New Lepidium (Brassicaceae) from New Zealand

Abstract A revision of the New Zealand endemic Lepidium oleraceum and allied species is presented. Sixteen species are recognised, 10 of these are new. The new species are segregated on the basis of morphological characters supported by molecular data obtained from three DNA markers (two rDNA and one cpDNA). One species, Lepidium castellanum sp. nov., is endemic to the Kermadec Islands where it is sympatric with Lepidium oleraceum. The North Island of New Zealand supports four species, with two of them, Lepidium amissum sp. nov. and Lepidium obtusatum, now extinct. The South Island supports six species, that, aside from Lepidium banksii, Lepidium flexicaule and Lepidium oleraceum, are all confined to the south-eastern half of the island (Lepidium aegrum sp. nov., Lepidium crassum sp. nov. and Lepidium juvencum sp. nov.). One of these, Lepidium juvencum sp. nov., extends to Stewart Island. The Chatham Islands support six species (Lepidium flexicaule, Lepidium oblitum sp. nov., Lepidium oleraceum, Lepidium oligodontum sp. nov., Lepidium panniforme sp. nov., and Lepidium rekohuense sp. nov.), one of which, Lepidium oligodontum sp. nov., extends to the Antipodes Islands group. The remote, subantarctic Bounty Islands group supports one endemic, Lepidium seditiosum sp. nov., which is the only vascular plant to be recorded from there. Lepidium limenophylax sp. nov. is known from islands off the south-western side of Stewart Island/Rakiura, The Snares and Auckland islands. Lepidium naufragorum, although not related to Lepidium oleraceum and its allies, is also treated because populations with entire leaves are now known. Typification is undertaken for Lepidium banksii, Lepidium oleraceum, Lepidium oleraceum var. acutidentatum, var. frondosum and var. serrulatum.

named varieties can be recognised, [though] there is considerable variation (possibly clinal) present within L. oleraceum. A critical investigation of the whole complex, based on genetic as well as morphological analyses, is required before any taxonomic changes within this species are formalised'.
In this paper we present a taxonomic revision of L. oleraceum s.l. (hereafter referred to as the Lepidium oleraceum group), along with the allied species L. banksii Kirk, L. fl exicaule Kirk, and the extinct L. obtusatum Kirk. Th is revision began in 1990 ( de Lange 2010) and has involved the critical study of type specimens and herbarium material, extensive fi eld work to examine natural variation of wild populations throughout their geographic range, supplemented with the use of DNA sequence data and, where possible, cultivating plants under uniform conditions. Although the current natural distribution of L. oleraceum group and its allies comprises small, highly fragmented populations that most probably represent only a small remnant of the original distribution, abundance, and morphological and genetic diversity of the species, suffi cient populations remain to enable the taxonomic revision off ered here.
In addition to revising the L. oleraceum group, we also treat Lepidium naufragorum. At the time of its formal naming L. naufragorum was regarded as having a relationship to L. fl exicaule and the L. oleraceum group (Garnock-Jones and Norton 1995). It is now known to have no close relationship to these species (Mitchell and Heenan 2000;Heenan et al. 2007). Nevertheless we treat it here because of the recent discovery of some populations that have entire rather than pinnatifi d rosette leaves, so necessitating a slight revision of the description provided by Garnock-Jones and Norton (1995).
To this sampling were added further plants of L. banksii, L. fl exicaule, L. oleraceum group and L. naufragorum collected by New Zealand botanists from a wide range of sites, and held by the Auckland Botanic Gardens, University of Auckland, Oratia Native Plant Nurseries, Motukarara Conservation Nursery and Landcare Research (Lincoln campus) to meet the ex-situ objectives of the Department of Conservation Coastal Cress Recovery Plan (Norton and .

Molecular phylogenetics
In conjunction with the taxonomic revision of the L. oleraceum group and allied species, phylogenetic studies were carried out to determine relationships among the taxa, as well as to gain a better understanding of the population variation in extant populations of Lepidium species. For simplicity and clarity we use the new taxonomic names described here throughout the manuscript.
All taxa accepted in the revision presented here were included in the phylogenetic analyses as well as other closely related New Zealand and Australian taxa (Table 1) (Mummenhoff et al. 2004;Heenan et al. 2007). DNA sequence data (nrDNA (ITS1-5.8S-ITS2 and ETS), cpDNA (trnL-F spacer regions)) was obtained from 65 representatives of the L. oleraceum group and seven from the L. sisymbrioides group, which were used as outgroups (Mummenhoff et al. 2004). Not all DNA regions were amplifi ed for each sample because it was diffi cult to access material. Only one species, the endemic and extinct Lepidium amissum de Lange & Heenan, described in this paper, was not included because three attempts failed to amplify DNA from the limited material available (three herbarium sheets > 100 years old). Th e phylogenetic relationship of both the L. oleraceum and L. sisymbrioides groups, and so the validity of using taxa from the L. sisymbrioides as outgroups, had already been established by Mummenhoff et al. (2004). Our sampling of the L. oleraceum group was intended to encompass the entire known range and extant variation within the species.
Total genomic DNA was extracted from 0.1-1.0 g of leaf tissue using a QIAGEN DNeasy® Plant Mini Kit or an INTRON Plant DNA kit using fresh material, fresh material stored in Silica Gel, and from herbarium specimens. DNA sequencing of nrDNA (ITS1-5.8S-ITS2 and ETS), and plastid DNA (trnL-F spacer regions) were carried out following the methods of White et al. (1990) and Mitchell and Heenan (2000) respectively.
PCR products were purifi ed using either the Perfectprep PCR cleanup kits (Eppendorf ), or diluted to approximately 1:4 with ultrapure water. Each sample was sequenced in the sense and antisense direction by the Centre for Gene Technology (University of Auckland), the Allan Wilson Centre Genome Service, (Massey University, Albany), or the Landcare Research Ecological Genetics Laboratory (Tamaki). Sequencing reactions were performed with the same primers as the PCR amplifi cations and the 3.1 ABI Prism TM Big Dye Terminator Sequencing Kit (Applied Biosystems, Scoresby, Vic.). Sequences obtained in this study have been assigned GenBank accession numbers (Table 1).

Matrix preparation
Sequence alignment was performed using ClustalX vers. 1.81 (Th ompson et al. 1997). Multiple alignment parameters were set to 12 for gap opening penalty and 6.0 for gap extension penalty. No further alignment was required.
Maximum likelihood (ML) analyses were conducted using RaXML (Stamatakis 2006) employing the full search algorithm with Gamma Distribution and General time reversible (GTR) model. In each of the datasets support for nodes was assessed by non-parametric bootstrapping (Felsenstein 1985) using 1000 replicates. Because of the simple nature of the dataset, no further phylogenetic reconstruction was attempted. Bayesian reconstruction in Mr Bayes (Huelsenbeck and Ronquist 2001), and parsimony analysis (Swoff ord 2006) in Paup* were undertaken, but preliminary analyses yielded similar results (not shown).

DNA sequence data
Phylogenetic analyses provided some resolution of relationships, with ETS the most informative ( Fig. 2), resolving the L. oleraceum and L. sisymbrioides groups. Within the L. oleraceum group there is a strong north-south dichotomy, with another group present on the Chatham Islands and sub-Antarctic islands. Th e samples from the North Island comprised only L. oleraceum sensu stricto (s.s.), and the southern group included L. aegrum Heenan et de Table 1. Herbarium vouchers and GenBank numbers of Lepidium study group and outgroup samples used in the present investigation.

AJ582424, AJ582481
AY015877, AY015963 Mummenhoff et al. (2004) L. panniforum de Lange & Heenan Chatham Is., Mangere I., and the Kermadec Islands' L. castellanum de Lange et Heenan. Chatham Island and sub-Antarctic species included L. rekohuense de Lange & Heenan, L. oblitum Houliston, Heenan et de Lange, L. panniforme de Lange & Heenan, and L. oligodontum de Lange Figure 2. Maximum likelihood phylogeny of ETS as calculated in RaXML, showing the Lepidium oleraceum (black text) and L. sisymbrioides (red text) groups, including the newly described taxa. Bootstrap support from 1000 replicates is shown for branches where it is greater than 50. Codes for each individual in the phylogeny are shown in Table 1. Th e main clade branches have been truncated to accommodate the tree within the journal format. AUS = Australia, NZ = New Zealand.
et Heenan from the Chatham Islands and L. seditiosum de Lange, Heenan & J.Rolfe from Bounty Island. ITS also resolved the L. oleraceum and L. sisymbrioides groups, but within ITS, there was no resolution within the L. oleraceum group including L. banksii (Fig. 3). Th e pattern for trnL-F was more complex (Fig. 4), and is discussed below, although it again provided no resolution within the L. oleraceum group and L. banksii.  Table 1. Th e main clade branches have been truncated to accommodate the tree within the journal format.

ETS Sequence Characteristics and Phylogeny
Eighteen unique ETS sequences were present among the 65 Lepidium samples that were amplifi ed for this region (58 from the L. oleraceum group and seven from the L. sisymbrioides group) (Fig. 2). Within the L. sisymbrioides group, the four samples of L. sisymbrioides all had diff erent ETS ribotypes, and L. naufragorum, and L. tenuicaule Kirk also contained more than one ribotype. Within the L. oleraceum group, one described species, L. fl exicaule, was found to contain more than one ETS ribotype, with the New Zealand and Tasmanian examples of this species diff ering by two base substitutions. Two pairs of species (L. rekohuense and L. seditiosum; L. oleraceum and L. banksii) shared the same ETS type (Fig. 2). Lepidium fl exicaule (Tasmanian sample), L. foliosum Desv. (Australian endemic) and L. desvauxii Th ell. (Australia and New Zealand samples) also share identical ETS sequences, however New Zealand L. fl exicaule have a diff erent ETS type (Fig. 2). Th e three most southern / central South Island species, L. crassum, L. aegrum and L. juvencum, share an identical ETS type with L. seditiosum from the Bounty Islands, while L. castellanum from Macauley Island (Kermadec Islands group) is also similar, diff ering by only one base pair (Fig. 2).
Th ere were three diff erent ribotypes in L. oleraceum s.s., with a single sample from Mangere Island diff ering by a single base pair to the other L. oleraceum s.s. samples.
Th e extinct Lepidium obtusatum and morphologically rather diff erent Chatham Islands endemic L. panniforme also share an ETS sequence type, which is almost identical to that of another Chatham Islands endemic, L. oblitum (Fig. 2). Notably, these three species also diff er from the other Chatham Islands endemic, L. rekohuense, and the Chatham /Antipodes islands endemic, L. oligodontum (Fig. 2). Lepidium limenophylax, known from the southern Titi Island of the south-western coastline of Stewart Island, the Snares and Auckland Islands, is sister to these Chatham / Antipodes islands species (i.e., L. oblitum, L. oligodontum, L. panniforme, and L. rekohuense) (Fig. 2).
Th is study also confi rms the taxonomic status of the Lord Howe Island endemic, L. nesophilum Hewson, which is a species that had previously been treated as L. oleraceum (see Green 1990).

ITS sequence characteristics and phylogeny
Th e 56 samples of Lepidium from which ITS data were collated (48 L. oleraceum group, 6 L. sisymbrioides group) are shown in Fig. 3. With the exception of L. limenophylax, all of the other species within the L. oleraceum group, including L. banksii and the Chatham Island sample of L. fl exicaule, shared the same ITS sequence (Fig. 3).
Th e New Zealand (Karamea) sample of L. fl exicaule grouped with Australian samples from which it diff ered by a single base pair (Fig. 3). In addition, the two Lord Howe samples of L. nesophilum, had markedly diff erent ITS sequences, though, notably, they clustered with other Australian species and shared an identical ETS sequence.

trnL-F sequence characteristics and phylogeny
Th e plastid data (trnL-F spacer region, 30 L. oleraceum group, 7 L. sisymbrioides group included) is interesting for the lack of strong divergence between the oleraceum and sisymbrioides groups, with two species (L. sisymbrioides and L. naufragorum) being found in both clades (Fig. 4); see Heenan et al. (2007) for discussion on sequence variation in L. naufragorum and L. sisymbrioides. Th is confl ict could be explained by hybridisation between representatives of each of the two groups. With the exception of L. limenophylax (which diff ered by one base pair), all of the species previously described in the L. oleraceum group shared the same trnL-F sequence (Fig. 4).
Although the trnL-F spacer region showed only low sequence variation, several patterns were evident. For example, the morphologically distinctive species L. nau- Figure 4. Maximum likelihood phylogeny of trnL-F as calculated in RaXML, showing the Lepidium oleraceum (black text) and L. sisymbrioides (red text) groups, including the newly described taxa. Bootstrap support from 1000 replicates is shown for branches where it is greater than 50. Codes for each individual in the phylogeny are shown in Table 1. fragorum (New Zealand), L. fl exicaule (Australia + New Zealand), L. muelleri-ferdinandi Th ell. (Australia), L. fasciculatum Th ell. (Australia) and L. oleraceum (Kermadec Islands and New Zealand) had identical sequences (Fig. 4), a pattern that suggests a recent origin for these taxa. Further, the two New Zealand samples of L. fl exicaule (South Island (Karamea) and Chatham Islands) have distinct trnL-F sequences (Fig.  4). Morphologically, the Chatham Island race of L. fl exicaule diff ers from the majority of New Zealand (North Island and South Island) L. fl exicaule plants in that they lack the marginal leaf, pedicel and stem denticles that are diagnostic of the species on mainland New Zealand (see Garnock-Jones 1988). Interestingly, the Australian (Tasmanian) samples of L. fl exicaule had the same trnL-F sequence as the Chatham Islands, and they too lack denticles. However, our study has found that, while denticles are usual for New Zealand (North Island, South Island) examples of this species, not all plants from western South Island have them, and, as we have not sequenced examples of these "non-denticulate" plants, further study into this pattern is warranted. Th e pattern is reversed for the ITS data set, in which the South Island (Karamea) sample of L. fl exicaule clusters with Australian L. fl exicaule (Fig. 3).

