The origin of Oxalis corniculata L.

Europe and North Africa

Europe has five extant species of Oxalis section Corniculatae, O. corniculata, O. exilis, O. stricta, O. dillenii and O. filiformis Kunth (syn. O. ferae L.Llorens, Gil & C.Cardona). Of these, O. exilis and O. filiformis are 20th century introductions with relatively narrow distributions (Llorens et al., 2005; Young, 1968). Owing to the presence of closely related species in Europe it is necessary to consider confusion of their identification in observations. Unless a complete specimen is available it is difficult to distinguish these taxa from one another, particularly if only fragments, such as seeds, are available. Taxonomic and nomenclatural confusion is a constant theme in the literature of this section (Eiten, 1955; Watson, 1989). Therefore, before investigating the situation of O. corniculata it is important to establish the time of introduction of O. stricta and O. dillenii to Europe.

It was not until 1680 that Morison described a species with an upright habit that we now refer to as O. stricta (Morison, 1680). However, this upright species was described from seed sent to him from Virginia and was not from Europe. By the mid-19th century the upright species (O. stricta) was so well established in Europe that Jordan (1854) considered that it must be a native species and described O. europaea Jord. He did so based purely on the geographic grounds that the Linnaean O. stricta was North American. He did not give any morphological characters that separate them. Herbarium specimens do exist of O. stricta collected in Europe before 1854, but these are few and only date from earlier in the 19th century (Table 1) (Lourteig, 1979).

O. dillenii was described and illustrated by Dillenius (1732) and was formally named later by Jacquin (1794) with direct reference to Dillenius. However, this was from material of North American origin. It was not until later that O. dillenii was recognised as naturalized in Europe. Perhaps the first record is that of Jordan (1854) who named O. navieri Jord., which is now considered a synonym for O. dillenii. Therefore, we agree with all modern authors that both O. stricta and O. dillenii are 19th century introductions and we can assume that earlier European observations of caulescent yellow-flowered Oxalis are of O. corniculata.

The earliest clear mention of a yellow-flowered caulescent Oxalis in European literature is from northern Italy in the 16th century (Turner, 1568). The earliest herbarium specimens are also from northern Italy, from Florence in 1600 and Bologna between 1551 and 1600 (Table 1). Watson (1989) reviewed the taxonomic situation of O. corniculata in Europe and cites Renaissance botanists of the 16th and 17th century from Belgium, France, Germany, the Netherlands, Spain and Switzerland. Therefore, there is good evidence, including accurate drawings, that O. corniculata occurred in Europe at least from the 16th century.

Earlier than this the information in literature is ambiguous. We have found six illustrations of Oxalis (Alleluia or Panis Cuculi in Latin) in medieval herbals from the 13th to 15th centuries. In each case only one taxa is illustrated. In Europe we would expect at least one taxon, the native O. acetosella L. This species is quite distinct from O. corniculata in having white flowers, short globular pods, a scaly creeping rhizome and no stem. The illustration in the 15th century Codex Bellunensis and an unnamed codex have no flowers, but depict the scaly rhizome and the lack of stem, which suggests this is O. acetosella (British Library, 2019a; Schoenberg Center, 2019). The illustration in Giovanni Cadamosto’s manuscript of the late 15th or early 16th century ambiguously shows a part caulescent and part acaulescent plant, (British Library, 2019b). The flowers of this illustration are in bud and are of little diagnostic use. A similarly ambiguous picture is in a codex dated 1475–1525, also from Italy (University of Vermont Libraries, 2019). Another herbal from circa 1440 shows a caulescent Oxalis with small yellow flowers in umbels. This illustration is close to the characters of O. corniculata and originates from Lombardy in northern Italy (British Library, 2019c). Finally, the oldest illustration is in the Tractatus de herbis (circa 1280–1350) (British Library, 2019d). This is a rather confusing illustration. It shows a horizontal creeping plant, rooting at the nodes, which could be O. acetosella or O. corniculata. The presumed flowers are small and erect, like O. corniculata, but solitary like those of O. acetosella. There are other four-pointed structures, which are presumably fruits. However, they do not resemble Oxalis fruits.

