a distinct evolutionary lineage (see section “Discussion”). It may be that further sampling and research supports the recognition of this name as a distinct taxon.

Material Examined. ARGENTINA. Chaco: 1st De Mayo, Laguna en Chacra, al lado del arroyo Ine, –58.849444, –27.416389, 02/03/2006, Mulgura de Romero, M. E., Anisko, T., Harbage, J., Illarrage, H.4249 (SI); San Fernando, Entre Barranqueras e Isla Antequera, –58.87944, –27.416389, 18/03/1967, Krapovickas, A. and Cristobal, C. L. 12752 (MO). Corrientes: Capital, Riachuelo, off Ruta 12 ca 17 km S of Corrientes, –58.749444, –27.554444, 06/04/1982, Schinini, A. Wiersema, J. H. 2243 (MO); Esquina, Isla Correntina frente a curuzu – Chali, en el Paraná medio, –59.630833, –30.341944, 10/04/1968, Burkart, A., Troncoso, N. S., Guaglianone, E. R. and Palacios, R. A. 26963 (SI); San Roque, R. Santa Lucia, –58.738056, –28.576944, 27/02/1957, Pedersen, T. M. 4486 (K, MO); Bella Vista, Cruce Ruta Nacional 12 Ey Puente Sobre Lel Río Santa Lucia, –58.72, –28.57, 12/04/2008, Mulgura de Romero, M. E., Aniśko, T., Belgrana, M. J. and Harbage, J. 4474 (SI); Dep. Esquina, Río Guayguiraro, –59.56, –30.374444, 26/02/1974, Quarin, C., Schinini, A. Gonzales, J. M., Ishikawa, A. 2196 (K). Santa Fe: La Capital, Ruta Nacional 168 y Puente no. 9, Sobre Arroyo mini, al E-SE de la Ruta Prov. 1. W, –60.575833, –31.672778, 12/04/2008, Mulgura de Romero, M. E., Anisko, T., Belgrana, M. J. and Harbage, J. 4477 (SI). BOLIVIA. Santa Cruz: Angel Sandoval, channel at southern end of Laguna Mandiore, –57.483333, –18.216667, 16/07/1998, Ritter, N., Crow, G. E., Garvizu, M. and Crow, C. 4562 (LPB, MO) [f. mattogrossensis], Ritter, N., Crow, G. E., Garvizu, M. and Crow. C. 4560 (LPB, MO). BRAZIL. Mato Grosso du Sul: Corumba, Cacimba da Saude, – 57.664167, –18.99833, 13/12/2002, Avellar, A. L. F. 13 (COR) [f. Mattogrossensis]; Ladario, terminal da Branave, no porto de Ladario, –57.584722, –19.0225, 12/08/1994, Sanches, A. L., Bortolotto, J. M., Damascenos Jr., G. 44 (COR) [f. mattogrossensis]; Corumba, Ladario, CODRASA Brejo (swamp), – 57.516389, –19.021111, 27/11/2004, Souza Jr., A. F., and Siqueira, C. S. 39 (COR) [f. Mattogrossensis]; Corumba, Ladario, –57.580556, –19.001111, –/07/1894, Anon s.n. [G. O. A. Malme] S07-785, S07-784

(S). PARAGUAY. Distrito Capital: L’Assumption [in the swamps], –57.604167, –25.256667, 17/03/1875, Balansa, R. 523 (K, P). Central: Piquete Cue, –57.666667, –25.166667, 29/11/2000, Zardini, E. M. and Guerrero, L. 55181 (MO); Bay of Ascuncion, wetlands, –57.61922, –25.273333, 15/03/1988, Ericsson, K. 577 (MO). Neembucu: Yataity, –58.040556, –26.788611, –/03/1975, Walter, M. A. 86 (K); Pilar Garden Club S, –58.276389, –26.867778, 28/01/2005, De Egea Juvinel, J., Pena-Chocarro, M.., Vera, M., Torres, M.. and Elsam, R. 738 (MO). Presidente Hayes: Riacho Pucu., –57.083333, –24.666667, 20/08/2000, Zardini, E. M and Guerrero, L. 54765 (MO); 12/02/1987, Sparre 2363/51 (P).

Victoria cruziana f. mattogrossensis Malme, Acta Horti Berg. 4(5): 12. 1907. Type: [Brazil, Matto Grosso do Sul, near Corumba] Anon s.n.[G.O.A. Malme] 1894 [July] [S (spirit collection) (lectotype (selected here): S (S07-784*); isolectotype: S (S07-785*)]

Victoria argentina Burmeister, Reise durch die La Plata Staaten 2: 5 (1861).

Victoria fitzroyana Hort. Ex Loudon, Encyc. Pl. Suppl. Ii. 1388. = Nymphaea gigantea Hook.