Phylogenetic implications
Bootstrap support for the trees was low to moderate, as would be expected for recently diverged taxa where resolution and divergence is low and where there has possibly been recent reticulation. An earlier study has indicated that hybridisation, and biocontinental origin for the New Zealand / Australian Lepidium is likely, although the age of this event and whether it is related to the divergence from the rest of the Lepidium group is unclear (Mummenhoff et al. 2004). Estimates for times of divergence vary between the L. oleraceum and L. sisymbrioides groups (0.7-1.3 Myr and 0.3-0.55 Myr respectively; Mummenhoff et al. 2004), although further hybridisation events and potential multiple dispersal events make accurate inference diffi cult. However, as noted by Mitchell and Heenan (2000), there was good support in the rDNA data for the L. oleraceum and L. sisymbrioides groups, though this was not found in the cpDNA dataset.
Lepidium naufragorum was found in the L. oleraceum clade for the plastid data, but is a member of the L sisymbrioides group for both nuclear regions. Th is incongruence suggests that L. naufragorum could have had a hybrid origin, with a parent from each of these groups (see in this respect also Mummenhoff et al. 2004). Although further study is required to address this issue, shared DNA sequences are common within New Zealand taxa. Th ey have usually been interpreted as evidence of genera that have undergone a recent and rapid species radiation (e.g. Wagstaff et al. 1999 (for Carmichaelia R.Br.), Wagstaff et al. 2004 (for Brachyglottis J.R.Forst. et G.Forst.), Mitchell and Heenan 2002 (for Sophora L.),  (for Crassula L.), de Lange et al. 2010b (for Kunzea Rchb.), Smissen et al. 2011 (for Simplicia Kirk), Prebble et al. 2012 (for Wahlenbergia Schrad.)). Th is explanation is also most likely for the shared ETS ribotypes between L. rekohuense and L. seditiosum; and L. oleraceum and L. banksii, which also show this pattern.
In general, the L. oleraceum and L. sisymbrioides groups formed two distinct clades in both nuclear datasets, but was not strongly supported by the plastid data, as was found by Mummenhoff et al. (2004). Signifi cantly for this study, the three DNA data sets confi rm that those informally recognised morphological units within the L. oleraceum group (see de Lange 2010;de Lange et al. 2010a) warrant formal taxonomic recognition. Th is is especially evident when, in our study, such morphologically discrete taxa as L. banksii (Garnock-Jones 1988) proved indistinguishable from the very diff erent L. oleraceum s.s based on the rDNA and cpDNA sequences used here. Th at such a distinct species is not distinguished using this data also further supports the idea that much of the New Zealand Flora is in a state of rapid speciation (see above references), and that the reliance of DNA data sets, in isolation, to resolve taxonomic uncertainty is fraught with diffi culty. In the case of the sequence data presented here. We conclude that the retention of a single morphologically divergent species, L. oleraceum is unacceptable, and that formal taxonomic segregation within the group, at species level is warranted.

Growth habit
Lepidium oleraceum s.s., its allied species, and the unrelated L. naufragorum are all perennial herbs that can be distinguished by their growth habit. Th is includes plants with an upright, spreading or sprawling, shrubby growth habit (L. castellanum, L. aegrum, L. banksii, L. crassum, L. juvencum, L. oleraceum, L. panniforme and L. seditiosum) and those with a prostrate or decumbent growth habit (L. amissum, L. fl exicaule, L. limenophylax, L. oblitum, L. obtusatum, L. oligodontum, and L. rekohuense). Lepidium obtusatum is, as discussed by Cockayne (1921), further distinguished from all other members of the L. oleraceum group by its uniquely suckering habit. A further separation, based on seasonal growth patterns, is also evident, i.e., whether plants die back to the rootstock or not during winter. Of the extant species, L. castellanum, L. aegrum, L. juvencum, and L. oleraceum have persistent stems that do not die back to a rootstock; L. oblitum and L. panniforme stems are usually long-persistent, though sometimes they die back to a basal rootstock. Th e remaining extant species all have stems that die back to their rootstock. Herbarium evidence indicates that the extinct L. obtusatum also died back in winter but there are insuffi cient herbarium specimens and no written accounts to enable us to determine whether the extinct segregate L. amissum did so as well. Unique amongst the L. oleraceum group, well-developed plants of L. limenophylax may develop prominent "swellings" around branch nodes; these may also be a point of seasonal die back and vegetative growth and they sometimes produce adventitious roots.

Leaf characters
Th e leaves of Lepidium oleraceum and allied species show a diversity of shape and degree of dentition that has been the basis for the recognition of a number of species and varieties (see Kirk 1899;Th ellung 1906). Lamina-shape immediately distinguishes the pinnatifi d Lepidium fl exicaule from the other members of the L. oleraceum group but not from the pinnatifi d-leaved forms of the unrelated L. naufragorum. Among the remaining members of the L. oleraceum group, lamina size, shape, and degree of dentition provide important diff erences between the species (Figs 5,6).

Floral and fruit characters
Th e infl orescences, fl owers and fruits of the Lepidium oleraceum group provide a number of characters enabling species recognition. Th e infl orescences of the species in the group are usually leaf-opposed. While in some species (L. fl exicaule and L. obtusatum) the infl orescences tend to be obscured by foliage, they are held well above the surrounding foliage in the majority of species. In most species the infl orescence rachises and pedicels are glabrous, but in L. castellanum, L. banksii, L. rekohuense, L. seditiosum, (and very rarely L. oblitum), minute, usually sparse, eglandular, clavate hairs are present. Th e sepals and petals show little variation with respect to size, shape, and degree of investiture. Th e size, shape and degree of notching of the silicles, and the length and shape of the stigma remnant are also critical characters (see Garnock-Jones 1988).
Th e presence or absence of an apical notch to the silicle in the New Zealand species was an important character emphasised by Kirk (1899), Cheeseman (1906Cheeseman ( , 1925, Allan (1961), andGarnock-Jones (1988). In all these treatments the lack of a notch in Lepidium oleraceum is noted, and a silicle of that species showing this is clearly illustrated by Garnock-Jones (1988). Th is is a signifi cant observation, because, of the new species segregated here from L. oleraceum s.s. a shallow apical notch is present in eight species (though minute and so scarcely evident in L. aegrum, L. limenophylax, L. juvencum and L. rekohuense), absent in two (L. castellanum and L. oleraceum), and unknown for L. seditiosum (see above comments). We also stress that care is needed to observe the presence or absence of a notch. It is best seen in freshly matured silicles because, in older, over mature silicles, the apically initiated disarticulation of the valves can be misleading as it gives the false impression of a notch. Notching was noted by past workers (Kirk 1899;Cheeseman 1906Cheeseman , 1925Allan 1961;Garnock-Jones 1988) as being especially prominent in L. banksii and L. obtusatum, as it also is in the new species L. amissum (which is segregated here from L. obtusatum) and L. oblitum. It is odd, therefore, that despite the prominence of the notch in L. banksii, Kirk (1899) in his dichotomous key emphasised the cordate silicle base rather than apical notch as diagnostic of L. banksii and he did not mention it in his key for L. obtusatum. In fact cordate silicle bases are not  unique to L. banksii as they are also present in L. oligodontum. Silicle wings that occur around the margin of the silicle are also important. Of the 16 species for which mature silicles are available, ten have slightly winged silicles and six species (L. castellanum, L. aegrum, L. juvencum, L. crassum, L. juvencum, L. oleraceum) do not.   Etymology. Th e specifi c epithet 'aegrum' is derived from Latin, meaning 'diseased and/or sick body' and refers to the condition of the plants in the single known extant wild population on the unnamed rock stack near Island Bay, many of which are infected with turnip mosiac virus (Fletcher et al. 2009).
Conservation Status. Lepidium aegrum is known from a single wild population and three additional populations that have been established with nursery-raised plants. Th e single wild population and unknown recruitment patterns means that this species is especially vulnerable to stochastic events. Furthermore, turnip mosiac virus, which deforms and retards growth, is common in plants at the wild population, and this will likely reduce reproductive success (Fletcher et al. 2009). Using the New Zealand Th reat Classifi cation System (Townsend et al. 2008), L. aegrum qualifi es as Th reatened/Nationally Critical. We recommend appending the qualifi ers 'CD' (Conservation Dependent -as it has been established and is being managed at Motunau and Quail islands, and at Stony Bay), 'DP' (Data Poor -to refl ect uncertainty over plant numbers and population trends), 'OL' (One location -since it is known from a single wild population). de  Holotype. New Zealand (Fig. 16): Cliff s between Karekare and Manukau Heads, January 1917, T. F. Cheeseman s.n., AK 4474! Isotypes: AK 206570!, WELT SP030095! Etymology. Th e epithet 'amissum' is derived from the Latin 'amissus' meaning 'lost' and is used here to refer to the loss though extinction, of this plant from the New Zealand fl ora.
Representative Specimen. New Zealand (North Island): North Auckland, Cliff s north of the Manukau heads, December 1870, T. F. Cheeseman s.n.,(AK 4473). Distribution (Fig. 15). Endemic. New Zealand, North Island, where it was recorded from 'between Manukau Heads and Karekare' (Cheeseman 1906(Cheeseman , 1925, not  the 'Titirangi' noted by Kirk (1899). Lepidium amissum was last collected from the west Auckland coastline in 1917 and is presumed to be extinct.
Recognition. Lepidium amissum had previously been included in L. obtusatum, perhaps because the species have no obvious fl oral diff erences and both have deeply notched and winged silicles with tapering bases. However, from Lepidium obtusatum, L. amissum diff ers by the non-rhizomatous, suberect growth habit; arching, sparingly leafy stems (Fig. 16); long persistent, deeply, and sharply incised rosette and lower stem leaves (Figs 17,18); upper stem leaves which are lanceolate, narrowly oblanceolate, to narrowly obdeltoid and mostly prominently toothed, with up to 6 pairs of deeply incised teeth, and with a truncate apex bearing 3-4 prominent teeth. Th e infl orescences of L. amissum are larger, only sparsely leafy when mature, and without associated vegetative stems (Fig. 19). Th e silicles are smaller than those of L. obtusatum (Fig. 20).

Ecology.
Little is known about the habitat preferences and ecology of L. amissum, beyond that it grew on sea cliff s (Cheeseman 1906(Cheeseman , 1925. Conservation Status. Lepidium amissum is considered to be extinct. Th ere have been repeated unsuccessful surveys for this species over the last fi fty years by various botanists. Aside from the type the species is known from only one collection made in 1917. Th is collection and the type came from an ill defi ned area of coastline stretching for some 6 km from the Manukau Heads to Karekare. Today this coastline is protected within the Centennial Park, Waitakere Ranges. However, during the period when this species was gathered, all of the coastline was being modifi ed as a result of kauri (Agathis australis (D.Don) Lindl.) logging (see Harvey 2001Harvey , 2006MacDonald and Kerr 2009), and this may have contributed to its eventual extirpation. It does seem that L. amissum was already uncommon at the time of its discovery and it is possible that the gatherings made by Th omas Cheeseman where suffi cient to cause its extinction as they are copious and comprise many stems.  Lectotype (fi de Allan 1961): Northwest Bay, Pelorus Sound, J. Rutland s.n., n.d., WELT SP079012! Etymology. Although Kirk (1899) did not explain his choice of epithet, the protologue makes it clear that "ovatum" was chosen to refer to the shape of the mature silicles.

Lepidium banksii
Note: Th e key provided by Kirk (1899, p. 34) used the name "L. forsteri" instead of L. banksii. Th is is considered to be a nomen nudum.