A considerable number of archaeobotanical studies have been conducted in Europe, but remains of Oxalis section Corniculatae remain rare. The RADAR database in the Netherlands contains four records, the earliest of which are from a site that spans the 17th and 18th centuries (Van Haaster & Brinkkemper, 1995). In Ferrara, Italy, remains of a single seed were found in a waste pit dating from the second half of the 15th century, which is notable because of the pre-Columbian dating (Bosi, 2000, Bosi et al., 2009). Although a single seed, its location is close to the specimen, illustration and literature reports from the 15th and 16th centuries, giving credence to this isolated observation. We have found no earlier archaeological evidence, even though some extensive studies have been conducted. Rinaldi, Mazzanti & Bosi (2013) summarize the seeds found in excavations in Modena, in northern Italy from the 2nd century BC to the 6th century AD. They list more than 400 taxa from 200,000 remains, including many weeds, but did not find O. corniculata. Likewise a similar study in Lleida, Spain, covering the period between the 2nd century BC and the 11th century AD report no O. corniculata, despite recovering almost 600,000 seeds and fruits.

We have found two other archaeological reports of O. corniculata in Europe, one from Neolithic Germany in the 6th century BC and the other from Visigothic Spain in the 6th-7th century AD (Herbig et al., 2013; Ollich et al., 2014). The reports are only of two and one seed, respectively, and there are no illustrations. Given the small number of seeds, the lack of supporting evidence and the potential for later contamination or misidentification, we have discounted these reports.

In all, we found no convincing evidence for the presence of O. corniculata in Europe earlier than the 15th century. Neither is there much information from North Africa, which has been in constant communication with southern Europe for many centuries. Preserved Oxalis seeds have been reported from late neolithic Hyrax middens in the Hoggar Mountains in the Sahara (Barakat, 1995). However, no illustration is provided and there is no way to verify this unlikely claim. A more reliable source is from excavations at ancient Carthage in modern day Tunisia (Van Zeist, Bottema & Van Der Veen, 2001). Here, preserved Oxalis seeds were reported from sediment from approximately the 4th century BC to the mid-6th century AD. However, there is again no illustration and we have been unable to access vouchers. The earliest direct observation from North Africa is from 1751 (Hasselquist, 1766). This was near Damietta, Egypt, in the Nile delta.

In conclusion, there is good evidence of O. corniculata being in Europe from the 16th century onward. Despite considerable taxonomic confusion, there is no evidence of O. stricta and O. dillenii occurring here much before the beginning of the 19th century, presumably as an introduction from North America, as is generally assumed. Earlier than this, despite a copious corpus of botanical, medicinal and archaeobotanical literature, the evidence is weak, leaving an open question about the long-term existence of O. corniculata in Europe. The observation of O. corniculata in Carthage may indicate that O. corniculata was present in the Mediterranean basin during the Iron Age, though more evidence than this would be preferable. Therefore there is little support for O. corniculata being native to Europe.

Sub-Saharan Africa

Today O. corniculata is a common weed in sub-Saharan Africa (Holm et al., 1991). However, Hooker & Bentham (1849) specifically noted its absence from the western intertropical coast, where it is present today (Hutchinson & Dalziel, 1954; Holm et al., 1991). The earliest sub-Saharan records are of Thunberg from near Cape Town, South Africa, circa 1772 (Table 1). Thunberg described its habitat as “Crescit prope urbem locis aquosis et in hortis” (Growing in wet ground near the city gardens) (Thunberg, 1823). Thunberg described O. repens Thunb. from these specimens, which has subsequently been synonymized with O. corniculata (Lourteig, 1979). The next specimens are from the 1830s from South Africa and Ethiopia (Table 1), but there are few records from sub-Saharan Africa until the 20th century and many records are of O. corniculata as an agricultural weed.

In Ethiopia Richard, Petit & Quartin-Dillon (1847) claimed three species, O. corniculata and two new species O. procumbens and O. radicosa. Lourteig (1979) accepted O. procumbens and O. radicosa as distinct species. Oxalis procumbens is said to grow in fields and fallow land at altitudes above 1,000 m in tropical Africa and O. radicosa has a larger distribution spanning much of Africa, Asia and Australasia. Examples of specimens are in Table 1. Many other authors, at least in Africa, have treated these species as synonyms of O. corniculata (Engler, 1894; Madagascar Catalogue, 2017). If one accepts that O. procumbens and O. radicosa are distinct species in Africa, then there is little reason to think that O. corniculata s.s. is native to sub-Saharan Africa. However, if these species are considered to be part of O. corniculata s.s. then it is perhaps native to the east coast of Africa and Madagascar and also to high altitude areas of tropical Africa. However, the lack of systematic collecting in Africa before the 20th century means that the indigenousness of specimens attributed to O. procumbens and O. radicosa can also not be confirmed.

