Wolffia columbiana and W. globosa (Araceae) new to Britain

In the course of surveys of ditches in the Pevensey Levels in October 2021, a large population of the non-native Wolffia columbiana H. Karst. was found, new to Britain. Subsequent investigation showed W. columbiana to be widespread on the Pevensey and Gwent Levels and present in other grazing marsh ditch complexes in Somerset, West Sussex and Kent, while W. globosa (Roxb.) Hartog & Plas was also found in the Gwent Levels. This paper describes the known distribution of W. columbiana in Britain, provides information on identification of different Wolffia species and considers the implications of this discovery for conservation of the native W. arrhiza (L.) Horkel ex Wimm.


Introduction
Wolffia L. is a genus of duckweed which includes the smallest flowering plant in the world and is distinguished from other duckweeds by a combination of swollen lower parts and the lack of roots. The genus includes 11 species worldwide (POWO, 2021) and, due to their small size, identification typically relies in part on microscopic characters. Since 2013, three non-native species have been recorded in Europe: W. columbiana H. Karst. (Schmitz et al. 2014), W. globosa (Roxb.) Hartog & Plas (Kirjakov & Velichkova 2013) and W. australiana (Benth.) Hartog & Plas (FLORON, 2021) and it has always been likely that they would appear in the UK (e.g. Adams, 2019).
In September 2021 whilst surveying parts of the Pevensey Levels in East Sussex (v.c.14) to update records of the status of Potamogeton acutifolius, RVL and EJ found a population of Wolffia in one of the larger drains. Initially thought to be the native W. arrhiza (L.) Horkel ex Wimm., closer inspection revealed a number of characters suggesting that it might, in fact, be one of the alien species. Microscopic examination then confirmed its identity as W. columbiana (subsequently also confirmed by W. van der Ven), a first record of this non-native species in Britain.
Subsequent investigation showed W. columbiana to be present in a number of areas, with photographic evidence of its earlier occurrence; and W. globosa was found at a single site in the Gwent Levels (v.c.35).
This paper describes the discovery of W. columbiana in Britain with information on its current known distribution, as well as the discovery of W. globosa, provides guidance on identification of Wolffia species and considers the potential period for which W. columbiana has been established in the country.

Recording Wolffia in Britain in 2021
Following initial discovery of W. columbiana in the ditch in the Pevensey Levels, the levels were re-visited in early October 2021, to attempt to locate populations of W. arrhiza. Sites with past records of W. arrhiza were visited but most were overgrown and shaded by reeds and willow scrub. Wolffia populations were only found at three sites and all involved W. columbiana (Table 1; Fig. 1).
A few days later a visit was made by RVL to the Gwent Levels and a total of six sites surveyed, representing all the main areas within the levels from which W. arrhiza had been reported. All sites supported populations of W. columbiana (Fig. 2  In November 2021 a Wolffia population was sampled in a pond in east Surrey (v.c.17), this did not survive postage particularly well but appeared to include W. arrhiza, possibly as well as W. columbiana and W. globosa however identification is not absolutely certain (see also Discussion). W. arrhiza was also tentatively identified in a sample (also including W. columbiana) from ditch on the Dowells in sheep pasture near Kenardington (TQ9784.3126) in East Kent (v.c.15).

Identification
All Wolffia species are very small; the largest is W. arrhiza, which is typically 0.7-1.3 × 0.6-1.0 mm but is only slightly larger than W. australiana and W. columbiana; the smallest, which is also the smallest known flowering plant, is W. globosa which is typically 0.5-0.8 × 0.2-0.4 mm ( Table 2). They are composed of a photosynthetic upper part which is green and opaque, below which there is an expanded lower layer which is firm and may be green or colourless.  Landolt, 1986;Landolt, 1994;Landolt, 2000;Bog et al., 2020) (Armstrong & Thorne 1984, Landolt 1986, Mazzeo et al. 1993, Landolt 1994 all occurring in Europe (Kirjakov & Velichkova 2013, Schmitz et al. 2014, FLORON 2021. An attempt has therefore been made here to provide guidance on all known species worldwide, while details are provided (Table  2) for the species known to occur in Europe.  (from Landolt, 1986;Landolt, 1994;Sree et al., 2014;Landolt, 2000;Ward et al. 2021;Bog et al. 2013 1 and 3). W. australiana is typically larger than the other species and narrower than all except W. globosa but is best recognised by the fact that fronds are parallel-sided in cross-section and can be much deeper than wide. W. globosa is typically much smaller than the other species, forming populations of uniformly tiny fronds which are narrower than W. arrhiza and W. columbiana. W. arrhiza (Fig. 4) is then recognised by the fact that it lacks the diagnostic characters of the other species. However, identification of plants can be far from straightforward.