Distribution (Fig. 15). Endemic. Kermadec Islands (Macauley Island, Curtis Island, Haszard Islet and Cheeseman Island).
Recognition. Lepidium castellanum is distinguished by its very robust, shrubby growth habit (Fig. 27), sometimes up to 1.8 × 2.0 m, erect, often closely packed, usually leafy stems, narrowly lanceolate to linear-lanceolate upper stem leaves, and by the very long, needle-shaped teeth of the leaves which reach well beyond the leaf margin . In this species, the pedicels (and often the infl orescence rachises) are minutely hairy (Fig. 31), whilst those of L. oleraceum are glabrous. Lepidium castellanum is sympatric with Lepidium oleraceum on Curtis Island. Elsewhere, in northern New Zealand, especially on the Poor Knights and Hen (Taranga) islands, narrow-leaved forms of L. oleraceum are also common. However, these plants have a much smaller growth habit, their leaves lack the distinctive long needle-shaped teeth and long-acuminate leaf apices of L. castellanum, and, as with L. oleraceum populations elsewhere, their pedicels are glabrous.
Ecology. Lepidium castellanum is a sparse associate of the vegetation that has colonised the soft, erosion-prone cliff tops of Macauley  Forst. On Cheeseman Island, numerous seedlings and a few adults were seen in 2011 growing on heavily petrel-burrowed soil covered ledges amongst Disphyma australe subsp. stricticaule, and also as a few sub-adults within an active petrel colony in a narrow gully, growing in association with Cyperus insularis, Tetragonia tetragonioides, and Parietaria debilis.
Conservation Status. Lepidium castellanum is confi ned to the Southern Kermadec Island group (see de Lange et al. 2004), part of the Kermadec Islands Nature Reserve. Within the southern Kermadec Islands group it has been recorded from Macauley Island, Cheeseman Island, Curtis Island, and from Haszard Islet. On Macauley Island it was recorded twice, in 2002(Barkla et al. (2008; T. C. Greene pers. comm.), at that time numbering at best fi ve or fewer individuals. However, it was not seen on Macauley Island in 2011 (de Lange 2012;de Lange in press). In 2002 and 2006, the summit of Haszard Islet (6 ha, of which c.2.1 ha is vegetated) was thoroughly explored during helicopter-assisted landings by ornithologists. During those visits, it was noted that Lepidium castellanum was one of the major shrub-forming species on the richly manured and petrel-turbated soils of that islet's summit slopes (T. C. Greene and K. Baird pers. comm.). While accurate counts of the number of plants present on Haszard Islet were not made, it would seem that on both visits the islet summit supported approximately 20-50 individuals (T. C. Greene and K. Baird pers. comm.). Further south, its status on Curtis Island remains unknown, as there have been no gatherings of Lepidium from there since 1969, and several visits during the 1980s and in 2002 did not see it there (G.A. Taylor and T. C. Greene pers. comm.). However, on nearby Cheeseman Island, a 7-ha island (of which c.5.6 ha is vegetated) To that assessment we add the qualifi ers 'DP' (Data Poor) and 'IE' (Island Endemic). Data Poor ('DP') is recommended because of the lack of accurate counts of wild plants from Haszard Islet, absence of any recent survey for the species on Curtis Island, and lack of any trend data.
Recognition. Lepidium crassum diff ers from the related L. oleraceum by its usually much smaller stature (Fig. 34) and seasonal growth habit (with plants dying back to a basal rosette overwinter). Lepidium crassum has distinctly petiolate, uniformly broadly elliptic, elliptic to obovate, thickly coriaceous, often doubly crenate leaves with obtuse to truncate apices (Figs 35,36). Its silicles are usually orbicular, sometimes orbicular-rhomboid, and with obtuse to shallowly notched apices (Fig. 37) and by its rDNA ETS sequence.  var. repens. Th is species has also colonised the man made 'Mole' at Aramoana (Fig. 34) where it is can be common amongst rubble, concrete blocks and landfi ll.
Conservation Status. Lepidium crassum is known from a small number of wild populations on the Otago coast. As most populations comprise few plants or occur on unstable coastal headlands and rocky outcrops, this species is vulnerable to stochastic events. We estimate there are between 250 and 1000 plants and the largest population comprises fewer than 300 plants. Using the New Zealand Th reat Classifi cation System (Townsend et al. 2008), L. crassum qualifi es as 'Th reatened/Nationally Endangered' (criterion B(1/1). We recommend appending the qualifi ers 'CD' (Conservation Dependent -because there have been several populations established by the Department of Conservation and these are being actively managed), 'DP' (Data Poor -to refl ect uncertainty over plant numbers), 'EF' (Extreme Fluctuations -because the largest known monitored population, that on Wharekakahu Island, appears to naturally undergo huge and apparently natural population fl uctuations (J. W. Barkla pers. comm.)), and 'RR' (Range Restricted -because it is naturally confi ned to a geographically small part of the New Zealand Botanical Region). Etymology. Th e exact meaning of the species 'fl exicaule' was not given by Kirk (1899)  Description (Figs 38-43). Tap-rooted, strongly pungent smelling, decumbent, summer-green, perennial herb forming densely leafy, patches up to 1.5 m diam., and arising from stout, semi-circular, whitish-grey (when exposed) rootstock 3-10 mm diam. when fresh. Tap root fl eshy, yellow to yellow-white when fresh, up to 300 mm long, deeply descending. Plants dying down in winter or in times of adversity to rootstock (in well-grown individuals the new season's branchlets may die back to the previous season's stem nodes). Stems prostate, weakly fl exuous, divergent to widely spreading, 100-800 mm long, 0.8-5.2 mm diam., woody near base, initially spherical in cross-section, pale yellow-green to dark green, sometimes tinged maroon and fi nely puberulent with short papillate or tapered hairs; becoming glabrous and prominently ridged and/or grooved with age, and usually bearing numerous leaf scars and withered petioles; upper portion of stems sparingly and openly to heavily branched; branches and branchlets, usually very leafy. Leaves glabrous, fi rmly fl eshy to succulent, glossy dark green to yellow-green, at senescence turning yellow. Rosette and lower stem leaves withering at fruiting; petioles distinct 10-50 × 1-3 mm, slightly concave in crosssection, fl eshy; leaves glabrous above, with papillae or denticles along midrib beneath and margins in New Zealand plants, or without denticles (absent in Chatham Islands and most Australian plants); lamina 50-90 × 15-30 mm, pinnatifi d, obovate to oblanceolate; pinnae in 1-3 pairs, bluntly toothed or crenate at apex and distal margins. Upper stem leaves glabrous above, usually with triangular denticles, dense on margins and sparse to dense beneath (although absent in Chatham Islands and most Australian plants); mostly apetiolate, petiole if present minute up to 1.5 mm long; lamina 5-30 × 3-10 mm, obovate, oblanceolate, or spathulate, apex bluntly toothed to crenate, base broadly to narrowly cuneate. Racemes 10-50 mm long, elongating up to 60 mm at fruiting, terminal and leaf-opposed; rachis and pedicels puberulent with short tapering hairs, and ± covered in triangular denticles (absent in Chatham Islands and most Australian plants) or glabrous; axillary; rachis and pedicels hairs if present, retrorse to patent, very short, 0.05-0.8 mm long, ± clavate, eglandular -glandular; pedicels, erecto-patent to patent, initially 1.04-1.27(-2.38) mm; elongating to 2.34-5.00(-6.02) mm long at fruiting. Flower buds dark green, apex bearing a conspicuous, caducous, crest of white, eglandular, antrorse hairs up to 0.9 mm long. Flowers sweetly fragrant, 1.4-1.8(-2.3) mm diam. Sepals, broadly ovate to oval, apex broadly obtuse, centrally green with a white margin, deeply concave, adaxially weakly keeled, adaxial midrib invested in conspicuous, caducous, white, eglandular, antrorse, hispid hairs, hairs sometimes scattered across rest of adaxial surface; abaxial surface glabrous; lateral sepals broad, 0.6-1.0 × 0.6-1.2 mm , median sepals narrower 0.4-0.8 × 0.4-0.7 mm. Petals white, 0.3-0.8(-1.0) × 0.2-0.8 mm, erecto-patent or patent, clawed; limb broadly obovate, apex obtuse, retuse or distinctly emarginate. Stamens 2, equal. Anthers c.0.16 mm long. Pollen bright yellow. Nectaries 4, subulate, 0.40 mm long. Silicles cartilaginous when fresh, subcoriaceous when dry, (2.1-)3.3(-4.1) × (2.2-)3.3-3.4(-4.0) mm, orbicular, obovate, to ovate, slightly winged in upper ¼, apex scarcely or shallowly notched, valves green maturing yellow-green, glabrous; style 0.8 (-1.0) mm long, free from the narrow wing, equal to or slightly exceeding the notch; stigma 0.2-0.4 mm diam. Seeds 2, 1.20-1.38 × 0.80-1.10 mm, ovoid to suborbicular, red-brown, dark red-brown or brownish black, not winged. FL. Oct-Feb. FR. Dec-Apr. Distribution (Fig. 44). Indigenous. Australia (Tasmania) and New Zealand (North, South, Chatham islands). In the North Island of New Zealand, Lepidium fl exicaule is now known from only two extant populations, one on Saddle Island off the west coast of Great Barrier Island (Aotea Island) ( de Lange and Cameron 2011) and the other at Stent Road north of Cape Egmont, Taranaki (Peet et al. 2003). Historically Lepidium fl exicaule was present on Rangitoto Island, along the Waitakere Coastline, near North Head, Waitemata Harbour and near Onehunga, Manukau Harbour (Kirk 1899;Cheeseman 1906Cheeseman , 1925Garnock-Jones and Norton 1995). It was also recorded from the Firth of Th ames, at Mercury Bay (Kirk 1899;Cheeseman 1906Cheeseman , 1925Garnock-Jones and Norton 1995) and near Wellington (Garnock-Jones and Norton 1995). In the South Island, the species is confi ned to the western coast of North-West Nelson from near Kaihoka to about Point Elizabeth. Lepidium fl exicaule was discovered on the northern portion of Rekohu in 2005 by Department of Conservation botanists.
Recognition  angular denticles on branchlet stems, leaf margins and petioles, and on the rachis and pedicels of the infl orescences (Fig. 42, see also Garnock-Jones 1988). Denticles are, however, absent from the Chatham Islands, most Australian collections, and some South Island ones. Th e absence of denticles prompted the brief listing of Rekohu and Australian plants as an unnamed subspecies (see de Lange et al. 2009). However, as denticles are sometimes absent from South Island plants, formal taxonomic distinction may not be warranted. Nevertheless, it is worth noting that Rekohu plants are consistently larger than those seen from the rest of this species' range, with well-grown specimens reaching up to 1.5 m diam., and that cpDNA data separates Tasmanian and Rekohu L. fl exicaule from New Zealand samples (Fig. 4). Further study, including obtaining sequences of non-denticulate South Island L. fl exicaule, is warranted. Of the indigenous species, it is perhaps most similar to L. naufragorum, which diff ers by its upright bushy shrub-forming habit with ascending to erect stems (Fig. 55), more sharply serrated pinnae of the leaves; racemes that are not so clearly leaf-opposed and are much longer (up to 150 mm, cf. 50 mm long), fl owers that have four rather than two stamens, and emarginate petals that are longer than the sepals (see Garnock-Jones and Norton (1995)

for other distinctions).
Lepidium fl exicaule is however, commonly confused in the fi eld with the naturalised L. didymum L. and L. coronopus (L.) Al-Shehbaz, species with which it sometimes grows. Both L. didymum and L. coronopus have long been treated in New Zealand under the segregate genus Coronopus but Al-Shehbaz et al. (2002) reinstated both species within Lepidium. Lepidium fl exicaule can be distinguished from L. didymum and L. coronopus by the valves of the silicles, which are smooth and which dehisce into 2 valves to leave a persistent replum. Th e silicles of L. didymum and L. coronopus have a reticulate-ridged or warty surface, and do not dehisce along the valves; in L. didymum they split into 1-seeded segments and in L. coronopus they are indehiscent.
Ecology. Th e ecology of Lepidium fl exicaule was described in some detail by Garnock-Jones and Norton (1995). Th ey concluded that it is a strictly coastal species of turf communities, rock crevices and the strandline of bouldery beaches. Lepidium fl exi- caule has also colonised tracksides, the compacted peaty ground of seal colonies, bird roosts, and even bird nests, where plants presumably arose from fruiting material that had been used for nest construction. In most locations, the species is found within the spray zone and always in sites prone to frequent disturbance.