The lack of data from much of Africa make it difficult to draw conclusions and this is further complicated by the ambiguity of the species concept of O. corniculata. The history of O. corniculata in sub-Saharan Africa will probably remain obscure without further evidence or discoveries.

Australasia

Australasia has several native species in section Corniculatae, but O. corniculata s.s. is considered introduced (Webb, Sykes & Garnock-Jones, 1988; Conn, Jeanes & Richards, 1999; Gray, 2011). Native taxa are O. novae-caledoniae R.Knuth & Schltr., O. exilis, O. rubens Haw, O. perennans Haw., O. chnoodes Lourteig, O. radicosa and O. thompsoniae B.J.Conn & P.G.Richards. Unlike authors of African Floras, authors of Australasian Floras accept O. radicosa as a distinct species. Australasian species live in a wide range of natural and anthropogenic habitats and although all are close in morphology to O. corniculata, all have distinct morphological features that separate them. Nevertheless, confusion of specimens and observations of O. corniculata with the native taxa must always be considered. Study of the Australasia situation is complicated because Eiten considered all Australasian members of the section Corniculatae to be part of O. corniculata s.l. and many specimens in herbaria retain his determinations (Eiten, 1963).

Of possible native species, the closest species to O. corniculata in Australasia is O. thompsoniae that has recently been described (Conn & Richards, 1994). It has a weedy habitat and is only known from collections made from the late 20th century. Conn & Richards (1994) suggest it may have been introduced to Australasia, but is of unknown origin. Likewise it may have evolved recently in Australasia from O. corniculata. Nevertheless, all reliable determinations of O. corniculata s.s. in Australasia date from the mid to late 20th century.

In summary, the prevailing consensus that O. corniculata is a recent introduction to Australasia seems to be consistent with the available evidence. However, the similarities of Australasian species with O. corniculata perhaps indicate a close evolutionary relationship.

North America

The Flora of North America lists several native members of Oxalis section Corniculatae. These are O. illinoensis Schwegman, O. macrantha (Trel.) Small, O. grandis Small, O. colorea (Small) Fedde, O. stricta, O. suksdorfii Trel., O. dillenii, O. albicans Kunth, O. pilosa Nutt. ex Torr. & A.Gray, O. californica (Abrams) R.Knuth and O. texana (Small) Fedde. Furthermore, in Mexico, Central America and the Caribbean there are other species, O. filiformis, O. rugeliana Urb. and O. thelyoxys Focke (Nesom, 2017).

Oxalis corniculata s.s. is considered an introduction to North America (Nesom, 2009). In Central America and the Caribbean the earliest herbarium specimens date from the 1820s (Lourteig, 1979). However, an illustration in Hernández (1651) has been interpreted as being O. corniculata (Pico & Nuez, 2000), although the illustrated specimen cannot be confidently distinguished from other North American taxa, such as O. albicans.

We found no convincing evidence that conflicts with the widely held view that O. corniculata has been introduced to North America.

South America

Several species in section Corniculatae have been reported in South America such as O. bisfracta Turcz., O. calachaccensis R.Knuth, O. corniculata, O. filiformis and O. sexenata Savigny (Eiten, 1963). Also, others, such as O. dumetorum Barnéoud and O. conorrhiza Jacq. can be included under a broader circumscription of section Corniculatae (Vaio et al., 2013). Vaio et al. (2013) used a molecular phylogeny to trace the evolutionary roots of section Corniculatae to South America. However, establishing whether O. corniculata s.s. is native to South America is complicated because of the other closely related species present on the continent. Where observations are not supported by specimens it is impossible to know if they have been correctly identified. The earliest specimens we have found of O. corniculata from South America (Brazil) are from 1815 to the 1830s (Table 1). However, in both cases the plants were either cultivated or weeds of cultivation.

Vaio et al. (2013) showed that O. corniculata is genetically close to the Bolivian species O. calachaccensis. In fact, O. calachaccensis does not possess any character that uniquely distinguishes it from O. corniculata (Eiten, 1963). O. calachaccensis was described by Knuth (1915) using material he collected in 1911 from La Paz, Bolivia and one is left to assume it is native to that location. However, as Vaio et al. (2013) indicate, the taxonomic status of O. calachaccensis is doubtful.

No earlier literature or archaeobotanical evidence has been found for O. corniculata s.s in South America. Even though O. corniculata has South American ancestry, there is no evidence for the species being in South America before the 19th century.