Discussion
There has been an increase in new regional records of non-native duckweed species in recent years (e.g. Lansdown 2008, Lansdown et al. 2015, Rumsey & Lansdown 2012) ( Table 2). It is clear that these species are being moved around the world to an increasing extent and there is a need for increasing scrutiny to recognise new invasions when they occur. Wolffia columbiana has clearly become well-established in Europe, where it is now known from the UK (this study), Belgium, France (J. Geslin pers. comm.), Germany (Schou et al. in prep.), Italy (Ardenghi et al. 2017) and the Netherlands (FLORON 2021). Random sampling of Wolffia demonstrated that in Germany and the Netherlands W. columbiana now is much more frequent than the native W. arrhiza (Schmitz et al. 2016, W. van der Wen pers. comm.). W. australiana has also been recorded from the Netherlands where it is known from only a few sites (FLORON 2021), while W. globosa is known from single sites in France (Niebler et al. 2021), Germany (Schou et al. in prep.) and Bulgaria (Kirjakov & Velichkova 2013).
The first published record of Wolffia in Britain was given by Trimen (1866), in relation to a population in a pond near Staines, Middlesex, although Gray (1866) mentioned having been shown specimens about 50 years earlier which were believed to have been collected from Putney Common. Until 2021, all populations were assumed to be W. arrhiza. However, the widespread occurrence of W. columbiana in 2021 strongly suggests that it has been established for some time; two sets of records are of particular note. The first record of Wolffia from the Pevensey Levels is from 1994 (BSBI Distribution Database); it would be surprising in an area so intensively botanised if it had been present in any abundance prior to that date. W. columbiana has certainly been established since 2018 (image EJ) and is now abundant (Fig. 5). It is credible that Wolffia arrhiza was never in the Pevensey Levels but that W. columbiana arrived at least as long ago as the 1990s. Similarly, the first record of Wolffia from the Gwent Levels was in 1982 and it seems credible that W. arrhiza was never present in the area, but that records involved W. columbiana or even W. globosa. In other areas, the Somerset Levels (v.c.5 and v.c.6), East Kent (v.c.15) (Lyons, 2011). When the decade in which Wolffia populations were recorded for the first time in a hectad is compared, there is a striking increase in the middle of the 20 th century and a second in the early 1990s (Fig. 6). It is tempting to think that at least part of one of these increases could be due to the arrival and spread of W. columbiana, although the 1950 peak is likely to be at least partly linked to fieldwork for the first atlas (Perring & Walters, 1962).
Duckweeds show a remarkable capacity for dispersal, and W. columbiana has even been shown to remain viable when transported through ingestion by birds (Silva et al., 2018). Whether by such means or by transfer by attachment to waterfowl, it is feasible that avian transmission may account for the Pevensey Levels  (Ardenghi et al., 2017). Inspection of the bases of plants of Pistia stratiotes and other non-native plant species sold in garden centres shows that many include non-native duckweeds such as Lemna aequinoctialis, L. minuta, L. valdiviana and Spirodela oligorrhiza (Rumsey & Lansdown, 2012;RVL unpublished data). It is highly likely that the inadvertent or deliberate introduction of cultivated aquatic plants to the wild brings with it some of these non-native duckweeds.  Images taken from the side of plants of W. columbiana, which have a globose cross-section, a differentiated margin when viewed from above and translucent upper cells, from the Gwent Levels (Fig. 3) more closely resemble those of W. arrhiza given by Landolt (1994, Fig. 2h) than those of W. columbiana (Landolt 1994, Fig. 2o), particularly in the extent of the differentiated upper part and the degree to which the upper part is domed. This character therefore appears to be unreliable.
Wolffia species are very small and there are only a very small number of features which can serve for morphological identification, particularly as counts of the number of stomata appear to be unreliable without recourse to SEM. Counts of stomata using a dissecting microscope suggest that populations of W. columbiana recorded in north-western France had <10 stomata per frond, whereas counts from material from the Gwent, Pevensey and Somerset Levels typically had (8-)10-20(-25). Schmitz et al. (2014) using SEM noted that counts from German populations found 3-14 stomata per frond (mean 5.5) and Dutch populations 3-11 per frond (mean 6.4). Landolt (1986) notes that "In Florida, there are clones of W. columbiana which have up to 30 stomata and therefore resemble W. arrhiza". The difference in counts in Europe could be because W. columbiana has spread within each country from one or only a few introductions, with most reproduction being vegetative and as a result, the populations involve a small number of clones. Apart from gross and typically unequivocal morphological structures, such as presence of a dorsal papule in W. brasiliensis or tapering into a root-like structure below in W. microscopica, most identification is based on poorly-defined descriptive terms or relative characters, such as "spherical", "ellipsoid" or "boat-shaped" (Landolt 1986), "Fronds intensely green and mostly shiny at the surface" (Landolt 1994), relative length to width, width to depth etc. There is also typically an overlap in characters presented and as the plants are so small, even small overlaps can render characters almost useless. It is therefore very difficult to achieve confident and, above all, reliable identification of Wolffia species other than those with particularly distinctive characters, such as W. brasiliensis, W. columbiana and W. microscopica. As noted by Landolt (1994) there is a need for more research.
The most striking aspect of the survey of Wolffia populations following the discovery of W. columbiana in the Pevensey Levels in 2021 was the very limited number of populations of W. arrhiza recorded. There is an urgent need to re-visit Wolffia populations to confirm the identity of the species involved, document all nonnative species present and as a high priority, confirm the distribution and conservation status of W. arrhiza.