Conservation Status. Previously
Lepidium fl exicaule had been assessed 'Th reatened / Nationally Vulnerable CD, EF ' (de Lange et al. 2009). Based on current evidence, this ranking is no longer appropriate, as Rekohu populations of this species (treated in that paper as an unnamed entity L. aff . fl exicaule but now included in L. fl exicaule (see recognition above)) and all of the monitored New Zealand L. fl exicaule populations are in decline, while the past total area of occupancy had been grossly under estimated. Based on current knowledge, Lepidium fl exicaule is more appropriately assessed as 'Th reatened / Nationally Endangered' (criterion A(3/1) of Townsend et al. (2008)) because the total area of occupancy of the sum of the populations is < 10 ha, and evidence obtained from monitoring indicates that there is an ongoing decline predicted to be up to 50% over the next ten years due to increased plant mortality (causes of which are as yet unknown) and loss of habitat caused by competition from weeds and coastal erosion. To this assessment we append the qualifi ers 'CD' (Conservation Dependent -because virtually all known populations are being managed), 'EF' (Extreme Fluctuation -because monitoring also indicates that this is a species that, despite the decline, is naturally prone to seasonal population fl uctuations) and 'TO' (Th reatened Overseas -because, for the time being, we include Rekohu and Tasmanian plants within L. fl exicaule (see recognition), and in Tasmania L. fl exicaule is also seriously threatened (A. Buchanan pers. comm.)).
Hybridism. On the Chatham Islands, Lepidium fl exicaule is occasionally found growing with L. oligodontum and L. rekohuense, and putative hybrids have been collected when these species occur together. Th ere is good evidence for the hybrid Lepidium fl exicaule × L. oligodontum (represented by AK 294939, P. J. de Lange CH391; AK 294942, P. J. de Lange CH394; AK 295119, P. J. de Lange CH442 & J. W. D. Sawyer). Th is hybrid is known from two sites on Chatham (Rekohu) Island where it is found in association with both parents. Like the parents, the hybrids have a decumbent growth habit. Th ey share with L. fl exicaule pinnatifi d rosette leaves, but vegetative features of L. oligodontum are evident in the mid to upper stem leaves that are mostly spathulate to cuneiform (but with occasional deeply lobed margins to weakly pinnatifi d leaves that are similar to L. fl exicaule). Th e fl owers of the hybrids have variable stamen numbers ranging from 1-5 per fl ower, compared to 2 in L. fl exicaule and 2-4-6 in L. oligodontum, while pollen stainability varied from 80 to 85%. Seedlings raised from AK 295119 presented a bewildering array of foliage types grading into either parent. Unfortunately, through mishap these plants failed to reach maturity, so vouchers showing this are unavailable.
Evidence for L. fl exicaule × L. rekohuense is less convincing. Th is putative hybrid is known from only one gathering (AK 295155, P. J. de Lange CH426 & J. W. D. Sawyer) which was collected in January 2006 from Cape Young, Rekohu, at a site where both parents grew. In the fi eld, this specimen resembled L. fl exicaule closely except that the plant was distinctly leafy and the upper stem foliage was more copious and larger than is usual for L. fl exicaule. As with all Rekohu gatherings of L. fl exicaule, this specimen lacked stem and marginal leaf denticles. Furthermore, as with both parents, there are two stamens per fl ower, and pollen stainability of this gathering was 98%. No fruiting material was present, and on a subsequent search of this site in 2007 the putative hybrid and its parents had gone. Th is specimen may be a very well grown example of L. fl exicaule but, as it appears anomalous alongside other Rekohu, New Zealand, and Australian gatherings of that species, we prefer to treat it as a putative L. fl exicaule × L. rekohuense hybrid.
Th e putative hybrid L. fl exicaule × L. oleraceum has also been recorded once, as a spontaneous plant arising in cultivation at the Auckland Regional Council Botanic Gardens (AK 228296 S. P. Benham s.n., AK 223492 S. P. Benham s.n.). Th e hybrid grew in a site where both parents (Scots Beach L. fl exicaule and Stephens Island L. oleraceum) were planted together. Th e hybrid appeared from a seed lot gathered from the cultivated L. fl exicaule plant. It initially appeared to be diff erent because it was much larger and more vigorous than L. fl exicaule, which it otherwise closely resembled (S. P. Benham pers. comm.). Over time, the hybrid developed weakly ascendant branches with rhomboid deeply serrated to pinnatifi d mid and upper stem leaves, and fl owers with 1-6 stamens. Th e hybrid appeared to be fully fertile, and seedlings raised from it showed clear segregation back to either parent (S. P. Benham pers. comm.). Again through mishap, these plants were lost before they reached maturity and specimens could be taken. Etymology. Th e specifi c epithet 'juvencum' from the Latin for 'youthful' refers to plants fl owering and fruiting only a few months after germinating from seed.
Recognition. Lepidium juvencum is distinguished by open, sprawling, straggling growth habit, and the stems are often trailing on the ground (Figs 46, 47). Th e leaves are elliptic, elliptic-oblanceolate, obovate to elliptic-obovate, and have attenuate bases; the marginal teeth are usually small and are mostly confi ned to the distal third of the leaf (Fig. 48). Th e silicles are most like L. oleraceum but diff er by possessing a rounded, slightly notched apex (Fig. 49).
Ecology. Lepidium juvencum grows in open and often disturbed sites on sandy soil. At Long Beach it grows on stabilised sand at the base of a cliff immediately behind the beach, and at Herekopere Island it occurred on a steep coastal bank comprising sandy loam. On Kaimohu Island, in the south-western Titi group, what is probably Lepidium juvencum has been recorded growing with L. limenophylax on the margins of coastal scrub. Lepidium juvencum has also been established and is self-seeding at the Yelloweyed Penguin Trust revegetation project at Tavora Reserve (near Palmerston, North Otago). Here it grows among Euphorbia glauca G.Forst. and Ficinia spiralis (A.Rich.) Muasya et de Lange on sand dunes at the back of the beach.
Conservation Status. Lepidium juvencum is presently known from three wild localities, and, while the number of plants at these localities is not known with certainty, it is estimated that there are < 100 in total. It is also acknowledged that L. juvencum is likely to be more widespread, especially in the vicinity of Stewart Island/Rakiura, and further fi eld surveys are required. Recruitment is likely to be sporadic, and the small and sparse populations are especially vulnerable to stochastic events, such as storm surges, that could erode its habitat. Based on our current knowledge of this species and using the New Zealand Th reat Classifi cation System (Townsend et al. 2008), L. juvencum qualifi es as 'Th reatened/Nationally Critical' (either criterion A(1) or A (3) of  applies). We recommend appending the qualifi ers, 'CD' (Conservation Dependent -because the species has been deliberately established at Tavora Reserve, where it is actively managed), 'DP' (Data Poor -to refl ect uncertainty over its distribution), and 'RR' (Range Restricted -because it is naturally confi ned to a geographically small part of the New Zealand Botanical Region). de  Etymology. Th e specifi c epithet 'limenophylax' derives from Greek (Ό λιμνοφυλαξ) and, though it strictly means 'lake watcher' (C. Blackford pers. comm.), it is used here in the sense of the alternative meaning of "harbour watcher or coast-guard", which Green (1994;p. 535) states is the more appropriate derivation (see his comments on the etymology of the Norfolk Island endemic orchid Phreatia limenophylax (Endl.) Benth.). It is given in allusion to the preferred habitat of L. limenophylax on the Snares, i.e. the low turf confi ned to the exposed ends of rocky headlands jutting out into and overlooking the sea. Historically, it was on these exposed headlands that the World War II coastwatchers were posted to keep watch for signs of enemy shipping (Fleming 1948, McEwen 2005.

Distribution (
Recognition. Lepidium limenophylax is recognised by the decumbent growth habit (Figs 51, 52), with plants developing a distinct woody network of branches. New season's growth arises from the nodes left from the previous season's growth. In this species, the leaves are consistently lanceolate, narrowly lanceolate, linear lanceolate or linear (Figs 50, 51B, 52), though slightly broader in seedlings and basal rosettes. Th e fl owers have two stamens (Fig. 53). It is morphologically most similar to L. rekohuense, which also has a decumbent growth habit and fl owers with two stamens. Lepidium limenophylax has lanceolate, narrowly lanceolate, linear lanceolate or linear leaves, rather than the spathulate to narrowly ovate or elliptic leaves of L. rekohuense. Furthermore, the silicles of L. limenophylax are elliptic, rhomboidal and unwinged (Fig. 54), rather than orbicular, obovate to ovoid and winged.
Ecology. On the Snares, Lepidium limenophylax has been described as an occasional component of cliff top vegetation where it is stated to grow in association with living or dead Poa astonii Petrie pedestals (and on occasion in association with Poa tennantiana Petrie and Hebe elliptica (G.Forst.) Pennell), often near the feeding sites of brown skuas (Stercorarius lonnbergi Mathews, 1912) and occasionally around the nests of Buller's mollymawk (Th alassarche bulleri (Rothschild, 1893)) or on dry peaty cliff tops (sometimes within the sandy quartz derived from the underlying granite) (Fineran 1969, Hay et al. 2004, Lake and Evans 2011. Further inland, it was also reported by Fineran (1969, p. 3) 'as uncommon... sometimes present on sites of old abandoned penguin rockeries, where it usually grew with great vigour'. Brian Rance (in litt.), on a visit to the islands in 2000, found that, as with Fineran (1969), L. limenophylax was virtually confi ned to the north-western cliff s. More recently, Lake and Evans (2011) failed to fi nd it at any inland sites. All these authors and individuals noted Lepidium limenophylax as uncommon in other habitats, though a few plants have been noted near Boat Harbour growing in association with Poa astonii, Asplenium obtusatum G.Forst. and Hebe elliptica in semi-open situations caused by Hooker's sea lions (Phocarctos hookeri (Gray, 1844)) (Lake and Evans 2011; B. D. Rance pers. comm.). Very little is known about the associations of Lepidium limenophylax on Kaimohu Island, though herbarium notes suggest that it grew there in association with what is probably Lepidium juvencum (see comments under that species), and B. A. Fineran (pers. comm.) noted that this species was present mostly along ridge crests and above cliff faces, always in association with the feeding sites of skua. Brian Rance (in litt.) has also observed plants apparently matching L. limenophylax on several of the Titi (Muttonbird) islands south west of Stewart Island, usually growing along tracks in association with a sparse ground cover of Apium prostratum subsp. prostratum var. fi liforme (A.Rich.) Kirk (Lake and Evans 2011). Lake and Evans (2011) were unable to count or estimate population size stating that 'it was not possible to estimate the number of plants due to the tendency [of the species] to grow in patches and not [as] isolated plants'. Nevertheless, they reported 'a good mix of mature and juvenile plants'. Outside these islands, the status of this species on the Auckland and south-western Titi (Muttonbird) Islands is unknown. Previously, Lepidium limenophylax, as L. aff . oleraceum (c) (CANU 5995; Snares), had been listed by de Lange et al. (2009, p. 89) under Appendix 2 Taxonomically Indeterminate Listings, as 'Acutely Th reatened/Nationally Critical' with the qualifi ers 'IE' (Island Endemic) and 'RR' (Range Restricted) appended. Th at assessment used data provided by B. D. Rance, on a visit to the Snares Islands in 2000. Lake and Evans (2011) provided a detailed update on the information obtained by Rance during a three week stay on the Snares between September and October 2010. Th ey concluded that the species was in good health, and much more abundant than had been believed previously. While these observations suggest that this is probably a narrow-range species secure within its Snares stronghold, we recommend retention of the current threat listing of de Lange et al. (2009) as a precautionary measure because: 1. Accurate information about the status of this species outside the Snares is absent (i.e. it is not a Snares endemic), 2. Th e total area of occupancy on the Snares is < 1 ha (criterion A(3) in Townsend et al. (2008)) and 3. Th ere is still no trend data available for the species. However, to refl ect the observations of Lake and Evans (2011) and our ongoing uncertainty about population trend and the status of this species outside the Snares we recommend that the addition of the qualifi er 'DP' (Data Poor) to the current threat listing. It should also be noted that because L. limenophylax is known from a number of islands and island groups it does not meet the defi nition of 'IE' in . Lepidium limenophylax also should be qualifi ed 'CD' (Conservation Dependent) because the largest known population occurs within a Nature Reserve and World Heritage site (Th e Snares) where it is vulnerable to the spread of disease, weed and rat incursions, all of which would have a profound impact on its long-term security. Because of these threats, which put at risk not only the Lepidium but all other Snares terrestrial biota, access to the islands is strictly controlled by the New Zealand Department of Conservation, which also undertakes regular biosecurity inspections of the island group.