Asia

Oxalis corniculata is widespread in Asia and is an important weed in countries such as India and China (Reddy, 2008; Liu & Watson, 2008). The authors of Asian Floras, such as those from China and Japan, do not commit to whether O. corniculata is native or introduced, but they conclude that the widespread dispersal by humans has made its origin obscure (Ōi, Meyer & Walker, 1965; Liu & Watson, 2008).

Nevertheless, there are several archaeobotanical reports of O. corniculata seed from China. The earliest of these are from the Majiabang culture (Hunan, 7000–5800 BP; 16 seeds in two excavations) and Daxi culture (Jiangsu, 7,000–5,300 BP; 17 seeds) (Nasu et al., 2012; Qiu et al., 2016). Then are reports from Henan Province 4,950–4,450 BP where 74 seeds were found (Deng et al., 2015). There are also reports of multiple preserved seed from the Huangguashan period in Fujian province from more than 3,000 BP; from the Bronze Age in Yunnan province, and from during the Tang dynasty in Lantau (Atha, 2012; Yao et al., 2015; Deng et al., 2017). The paper of Nasu et al. (2012) has a particularly convincing illustration of a seed. At all of these sites there was evidence of early agriculture where O. corniculata may have been a weed.

Apart from O. corniculata the only other possibility is that these are seeds of O. stricta, which is believed to be native to China (Liu & Watson, 2008). However, this Flora of China describes O. stricta as growing in “Forests, ravines; 400–1,500 m” in Guangxi, Hebei, Henan, Hubei, Jiangxi, Jilin, Liaoning, Shanxi, Zhejiang. This is not consistent with the presence of this Oxalis in either Hunan, Fujian, Yunnan or Jiangsu province as an agricultural weed.

Also in Asia, Oxalis seeds have been found in Japan from the Early Yayoi period (ca. 2,820–2,530 BP); in the Philippines in the 1st millennium BC and in Vietnam between 3,450 and 3,250 BP (Paz, 2005; Nasu & Momohara, 2016; Castillo et al., 2018). Castillo et al. (2018) provide a photograph of the single seed they found. Nasu & Momohara (2016) indicate that they found between 11 and 58 seeds across two different excavations. O. stricta has neither been found in the Philippines nor Vietnam, so it seems reasonable to assume these are seeds of O. corniculata.

There is early documentary evidence of the presence of O. corniculata in China from 1751 (Osbeck, Torén & Ekeberg, 1771). This is from Guangdong, outside the known range of O. stricta in China and in a region where O. corniculata is common today. There is also a specimen from Osbeck in the Bergius Herbarium from 1751 and this has been identified by Lourteig as O. corniculata (Table 1). The distribution of first records illustrated in Fig. 1 demonstrates the cluster of early records from south–east Asia.

Oxalis corniculata is also the main food plant of the larvae of Pseudozizeeria maha Kollar, 1844 (pale grass blue butterfly) (Saji et al., 2017). This species is found across India, south–east Asia, China, North and South Korea and Japan. The presence of this largely monophagous insect suggests that the original native range of its host plant lies within Asia.

While it is an indirect indication, the use of plants in traditional medicine can perhaps indicate a long association of a plant with human culture. In our survey of the medicinal uses of O. corniculata we discovered a total of 182 publications containing 892 remedies for ailments at different locations. It is evident from this map that Asia is at the centre of where O. corniculata is used medicinally (Fig. 2).

In conclusion, there are several high-quality paleobotanical records of O. corniculata in eastern Asia and early documentary evidence. Furthermore, there is at least one insect that feeds on O. corniculata in the continent and there is ample documentation of the use of O. corniculata in traditional Asian medicine. Therefore, there are several lines of evidence for O. corniculata having a long presence in Asia, although the association of O. corniculata with human disturbance and agriculture has probably led to its further spread within Asia. This makes it difficult to point to a more specific area of origin. However, the archaeobotanical remains from eastern China, Japan and the Philippines do point to a long history in this area.

Polynesia

Oxalis corniculata was present on Hawaii (1779) and Tahiti (1769) when they were explored during the scientific expeditions of Captain James Cook (St John, 1978). Indeed, archaeological excavations have found O. corniculata seed in sediment dated from between the 15th and 17th centuries (McCoy, 1977; Allen, 1984). Therefore, O. corniculata is either native to Polynesia or, as suggested by Ridley (1930), was introduced by Polynesian colonizers.