Etymology.
Th e epithet 'naufragorum', the genitive plural form of 'naufragus' meaning a 'castaway or ship wrecked person', was given by Garnock-Jones and Norton (1995) to commemorate a sealing gang that was set down on the type locality of the species, the Open Bay Islands, in 1810 where they were left for four years until they were rescued.
Recognition. Lepidium naufragorum is easily distinguished from all other Lepidium species in New Zealand by the upright bushy shrub habit (with plants dying down to a central root stock in winter), erect stems, mostly sharply serrated pinnatifi d rosette and stem leaves (rarely entire) (Fig. 55), and by the emarginate silicles (Fig. 56). Of the New Zealand species, it is most similar to L. fl exicaule in which it was included by Garnock-Jones (1988) before fi eld work and further study recognised its distinctiveness. For differences between L. naufragorum and L. fl exicaule, see under L. fl exicaule.
Although most plants of L. naufragorum have pinnatifi d foliage, occasional specimens with simple or weakly pinnatifi d leaves are also known, and these can at times be locally common (Knightbridge and Newton 2006). It was plants such as these that were the basis for literature records (e.g. Burrows 1972) and collectors such as L. Cockayne and B.C. Aston (WELT!) referring this species to Lepidium oleraceum and L. banksii (as var. ovatum). From Lepidium oleraceum these forms can easily be distinguished by the seasonal growth habit (with plants dying back to the rootstock over winter), and by the distinctly emarginate silicles. So far, Lepidium oleraceum has not been found within the Westland range of L. naufragorum. Th e confusion with Lepidium banksii partly stems from these early collectors' uncertainty as to what this species is. In any case, non-pinnate leaved forms of L. naufragorum are readily distinguished from L. banksii by their much smaller (2.8-4.0 × 2.3-3.2 mm) broadly elliptic and shallowly notched silicles. Th e silicles of L. banksii are much larger (4.5-5.5 × 4.0-5.0 mm), broadly ovate and deeply notched (Fig. 25).
Ecology. Th e ecology of Lepidium naufragorum was described in some detail by Garnock-Jones and Norton (1995) who noted that it preferred sites frequented by seals and nesting sea birds or sea bird roosts. Th ey concluded that the species required these species not only for nutrient enrichment but also because their disturbance kept the habitats of this species open. Th e authors also noted that Lepidium naufragorum was restricted to base-rich substrates.
Conservation Status. Lepidium naufragorum has a conservation status of "Th reatened / Nationally Vulnerable CD, RR " (de Lange et al. 2009). Based on current evidence this ranking is still appropriate. Etymology. Th e epithet "oblitum" from the Latin meaning "forgotten, disregarded or neglected" refers to the circumstances in which this species was fi rst discovered. At an early stage of this revision L. oblitum had been included within the range of Lepidium panniforme, with some plants treated as possible hybrids between L. panniforme and L. oleraceum. It was only following critical analysis of plants grown in New Zealand from samples collected from Mangere Island that the distinctive nature of this species was realized using DNA and morphological data.
Recognition. Lepidium oblitum is recognised by its sprawling growth habit (Fig.  58); by the branches often widely spaced, trailing and mostly devoid of foliage except for the ends which are typically densely terminated by foliage (Fig. 57); and by the prominently notched silicles (Fig. 62). Of those species with which it grows (L. oligodontum, L. oleraceum, L. panniforme and L. rekohuense), it can be easily distinguished from L. oleraceum by its sprawling growth habit, fl owers with 2-4 stamens, and distinctly notched silicles (Figs 58, 61, 62).
Lepidium oblitum and L. panniforme grow together on Mangere Island (Fig. 63). As they have similar silicles and fl owers with 2-4 stamens but very diff erent foliage, L. oblitum was initially thought to be hybrid between L. panniforme and L. oleraceum. At that time it was not realised how widespread L. oblitum is, nor that it comprised stable, true-breeding populations (G. Houliston unpubl. data). It was only following DNA-based investigations undertaken for this paper that the putative hybrid status was rejected and its status as a species recognised. Nevertheless, L. oblitum is still closely allied to L. panniforme, and both also share a relationship to the extinct L. obtusatum, a relationship evident by all three species sharing prominently notched silicles. Lepidium oblitum has widely spreading, lax (rather than erect to suberect), sprawling stems and rosette leaves that distinguish it from L. panniforme (Figs 82, 83). Th e leaves of L. oblitum are shallowly toothed but they are never as deeply serrated or lacerate as those of L. panniforme (Figs 81,83,84). DNA data suggests that gene fl ow between L. oblitum and L. panniforme has occurred at some sites on Mangere Island (G. Houliston unpubl. data). Lepidium oblitum also grows with L. rekohuense on Rabbit Island. From that species it is easily separated by its smaller growth habit; by the glabrous rather than sparsely papillate-hairy upper branch stems, and glabrous rather than fi nely hairy infl orescences (the pedicels of L. oblitum may, very rarely, bear a few caducous glandular hairs). Fur-thermore, the fl owers of L. oblitum have 2-4 rather than 2 stamens, while the silicles of L. rekohuense are also mostly orbicular (rarely obovate) and slightly larger (up to 4.1 × 4.0 mm cf. 3.3 × 3.0 mm).
Ecology. Lepidium oblitum is one of three Lepidium species recorded from Mangere and Rabbit islands. On these islands, it is frequently found growing in association with petrel burrows, and on exposed, often wind eroded cliff faces, associated soil blowouts and ephemeral drainages and seepages. On Mangere Island, it is perhaps most common among the boulders and rocks forming the northern summit of the Top Plateau. In these habitats, L. oblitum commonly grows with Asplenium obtusatum, Festuca coxii Conservation Status. Lepidium oblitum is so far known only from the Chatham Islands group where it has been collected from Mangere and Rabbit Islands (Fig. 63). On Mangere Island (a Nature Reserve, with strict permit controlled access) it is known from several populations, the largest of which occurs along the northern western cliff faces of 'Top Plateau'. Th at population may number in the tens of plants; the only other ones known occur in steep runnels draining the western cliff s south of the main Landing area. When visited by PdL and PBH in 2006, there were fewer than 10 adults in total. Subsequent visits to Mangere Island by Department of Conservation rangers suggest that there has been little change in the population sizes on that island. However, hard data is unavailable and so it would inappropriate to infer from these observations that L. oblitum populations on Mangere Island are stable. On nearby privately owned Rabbit Island, two plants of L. oblitum were seen in 2006 and the island has not been visited by Department of Conservation staff since.  Th erefore, based on available information Lepidium oblitum is known from two sites, totalling 4 populations, which collectively are unlikely to exceed 100 mature individuals. Further, the total area of occupancy is considerably less than 1 ha. On the basis of that data, using , Lepidium oblitum qualifi es as "Nationally Critical" (using either criterion A(1) or A(3)) because there are fewer than 250 mature individuals known and the total area of occupancy is ≤ 1 ha. We favour criterion A(3) because a precise survey of the numbers of plants of L. oblitum has yet to be carried out, and available data (beyond that obtained by PdL and PBH) is unreliable because of understandable past confusion with L. oleraceum s.s. and L. panniforme.
Th is conservation assessment should also be qualifi ed 'CD' (Conservation Dependent) because Mangere Island is subject to ecological restoration, and ongoing surveillance to ensure it remains predator-free. Should rodents establish on the island, they will aff ect that island's Lepidium species, both directly through browse and indirectly through predation of the sea birds that maintain this species habitat. Other necessary qualifi ers include 'DP' (Data Poor) because of the lack of precise information on population size and trend data, 'IE' (Island Endemic), and 'RR' (Range Restricted) because of its precise habitat requirements and geographically narrow-range.

Lepidium obtusatum Kirk, Trans. & Proc. New Zealand Inst. 24, 423, (1892) http://species-id.net/wiki/Lepidium_obtusatum
Holotype. New Zealand (Fig. 64) Kirk), type locality, and labelled in the naming authors hand could be located in the WELT (the main herbarium where Kirk material has come to be lodged) so this sheet is regarded here as the holotype.
Etymology. Although the exact meaning of the species "obtusatum" was not given by Kirk (1891) it seems that he took the name from the shape of the hypogynous glands (i.e. the nectaries) which he specifi cally noted (Kirk 1899) are "very short and obtuse".
Representative Specimens. New Zealand: Entrance to Port Nicholson, n.d., T. Kirk s.n Distribution (Fig. 44). Endemic. New Zealand, North Island, Wellington Harbour, Miramar Peninsula, in a small area of coastline between Worser Bay and Breaker Bay.

L. amissum
Ecology. Very little is known about the habitat preferences and ecology of this species. It apparently grew on rocky headlands, sea cliff s, coastal rocks and beaches (Kirk 1891(Kirk , 1899Cockayne 1921), and some herbarium specimens state that it also grew on sandy and gravel beaches at the high tide mark (e.g., WELT SP030104, SP030107). Th e species evidently had a most remarkable appearance when fresh, Cockayne (1921) noting that 'in appearance L. obtusatum represents, even when in fl ower, a huge crusty saxifrage. Its leaves arranged in rosettes are very thick, glossy and rather dark green'. Its thick, fl eshy leaves may have the been the basis for a vernacular name 'sea kale lepidium' recorded by J. H. MacMahon (see WELT SP081926).
Conservation Status. It is a matter of some irony that when it was described, Kirk (1891) stated, 'happily for Lepidium obtusatum, it grows in a few spots which are inaccessible to sheep, so that it will probably hold its ground for many years'. Th is prediction did not come to pass; the species was fi rst treated as 'Presumed Extinct' by Williams and Given (1981), a status that has been maintained in all subsequent threatened indigenous vascular plant listings to the present (see Cameron et al. 1993;de Lange et al. 2009). Th e factors leading to the demise of Lepidium obtusatum are well documented (Cockayne 1921;Norton et al. 1994;. Th e species was a narrow-range endemic whose location on the margin of an actively expanding city meant that it was extremely vulnerable to habitat destruction (see de Lange et al. 2010a;de Lange et al. 2010c). Nevertheless it seems that, at least initially, it was very common within the small area it occupied, and Cockayne (1921) mentioned that the quarrying of rock from its cliff habitat, had, temporarily at least, caused an expansion in the population through the spread of 'underground stems'. Nevertheless, by 1951 it had gone extinct. From available evidence it seems that the biggest factor was not habitat loss but over collection (Norton et al. 1994). Herbarium specimens show that following its formal recognition by Kirk (1891), it was repeatedly gathered by botanists who made copious collections (often gathering whole plants). Most of these gatherings occurred between 1900 and 1939 (and they intensifi ed between 1920 and 1939) ending with the last known collection in 1951 ( . During this period it is also likely that the modifi cation of most of its habitat for housing and defense structures associated with military operations at Fort Dorset would have impacted on the population, as would the spread of weeds, especially those derived from nearby urban gardens. Today, all of the former range of L. obtusatum is now choked in a rank growth of weeds, most especially veldt grass (Ehrharta erecta Lam.).
Hybridism. A suite of specimens (WELT SP081927A, B, C, J. H. MacMahon s.n., 1931, Seatoun) are tentatively identifi ed here as a putative natural hybrid between Lepidium obtusatum and L. oleraceum. Th ese specimens, which appear to be pieces split from the same plant seem to have had an erect, shrubby growth habit, with somewhat woody, fl exuous branches, while the basal and upper cauline leaves are narrowly elliptic to lanceolate, deeply and sharply incised to weakly pinnatifi d. Th e fl owers, though in poor order, have four stamens (pollen stainability 0%). Only a few silicles in the specimens are mature and these measure up to 4.05 × 2.75 mm, are suborbicular to ovate, narrowly winged, with the apex weakly notched. Signifi cantly, they contain no seeds. Th e putative parentage is based on the knowledge that both L. obtusatum (see collections cited above) and L. oleraceum were sympatric at Seatoun (WELT SP09712, W. R. B. Oliver s.n.) and that no other Lepidium species was known to be present in that area at that time. Th e erect, shrubby growth habit is a feature of Lepidium oleraceum, while morphologically the silicles are a close match for L. obtusatum. Th e shape of the cauline leaves fi ts within the range seen in L. oleraceum, though their degree of dentition also approaches that seen in L. amissum, L. banksii and L. panniforme. Neotype (Fig. 69, designated here): S, Sparrman, Nova Zelandia. Notes: Nicolson and Fosberg (2004) were unsure about typifi cation of L. oleraceum G.Forst. ex Sparrm., and the name has not been typifi ed. Th ey listed two Sparrman specimens at S, but did not select a type specimen since neither of these appeared to have been gathered from cultivated plants. Given the revised taxonomic treatment presented here for the L. oleraceum complex, it is necessary to be certain of the application of L. oleraceum G.Forst. ex Sparrm. Th erefore, we select one of the Sparrman sheets as neotype.

Lepidium oleraceum
When describing L. oleraceum, Sparrman (1780)  It is unusual that this description was based on cultivated plants since it was the practice of A. Sparrman and G. Forster to prepare descriptions from freshly collected material during Captain Cook's second voyage (Forster andForster 1776 (see Edgar (1969) for translation). Plant material of L. oleraceum, along with other edible, herbaceous coastal plants was frequently collected during the New Zealand part of the voyage to be utilised for their antiscorbutic properties (de Lange and Norton 1996). Th erefore, there would have been ample opportunity to prepare the description from freshly collected material. Indeed, as noted by Nicolson and Fosberg (2004), the published description of the later homonym L. oleraceum G.Forst. diff ers from that of L. oleraceum G. Forst. ex Sparrm., and it is most likely that the description of the homonym was prepared on the voyage from fresh material.
Th e wild locality of the neotype of L. oleraceum is not known, except that it is from New Zealand. While anchored in Queen Charlotte Sound during May 1773, G. Forster noted 'the antiscorbutic plants grew on every beach' and 'we immediately gathered vast quantities... of a well-tasted scurvy-grass (Lepidium)' (Forster 1777, p. 126). Th ey also collected it from Long Island (Hoare 1982, vol. II, p. 287), and saw it in the vicinity of Maori settlements: 'near all the places where the Indians have their huts, there grows a kind of Scurvy-Grass or Lepidium' (Hoare 1982, II, p. 297). Th erefore, since J. R. Forster, G. Forster, and A. Sparrman would have had ample opportunity to collect L. oleraceum in Queen Charlotte Sound it is very likely that the neotype was collected from there. Dusky Sound, Fiordland, is the only other location where landfall was made in New Zealand during Cooks 1773 voyage, but the species is not known from the Fiordland coastline so is unlikely to have been collected from there.
Etymology. Th e exact meaning of the species 'oleraceum' was not given by Sparrman (1780). However the epithet derives from the Latin 'oleraceus' meaning 'used for herbs or vegetables' (Taylor 2002), a point that Sparrman had alluded to in a brief note he wrote in Latin on the backside of the neotype sheet.
= Lepidium oleraceum var. acutidentatum Kirk, Stud. Fl. N.Z., 35 (1899) Type Collection:'NORTH and SOUTH Islands; STEWART Island; the SNARES; AUCKLAND Islands; CHATHAM Islands. In places near the sea.' Neotype (Fig. 70, designated here): T. Kirk 367, March 1869, Taranga Islandslabelled as 'Lepidium oleraceum var. incisum' in Kirk's hand. WELT SP027646! Notes: Kirk (1899) described var. acutidentatum thus 'stems with slender leafy branches. Leaves 1in.-1½in. long, narrow, cuneate or oblong-spathulate, the upper portion acutely serrate or almost dentate'. Plants matching this vague description span large parts of the range of L. oleraceum s.l. Our searches of the herbaria where Kirk traditionally lodged specimens (see comments by de Lange 2007) have failed to fi nd any specimens labelled by Kirk as var. acutidentatum. However, there are a range of specimens in WELT (WELT SP027646!, SP027647!, SP027648!, SP030087!) bearing in Kirk's hand the manuscript names 'var. erectum' and 'var. incisum', which, being unpublished, have no nomenclatural status whatsoever. Th ese specimens, along with the generally confusing notes provided by Kirk (1899, p. 34-35) in his entry for L. oleraceum, suggest that Lepidium was a genus he was still working on close to his death  (Moore 1973, but see also comments by de Lange 2007 andGardner 2002). At the time of his death, Kirk was using these manuscript names on his herbarium specimens, but he was either undecided on his fi nal choice of epithet or had yet to relabel his specimens as var. acutidentatum, a name which was then later published posthumously when his unfi nished Flora manuscript was uplifted and published (Kirk 1899). Despite this confusion, Kirk's description of var. acutidentatum matches most closely those WELT specimens (T. Kirk 367, WELT SP027646, T. Kirk 368, WELT SP027647) that he had collected from Taranga Island, which is the largest of the Hen and Chicken Islands group and labelled 'var. incisum'. Notably, both specimens match his protologue as to the description of var. acutidentatum, e.g., 'slender leafy branches... Leaves 1in.-1½in. long, narrow, cuneate…upper portion acutely serrate'. Th erefore, in the absence of any suitable material for a lectotype we designate T. Kirk 367 (WELT SP027646) as the neotype of L. oleraceum var. acutidentatum Kirk. Etymology. Th e meaning of the epithet 'acutidentatum' was not given by Kirk (1899). However, the epithet as indicated by the protologue probably derives from the sharply toothed leaves.
= Lepidium oleraceum var. frondosum Kirk, Stud. Fl. N.Z., 34 (1899) Type Collection: 'Banks and Sol., MSS and Ic.' Lectotype (Fig. 71, designated here): WELT SP063976/A! IsolectotypeWELT SP063976/B! Notes: Kirk (1899, p. 34) despite his rather confused and evidently incomplete treatment of Lepidium oleraceum nevertheless clearly described Lepidium oleraceum var. frondosum in the following manner: 'var. frondosum, (sp.), Banks and Sol. MSS. and Ic. Robust, leaves large, fl eshy, broadly cuneate-oblong or oblong, sometimes 3in-5in. long and 1in. wide, sessile or narrowed into a broad petiole, serrate'. Although no specimens or locations were cited by Kirk in his protologue, his wording makes it clear that any specimens collected by Banks and Solander and labelled by them 'Lepidium frondosum' and any accompanying illustrations made from these specimens constituted the type collections from which he formally recognised his L. oleraceum var. frondosum. Th erefore, we have selected WELT SP063976/A as lectotype because this sheet, comprising two fl owering and fruiting, leafy specimens, is labelled both by Kirk 'Lepidium oleraceum Forster Prodr. N. 248 var. frondosum' and also in Solander's hand as 'Lepidium frondosum Mscr'.
Etymology. Th e meaning of the epithet 'frondosum was not given by Kirk (1899) who took the name from the unpublished Banks and Solander manuscript. Nevertheless it seems reasonable to assume that the epithet refers to the large, much tooth, arched leaves typical of the robust forms of L. oleraceum that Kirk referred to this variety.
Notes: Th ellung (1906, p. 293) described L. oleraceum var. serrulatum from a single undated gathering which, according to the specimen details, came from 'New River' where it was collected by Godey. Allan (1961, p. 177) equated this location thus: 'S[outh Island]. New River estuary, Riverton, Southland'. Th e holotype bears two labels, one written in the hand of the naming author, Th ellung, and the other in an unknown hand, presumably Godey, which reads 'Lepidium oleraceum Forst var. New River, Nouv. Zélande, Godey'. We have been unable to fi nd out who the collector Godey was.
Th ellung distinguished his new variety from L. oleraceum s.s. thus: 'folia obovata, a medio ad apicem regulariter subtiliter et acute serrator' meaning 'foliage obovate, regularly and acutely, fi nely serrated from middle of the leaf to the apex'.
Signifi cantly the silicles of the holotype are acute rather than notched, a condition seen only in L. castellanum and L. oleraceum, neither of which are known from the southern South Island. For this reason, we suspect that the gathering was collected from the northern part of New Zealand, rather than the southern South Island, and that Allan's interpretation of location of 'New River' is incorrect. We also suspect that the exact location of the 'New River' will now never be known. In all probability this name may have simply been one used locally for some other part of New Zealand.
Th e obovate leaves of Lepidium oleraceum var. serrulatum place this variety within L. oleraceum (as treated here) rather than L. castellanum. Within L. oleraceum s.s., there are a few vegetatively similar, obovate-leaved plants with fi nely serrated leaf margins that approach the extreme condition seen in var. serrulatum. Th ese gatherings all come from off shore islands in the northern part of the North Island (e.g., Th ree Kings Group, A. E. Wright 6072, AK 173005; Rock Stack north of Motuopao Island, R. Parrish s.n., AK 196229;Motutakapu Island, R. Parrish s.n., AK 209112). However none of those obovate-leaved plants have such prominently serrated leaf margins as the holotype of var. serrulatum. Based on these observations, we think that var. serrulatum was collected from somewhere in the northern part of the range of L. oleraceum, and that it represents nothing more than an extreme form of that species, which even following our treatment here remains a highly variable species.
Ecology. Much has been written about the ecology of Lepidium oleraceum, of which the summary provided by Norton et al. (1997) still applies despite our segregation here of 10 new species from it. Lepidium oleraceum is intimately associated with sea-bird trails, roosts and nesting grounds. Plants are dependent on these birds not only for the habitats they create through disturbance but also the nutrients they bring from the sea in the form of guano and discarded or regurgitated food, and because these birds assist with seed dispersal. Seed dispersal is most readily facilitated by birds because Lepidium oleraceum, as well as growing around bird nests and burrows, is often used for nesting material. In this way, the seeds, which are mucilaginous when wetted, can easily stick to the feathers and feet of birds.
Lepidium oleraceum tends to be short-lived, with individual plants rarely persisting for more than three years in the wild and up to fi ve in cultivation. It has also been observed that some wild populations are prone to sudden crashes, and may even completely die out, only to reappear some years later. It is unknown whether the reappearance is from a residual seed bank or from seeds dispersed by birds from another site. Plants may even behave as annuals in some locations because, despite its predilection for open, drought-prone coastal habitats, L. oleraceum does not relish drought. Norton et al. (1997) stressed the importance of sea birds in providing the nutrient regime necessary to sustain the species, arguing that the loss of sea bird breeding grounds from large parts of the country best explains the rapid demise of Lepidium oleraceum. However, while that view is probably still valid, some anomalies remain to be explained. For example, Lepidium oleraceum appears to have always been scarce on the Th ree Kings Islands, which are naturally predator-free and harbour large sea bird breeding colonies. On those islands, Lepidium is confi ned to white-fronted tern (Sterna striata Gmelin, 1789) and red-billed gull (Chroicocephalus scopulinus (Forster, 1844)) nesting grounds, and is absent from all other sea bird colonies, including the heavily burrowed ground left by breeding black-winged petrel (Pterodroma nigripennis Rothschild, 1893). Th is requires further study, as on many Hauraki Gulf islands Lepidium oleraceum is usually absent from tern and gull nesting sites, and more usually associated with petrels such as grey-faced (Pterodroma macroptera gouldi (F.W.Hutton, 1869)), black-winged, and Pycroft 's (Pterodroma pycrofti Falla, 1933), or with shearwaters (Puffi nus spp.), white-faced storm petrels (Pelagodroma marina Latham, 1790) and common diving petrels (Pelecanoides urinatrix (Gmelin, 1789)). On some islands supporting these birds and gannets (Morus serrator Gray, 1843), Lepidium oleraceum is found only near gannet colonies. Interestingly, Lepidium oleraceum was regarded as one of the most common plants on the foreshore of Aorangi Island, Poor Knights until the removal of pigs in 1936 (Reynolds 1988; de Lange and Cameron 1999), yet now, despite the massive sea bird colonies covering that island, this species is scarce, being mostly confi ned to cliff habitats and recent slip scars (de Lange and Cameron 1999).
Th ese patterns suggest that the overriding need for this species is a combination of nutrient rich soils and frequent habitat disturbance to keep sites free from competition. Yet why Lepidium oleraceum is so uncommon on some naturally predator free islands which appear to meet these criteria, and why it seems to show preferences for particular bird nesting associations that can vary from island to island remains to be elucidated.
It is also possible that Lepidium oleraceum was deliberately cultivated and utilised as a pot herb by Māori who knew the species (and probably those allied to it) as "nau" (see de Lange and Norton 1996). Th e early writings of Banks, Solander and Forster make frequent mention of its abundance near Māori dwellings and settlements and that these people often directed Cook's shore foraging crews to places where it could be collected (Beaglehole 1962(Beaglehole , 1967(Beaglehole , 1968Forster 1777;Hoare 1982;de Lange and Norton 1996). While it could be argued that Lepidium oleraceum fl ourished around coastal human habitations because these were sites of frequent disturbance and nutrient enrichment, the fact that Māori had a specifi c name for the plant and knew that it was edible suggests that they may also have cultivated it, a possibility that needs further critical ethnobotanical investigation.
Conservation Status. Prior to this revision, Lepidium oleraceum was assessed as 'Th reatened/Nationally Vulnerable CD, EF, RR, Sp '. Following the segregation of Lepidium oleraceum s.l. in this paper a new threat listing of L. oleraceum s.s. is now necessary. As circumscribed here, Lepidium oleraceum is now confi ned to the Kermadec Islands, North Island (and adjacent off shore islands), northern South Island and Chatham Islands (Mangere Island only). Within this area, Lepidium oleraceum is only "common" (i.e. occurring in numbers greater than 200 individuals) on a very few islands and is-lets, notably Mahuki Island west of Great Barrier (Aotea Island), Karewa Island in the Bay of Plenty, Waioioi Reef off the west coast of Albatross Point, and on several islands (notably Stephens Island) within the Marlborough Sounds. It is now a very uncommon species elsewhere within this range, with most known populations comprising 50 or less mature individuals. At all known sites, populations are naturally prone to the sudden collapse and boom cycles briefl y described by Norton et al. (1997) and Norton and de Lange (1999). Also, because there are few places where the Lepidium oleraceum s.s. is being closely monitored, available data on natural population fl uctuations is insuffi cient from which to provide an overall general assessment of total population size and stability. Irrespective, the species is still very widespread in northern New Zealand and the Marlborough Sounds, and whilst most of the remaining populations are small (i.e. < 50 mature plants) this may refl ect a natural state of aff airs, as past accounts of this species' abundance do seem to have been exaggerated (see comments by de Lange and Norton 1996). Nevertheless, in the absence of key data on trend and rate of decline, it is diffi cult to provide a meaningful threat assessment, though the situation is not so diffi cult as to recommend a conservative assessment of 'Data Defi cient'. Suffi cient information exists to attempt a threat listing here, from which we recommend a conservation listing of 'Th reatened/Nationally Vulnerable' applying criterion 'C' of Townsend et al. (2008, p. 20-21). Th is assessment is based on data collected and stored by the Department of Conservation-sponsored coastal cress recovery team (P. I. Knightbridge pers comm.), from which we estimate the total population for this species at c. 3000-3500, and a rate of decline of c.10% over the next 10 years. As few wild plants last longer than three years, we have opted for (as recommended by the manual (Townsend et al. 2010)) the rate of loss rather than generation time. To that assessment the qualifi ers 'CD' (Conservation Dependent), 'EF' (Extreme Fluctuations), 'RR' (Range Restricted) and 'Sp' (Sparse) still apply but we recommend the addition of 'DP' (Data Poor) to refl ect the overall absence of trend data.

Lepidium oligodontum
Description (Figs 78-80). Tap-rooted, decumbent, rather fl accid, pungent-smelling, summer-green, succulent, perennial herb forming sparse to densely leafy, circular ± fl at masses up to 1 m diam., and arising from stout, semi-circular, greyish-white or reddish-grey (when exposed) rootstock 6.0-30.0 mm diam. Tap root woody, up to 0.2 m long, deeply descending. Plants dying down to rootstock and/or previous seasons stem nodes, at fruit set or soon thereafter. Stems decumbent, widely spreading, up to 0.8 m long and 3.20-4.56(-5.20) mm diam., ± woody throughout, ± square to somewhat angular-spheroidal in cross-section and prominently ridged on angles (this especially conspicuous when dry), dark reddish-green to dark green when fresh, drying dull grey; stems usually heavily branched from base, branches and branchlets numerous, prostrate, widely spreading, and usually very leafy; basal portion of stems, glabrous. Leaves glabrous, succulent, dark green, green to yellow-green at senescence turning yellow. Rosette leaves 5-10(-14), mostly present in autumn -early spring but not persisting (very rarely so) at fruiting; petioles distinct up to 70 × 2 mm, fl at or slightly concave in cross-section, succulent; lamina narrowly spathulate, cuneiform, obdeltoid, obovate or rarely elliptic-lanceolate, up to 60 × 22 mm, margins entire, or sparingly  dentate in upper ⅓, if teeth present then in 1-3(-5) pairs running to and including the usually tridentate apex, basal teeth often asymmetric, base narrowly cuneate, cuneate to attenuate. Middle stem leaves persistent or not at fruiting; petiole usually distinct (rarely not) up to 14.0 × 1.13 mm, mostly fl at in cross-section, sometimes slightly concave, succulent; lamina spathulate, cuneiform, linear-cuneiform, oblanceolate, narrowly ovate, to narrowly obovate, or rarely orbicular, 10. 6-22.8(-33.2) × 5.4-9.2 (-17.5) mm; margins entire, or sparingly dentate with 1-2-3(-5) pairs of teeth in upper ⅓ apex usually tridentate, basal teeth often asymmetric, lamina base narrowly attenuate, attenuate, cuneate or rarely acute. Upper stem leaves usually without a distinct petiole, petiole if present 1.0-3.6 mm, fl at or slightly concave; lamina narrowly cuneiform, oblanceolate, or narrowly obdeltoid, 7. 6-10.8(-11.9) × 2.7-3.1(-3.5) mm; margins entire or weakly dentate to deeply incised, if dentate or incised then with 1(-2) often asymmetrical teeth present in the upper ⅓, apex entire or tridentate, lamina base cuneate to narrowly cuneate. Racemes (5.0)-9.7(-28.9) mm long, usually congested, elongating up to 60 mm at fruiting, terminal and axillary; rachis and pedicels glabrous; pedicels, erecto-patent to patent,0.82-1.00(-2.08) mm, 2.1-3.5(-6.1) mm long at fruiting. Flower buds dark green to green, apex glabrous. Flowers sweetly fragrant, 1.2-1.5(-2.2) mm diam. Sepals 4, saccate, pale to dark green with a broad white, ± undulose margin, pale to dark green with a broad white, ± undulose margin, deeply concave, adaxially weakly keeled or not; lateral sepals 0.9-1.4 × 0.8-1.2 mm, broadly ovate to oblong, ± overlapping at base, apex rounded to obtuse, adaxial surface mostly glabrous sometimes diff usely papillate, abaxial surface often hairy, hairs patent, weakly fl exuous, 0.1-0.4 mm long, eglandular, mostly shedding at anthesis except near base; median sepals 1.0-1.4 × 0.9-1.2 mm, broadly ovate to oblong, apex rounded to obtuse, adaxial surface glabrous, abaxial surface usually glabrous, rarely sporting a small tuft of patent, eglandular, fl exuous hairs 0.1-0.2 mm long. Petals usually present (occasionally absent) usually equal to or slightly overtopping sepals (rarely > sepals), white, 0.8-1.6(-1.9) × 0.6-1.2(-1.8) mm, patent, clawed; limb broadly obovate, apex obtuse, rarely retuse. , equal. Anthers c.0.14 mm long. Pollen bright yellow. Nectaries 2, subulate, 0.32 mm long. Silicles somewhat fl eshy and distinctly turgid when fresh, on drying collapsing to form a coarse reticulum, broadly orbicular Ecology. Lepidium oligodontum is a strictly coastal, island endemic inhabiting richly manured, frequently disturbed habitats. On Rekohu it is extremely uncommon and now virtually confi ned to coastal cliff s, near-shore rock stacks, and, very occasionally, coastal turf communities, especially in places frequented by New Zealand fur seals (Arctocephalus forsteri Lesson 1828). In these habitats it is usually associated with Disphyma papillatum, Apium prostratum subsp. denticulatum P.S.Short, Crassula moschata G.Forst., Dichondra spp., Hebe chathamica, Leptinella potentillina F.Muell., and Puccinellia chathamica. In a few sites, such as at Point Somes and Ocean Bay, Lepidium oligodontum grows intermingled with L. fl exicaule and hybrids between both species have occasionally been collected where they co-exist. Habitats occupied by L. oligodontum vary on the outer islands, islets and rock stacks of the Chatham archipelago. It is an occasional associate of coastal turf and cliff vegetation on islands such as South East (Rangatira). On the other outer islands (e.g., Th e Sisters (Rangitatahi), Star Keys), islets (e.g., Th e Pyramid (Tarakoikoia)), reefs and rock stacks L. oligodontum is a prominent species of the distinctive guano-dependent vegetation that has developed in the presence of New Zealand fur seals, and sea birds such as the albatrosses (Diomedea sanfordi Murphy, 1917, Th alassarche eremita (Murphy, 1930, Th alassarche bulleri), northern giant petrel (Macronectes halli (Mathews, 1912)), Chatham Island fumar prion (Pachyptila crassirostris pyramidalis C.A. Fleming, 1939), and white-fronted tern (Sterna striata Gmelin, 1789) (Aikman and Miskelly 2004). In these guano-enriched habitats, L. oligodontum is usually found growing in close association with the woody shrub, Leptinella featherstonii F.Muell., and herb, Senecio radiolatus subsp. radiolatus (see de Lange and Sawyer 2008). On Rabbit Island, L. oligodontum was found growing at the entrance to petrel (Pterodroma spp.) and shearwater (Puffi nus spp.) burrows, sometimes in association with L. oblitum and L. rekohuense, and other coastal turf plants such as Disphyma papillatum. Detailed information on the habits of L. oligodontum on the Antipodes is lacking, although herbarium specimens and people 's observations (G. A. Taylor and S. P. Courtney pers. comm.) note that it is seemingly restricted to a few coastal headlands where it grows at the top of steep cliff s in sites kept free of taller vegetation by wind and salt burn. Oddly, despite the abundance of sea birds and seals on the Antipodes, L. oligodontum was not observed growing anywhere near sites frequented by these animals, leading us to infer that the habitat is similar to that occupied along the southern coastal cliff s of Rekohu, e.g., Moriori Creek and Otauwae Point.
Lepidium oligodontum, more than any other member of the L. oleraceum complex, has a very seasonal growth pattern, with most vegetative growth occurring from late winter to summer. Lepidium oligodontum plants are scarcely visible during autumn and early winter because growth virtually ceases and most of the foliage, branches, branchlets and infl orescences wither away. During this time of 'dormancy', plants persist as minute leafy shoots clustered around the rootstock apex, or as a single leafy rosette.
It has also been observed that the annual growth cycle of L. oligodontum seems to be dictated, perhaps more than any other New Zealand Lepidium, by the onset and spread of the reproductive phase of the oomycete Albugo candida (Pers.) Kuntze. Th e spore-bearing pustules of A. candida appear on wild and cultivated plants within weeks of the initiation of spring growth, and by late January/February have usually erupted across most of the stem leaves, fl owers and fruits. Further vegetative growth from the host plant is either completely retarded or aborted. Observations of wild population of Lepidium oligodontum suggest that these seasonal eruptions of Albugo candida have a tremendous impact on this species' vigour, and we have found that the intensity of these A. candida infections make it impossible to maintain L. oligodontum in cultivation. It is not clear if the severity of these attacks in L. oligodontum populations is natural or a consequence of other, as yet undetermined, external factors, or even if the strain of A. candida infecting L. oligodontum is endemic to it (E. H. McKenzie pers. comm.). Whether indigenous, endemic or naturalized, this oomycete is now, at least, a critical constraint on the growing season, fl owering and fruiting of this species, especially in what appear to be ecologically suboptimal sites. Field observations suggest that Lepidium oligodontum is more abundant and the plants clearly thriving in sites where nutrient levels remain high, such as on the guano-enriched Pyramid (Tarakoikoia), or within the seal colonies of the Star Keys and Western Reef (de Lange and Sawyer 2008). Despite the presence of A. candida, their growth season is longer and the plants more persistent than at any of the other Chatham Islands' L. oligodontum sites that we have been able to visit.
Conservation Status. Lepidium oligodontum as L. aff . oleraceum (b) (AK 208579; Antipodes -Chatham Islands) was listed in Appendix 2 of de Lange et al. (2009, p. 89) as "Th reatened/Nationally Critical DP, EF, RR ". With its formal description and recognition, a reassessment of that conservation listing is now appropriate.
Aside from a few, small, isolated islands, islets and rock stacks in the Chatham archipelago and possibly Antipodes Island, trend data gathered by the New Zealand Department of Conservation over the last 15 years confi rm that Lepidium oligodontum, though naturally prone to extreme population fl uctuations, is now in fi nal stages of a major terminal decline on Rekohu (A. Baird unpubl. data).
Based on our knowledge of this species' ecology, its loss from Rekohu appears to be the result of an initial loss of habitat as key sea bird and seal populations on the larger islands of the group went extinct over the last 600 or so years (King 1989). Habitat loss has accelerated since the early 19th century following the settlement of Rekohu by Maori and Europeans whose impacts on the remaining seal and sea bird breeding grounds have been severe (Wills-Johnson 2008;King 2008). Land clearance and ongoing modifi cation of the main Chatham Islands has also facilitated the spread of competing plants that were previously unknown there and which had been either deliberately or inadvertently introduced .
Another factor in the decline of this species is the oomycete Albugo candida, discussed above. It remains unclear whether Albugo candida truly poses a serious threat to indigenous New Zealand Brassicaceae, and in particular Lepidium oleraceum s.l. (Armstrong 2007). Nevertheless, our fi eld observations and the remarks made by collectors on herbarium specimens note that Albugo candida strongly retards the growth of Lepidium oligodontum and that it continues to be a major factor in the decline and loss of many of the smaller populations on Rekohu. Lepidium oligodontum seems to be buff ered from the severity of Albugo candida outbreaks only on less modifi ed outer islands, islets and rock stacks of the Chatham Island archipelago and, presumably, the Antipodes Islands, where there are intact, functional guano-and marine mammalbased ecosystems (de Lange and Sawyer 2008). Furthermore, Albugo candida has prevented Lepidium oligodontum from being successfully cultivated beyond a single season or translocated to new sites (A. Baird, S. Benham, G. Davidson, J. Santos, T. Silbery, and R. Smith pers. comm.).
Aside from Albugo candida, some of the Rekohu populations of L. oligodontum are also subjected to a diverse range of threats. Th ese include browse by feral sheep (Ovis aries Linnaeus, 1758), cattle (Bos primigenius taurus (Linnaeus, 1758)) and possum (Trichosurus vulpecula (Kerr, 1792)). Possums, in particular, have been found avidly browsing Lepidium oligodontum plants growing on near-shore rock stacks at Point Somes and Te Koparuparu Bay on Rekohu.
Following extensive surveys of the Chatham Islands by the Department of Conservation during the summer of 2005/2006, 14 functional subpopulations were recognized, down from 28 known in 1996 (a decline rate of 50 % over 10 years). Th ese gave a total population for that island group of 1321 adults, of which 748 (57%) were confi ned to one site, Western Reef. Th ese fi gures exclude population data for Th e Sisters and Th e Pyramid (Tarakoikoia), which are privately owned and inaccessible to the Department of Conservation. To date, the few reports of L. oligodontum from these islands suggest that it is "common" (R. Emberson and P. Schofi eld pers. comm.). Monitoring at Te Wakaru Island, which has an intact guano and marine mammalbased ecosystem, shows that the population there fl uctuates greatly from year to year. Th erefore, trend data needs to be gathered over several years before any evidence of decline can be distinguished from natural population "boom/bust" cycles. Little is known about the species' status on the Antipodes. In part because these remote islands are infrequently visited, and then usually by ornithologists at times when our data for the Chatham Islands suggests L. oligodontum was already undergoing its seasonal decline. For example, an ornithological party considered L. oligodontum to be very uncommon there in 1995, seeing perhaps only a few tens of plants (G. A. S. Taylor pers. comm.). Th ese were restricted to one site at Reef Point, where, because of their close proximity to a former castaway hut and the unusual habitat they occupied, it was though that the few plants seen may have been introduced there (G. A. S. Taylor pers. comm.). Th is contrasts somewhat to the observations published by Godley (1989) (a botanist) who visited the islands in 1969. Godley recorded L. oligodontum (as L. oleraceum) as being 'common there [at Reef Point] on coastal rocks and occasionally on cliff s', though he then went on to state that 'it was not seen by any of the 1969 party at other possible localities'. Nevertheless, based on what we now know of the Lepidium oleraceum group, the possibility that L. oligodontum is introduced to the Antipodes seems unlikely, and based on the historical observations summarised by Godley (1989, p. 546) it would seem that, if anything, observers were witnessing diff erent stages of the 'boom/bust' cycles observed elsewhere on the Chatham Islands. Nevertheless, as the Antipodes Islands are so infrequently visited, it is important that future visits focus on gathering better data and setting up long-term monitoring of that island group's Lepidium oligodontum population.
In summary, we consider that Lepidium oligodontum is more appropriately assessed as 'Th reatened/Nationally Vulnerable' (criteria B(1/1) of . Th is is based on a population size that is likely to be between 1000 and 5000 mature individuals spread across the Chatham and Antipodes island groups (noting that our best available data (fi eld season 2005/2006) recorded 1321 plants from those accessible locations on the Chatham Islands; that outside those areas on other islands in that group, such as Th e Sisters (Rangitatahi), Star Keys and Th e Pyramid (Tarakoikoia) there are likely to be at least as many individuals; and that there no reliable population data for the Antipodes Islands). Further, it would seem that, aside from the ongoing decline of what are eff ectively non-functional, residual populations on Rekohu, the species is secure on the outlying islands, islets and rock stacks of the Chatham Islands group where the ornithocoprophilous ecosystem remains intact and functional. Accepting the absence of hard data for the Antipodes Islands, observations by fi eld parties suggests that L. oligodontum, while less abundant there, is probably also secure due to the intact nature of the ornithocoprophilous ecosystem there.
To this recommended threat listing we suggest appending the following qualifi ers 'DP' (Data Poor -to refl ect uncertainty over the status of the species on the Antipodes Islands), 'EF' (Extreme Fluctuations -to refl ect the wide population fl uctuations experienced throughout a normal growing season), and 'RR' (Range Restricted -because L. oligodontum is naturally confi ned to a geographically small part of the New Zealand Botanical Region). de  Etymology. Th e name "panniforme" from the Latin meaning "like a shredded rag" (N. G. Walsh pers. comm.) refers to the ragged appearance of the deeply lacerate, tattered basal and lower stem leaves characteristic of this species.
Recognition. Lepidium panniforme has an erect, suberect, or spreading growth habit (Figs 82, 83), which immediately separates it from the decumbent L. oligodontum, and L. rekohuense. It is further separated from these taxa by the long persistent, much larger, usually deeply toothed or lacerate and often rather tattered basal and lower stem leaves (Figs 81-84), and from L. oligodontum and L. rekohuense by the fl owers which consist-ently have (2-)4 stamens. In growth habit, the species is most similar to L. oblitum and L. oleraceum, species with which it grows on Mangere Island. Th e key distinctions between L. oblitum and L. panniforme are described under L. oblitum, however; in brief, the deeply toothed and/or lacerate leaves of L. panniforme serve to readily distinguish it from L. oblitum. From L. oleraceum it is also easily separated, especially when fruiting when the notched rather than acute silicle apex can be seen, but also when vegetative, as the leaves of L. panniforme are diagnostically deeply toothed and/or lacerate. Th is is a condition seen otherwise only in the Bounty Islands endemic L. seditiosum which diff ers by having hairy infl orescences (Fig. 96) and distinct rDNA ETS sequence. Lepidium panniforme shares the same rDNA ETS sequence as the morphologically diff erent L. obtusatum. From that species it is recognised by its upright shrub-habit, deeply lobed to lacerate leaves and much smaller, minutely notched silicles.
Ecology. On the highly modifi ed landscape that is Mangere Island, Lepidium panniforme is known only from a very few sites where it grows in coastal herbfi eld along cliff tops, in rough pasture, shrubland, regenerating forest and sites kept artifi cially open, such as track sides. Because Mangere Island is being actively restored to coastal forest, few of these habitats are natural, and as such it is diffi cult to determine   Subsequent fi eld work by Department of Conservation staff on Mangere Island indicates that L. panniforme is a very uncommon species, which is potentially further threatened by the re-vegetation of that island. Th is is in part a natural process, though it is one which has been augmented by deliberate plantings as part of that island's long term restoration management as a wildlife refuge for threatened endemic fauna (D. Houston pers. comm.). Nevertheless replanting within the Lepidium panniforme sites has been carefully managed to date, and the species remains locally abundant along the disturbed and open track margins from the hut to the isthmus. However, with such a small population L. panniforme remains vulnerable to any change in management, and, over time as the surrounding shrubland matures to forest, some decline through natural succession is to be expected. Th erefore, this remarkable species will need to be closely monitored to ensure that it is not lost from the island. Th ere is also a need to survey for the species on the almost inaccessible, closely adjacent Little Mangere Island, as the species is still known from there only from those few images taken by the late D.V. Merton.
Th erefore, with < 250 mature plants known, and a total area of occupancy of < 1 ha, using the New Zealand Th reat Classifi cation System (see , Lepidium panniforme is rated as 'Th reatened/Nationally Critical'. In this case criterion A(1) or A(3) equally apply. To that listing we recommend appending the qualifi ers, 'CD' (Conservation Dependent -as the species is being actively managed), 'DP' (Data Poor -because accurate information about its status on Little Mangere Island is still needed), 'IE' (Island Endemic), 'OL' -because the species occurs on two closely associated islands such that it is at risk of elimination through catastrophe), and "RR" (Range Restricted -because of the narrow geographic range this plant occupies in relation to the rest of the New Zealand Botanical Region).
Lepidium rekohuense de (King 1989). Th is name was chosen to refl ect the endemic status of this species on the Chatham Islands group.

Recognition.
Healthy specimens of Lepidium rekohuense can form patches up to 2 m in diameter, which is the largest of the New Zealand endemic Lepidium species (Fig.  89). Within the L. oleraceum group, L. rekohuense is morphologically most similar to L. oblitum and L. oligodontum. From these species by is easily separated by the fl owers which consistently have two rather than 2-4 (L. oblitum) or 2-4-6 (L. oligodontum) stamens, by its much larger overall stature (up to 2 m diam.), by the sparsely papillatehairy upper branch stems, and by the presence of retrorse to patent, very short, ± clavate, eglandular-glandular hairs on the infl orescence rachis and pedicels. Th e silicles of L. rekohuense are orbicular (rarely obovate) and consistently, though minutely, notched (Fig. 92), while those of L. oligodontum, orbicular to suborbicular and not or scarcely notched.
Ecology. Lepidium rekohuense is currently known from salt-marsh and meadow at Kaiangaroa, from steep, eroded basaltic tuff erosion gullies and cliff faces at Cape Young and on Rabbit Island, and from the crevices and ledges of greywacke rock outcrops of the Forty Fours (Motuhara). At Kaiangaroa, L. rekohuense is a seasonally conspicuous member of the salt marsh and meadow vegetation that has developed behind the cobble beach and shallow shelving schist shore platform in and around Kaiangaroa Point. Here, plants grow in a variety of situations ranging from fully exposed and eroded habitats to low windswept thickets dominated by Hebe chathamica, and H. chathamica × H. dieff enbachii hybrids. In the salt marsh plants are usually found growing within dense Sarcocornia quinquefl ora (Bunge ex Unq.-Sternb.) A.J.Scott. subsp. quinquefl ora, Samolus repens var. repens, and Selliera radicans Cav. turf. In this habitat, plants are often lost during storm surges or during the winter months, and it would seem, from the presence of seedlings and young plants along drainage channels and in and around eroded sections of salt marsh, that these storm events are necessary to exhume and disperse buried seed. At the back of the salt marsh, where the salt meadows are dominated by taller plants such as Apodasmia aff . similis, Ficinia nodosa (Rottb.) Goetgh. Muasya et D.A.Simpson, and occasional Myosotidium hortensium, L. rekohuense plants are also present, and here they often grow intermixed with Apium prostratum subsp. denticulatum, Selliera, Samolus, and Leptinella potentillina. Higher up, where thickets of Hebe chathamica and hybrids form the dominant cover, L. rekohuense is less common, in part because they are often easily missed as they grow with Apium prostratum subsp. denticulatum threaded through H. chathamica. Lepidium rekohuense is also occasionally found growing on and around the small schist rock stacks around Kaiangaroa Point.
At Kaiangaroa the highly exposed and dynamic habitat means that many L. rekohuense plants, especially the younger plants are often lost through coastal erosion and from storm surges. However, in favourable sites, plants are remarkably resilient and long-lived once established. For example, mature plants fi rst observed in 1996 are still present at the time of writing (2012) 16 years later, making this species easily the longest lived member of the L. oleraceum complex in New Zealand. Th e key to this species success at Kaiangaroa seems to be its remarkable tap root, which, once established, fi rmly anchors the plant into the substrate such that coastal erosion often leaves mature plants exposed, festooned in driftwood and kelp, while the surrounding salt-marsh turf has been destroyed.
Th e habitat occupied at Cape Young and on Rabbit Island is markedly diff erent. Here the species grows at the apex of steeply descending, erosion gullies that have developed within the easily eroded basaltic tuff . In these sites it is often the only plant present though, in a few places on Cape Young, it grows with Lepidium fl exicaule, with which it occasionally hybridises. On Rabbit Island, large plants grew at the head of an erosion gully under a sparse canopy of the introduced tree mallow (Malva arborea).
Little is known about its habitats on the remote Forty Fours (Motuhara). From the limited information available (P. N. Johnson pers. comm.) it seems that the species is very uncommon there, and that it grows mainly within crevices and on ledges on the cliff faces of those rock stacks.
Conservation status. Th e most recent census data that we have (July 2007) recorded 114 adult plants of L. rekohuense from just three accessible sites; two on Rekohu (Kaiangaroa and Cape Young) and one on Rabbit Island. Th e status is uncertain of the species on the Forty Fours, privately owned land from which the Department of Conservation has not been granted visiting rights. Nevertheless, observations made in 2005 by a private landing party of geologists, entomologists and ornithologists suggest that there are probably fewer than 10 plants on the larger of the two main islets making up the Forty Fours. Using the New Zealand Th reat Classifi cation System , L. rekohuense is rated "Th reatened/Nationally Critical" using criterion A(1) because there are < 250 adult plants known from the wild. To this threat rating we recommend appending the qualifi ers 'CD' (Conservation Dependent -due to need for ongoing management of the Kaiangaroa population), 'IE' (Island Endemic -because L. rekohuense is naturally confi ned to the Chatham archipelago). It is worth noting that, without management, the largest population of L. rekohuense known to the Department of Conservation, that at Kaiangaroa, would probably now be extinct.  0.4-0.6 mm long, white, clavate hairs; pedicels 1.6-2.2 mm long at fl owering, erecto-patent, densely and mostly circumferentially covered in 0.4-0.6 mm long, white, clavate hairs. Flowers 2.3-2.6 mm diam. Sepals 4, saccate, dark green usually with a narrow white, ± undulose margin; lateral sepals broad, 0.8-1.2 mm diam., obovate to broadly obovate, ± overlapping at base, apex rounded to obtuse, abaxial surface densely hairy, hairs 0.1-0.4 mm long, eglandular or glandular, mostly clavate, some setose, median sepals 0.8-1.0 mm diam., broadly obovate, dark green, usually with a narrow white, ± undulose margin, apex rounded to obtuse, abaxial surface densely hairy, hairs 0.1-0.4 mm long, eglandular or glandular, mostly clavate, some setose. Petals white, 1.3-2.0 × 1.0-2.3 mm, mostly recurved over stigma some spreading, claw 0.4-0.9 mm long; limb obovate, obovate-spathulate rarely orbicular, apex obtuse or slightly emarginate, margins smooth. Stamens 4, fi laments 1.2-1.8 mm long, white; anthers 0.3-0.4 mm long, yellow. Ovary 1.1-1.8 × 0.6-1.3 mm, broadly ovate to elliptic, dark green, apex round or weakly notched; style 0.11-0.4 mm long, cylindrical below, broadly spreading at apex; stigma 0.2-0.4 mm diam. Nectaries 4, 0.2-0.3 × 0.1-0.15 mm, narrow-oblong, pale translucent green. Mature silicles not seen. FL Nov. FR unknown. Distribution (Fig. 76). Endemic. Bounty Islands (Funnel Island, Molly Cap).
Recognition. Amey et al. (2007) initially reported the presence of Lepidium oleraceum s.l. from the remote Bounty Islands. In that paper they noted that unpublished rDNA ETS data from their gathering placed the Bounty Islands plant within a southern L. oleraceum clade of samples collected from Banks Peninsula south to Stewart Island rather than plants from the nearby Antipodes Islands. On the basis of that information they suggested that the Bounty Islands plant was a recent arrival, probably coming from the Otago coastline of the southern South Island. Subsequently, the rDNA ETS phylogeny presented here confi rms that L. seditiosum is related to the South Island species L. aegrum, L. crassum, and L. juvencum (Fig. 2). However none of these species are morphologically similar to L. seditiosum. Morphologically, the most similar species is the Chatham Islands endemic Lepidium panniforme, which also has deeply toothed (though also lacerate leaves) but whose stems and infl orescence rachis are glabrous (only very rarely furnished with sparse silky hairs), and whose rDNA ETS sequence places it as sister to L. obtusatum. Th e Bounty Islands plant diff ers from L. panniforme (and indeed all other members of the Lepidium oleraceum complex) by its distinctly clavate-hairy upper stems and infl orescence rachises.
While our material is inadequate to furnish a full species description, the absence of mature silicles and basal leaves should not preclude formal taxonomic recognition. Th e Bounty Islands are perhaps the most remote outlier of the New Zealand Archipelago, and are very rarely visited (Amey et al. 2007), so the alternative, of waiting for better fruiting material before describing what in our view is clearly a new species is, unacceptable. In the authors' opinion, there is suffi cient data available to justify this plant's taxonomic recognition at species rank.
Ecology. Prior to the remarkable discovery of Lepidium on the Bounty Islands group, these islands had always been considered devoid of any form of terrestrial vegetation beyond a few lichens and algae (Amey et al. 2007). None of the 15 islands, islets and rock stacks comprising the group reach any higher than 88 m a.s.l., and virtually all available dry land is occupied by some of the densest concentrations of nesting seabirds seen anywhere in the world (Amey et al. 2007). Amey et al. (2007) recorded Lepidium seditiosum (as L. oleraceum s.l.) as growing within rock crevices near the summits of one island (Funnel Island) and an inaccessible (by boat) rock stack (Molly Cap) within the Bounty Islands group. At the time of their discovery, on Funnel Island plants grew on the margins of a colony of nesting seabirds, in a site inaccessible to albatross and penguins but used by Fulmar prions (Pachyptila crassirostris) and possibly Cape pigeons (Daption capense). Th ey noted that where the plants grew, a 'rich friable soil' had accumulated, and that the crevice walls provided some shelter from the most extreme winds. Because of the inaccessible nature of Molly Cap, no further observations on the plant they had seen from their boat could be made.
Conservation Status. Amey et al. (2007), during a November 2004 visit to the Bounty Islands group, recorded 'at least 13' plants at two sites (Funnel Island (12 plants) and Molly Cap (1 plant). On this basis, L. seditiosum qualifi es as 'Th reatened / Nationally Critical' (either criterion A(1) or A(3) of Townsend et al. (2008)) as there are fewer than 250 mature plants known and the total area of occupancy is < 1 ha. To this assessment we add the qualifi ers 'CD' (Conservation Dependent -as the Bounty Islands are a Nature Reserve and World Heritage Site with strict controls in place to regulate landings, and prevent the spread of diseases, weeds and foreign predators), 'DP' (Data Poor -because accurate data on the total number of individuals and trend data is not available), 'IE" (Island Endemic), and 'OL' (One Location -because, following the defi nition in , the species is confi ned to the Bounty Islands group, where it grows on an island and rock stack in close proximity, and so the chances of losing the species to a single event are greater than for a species found in one island group on several widely separated islands).  de Lange et al. (2009) and de Lange and Rolfe (2010) in treating Lepidium desvauxii as indigenous within the southern part of its New Zealand range (including the Chatham Islands), and naturalised within the more modifi ed parts of central and northern New Zealand. 3 Lepidium ruderale is accepted by Webb et al. (1995) as naturalised in New Zealand. 4 Lepidium panniforme usually has 4 stamens but occasional plants with 2 or 2-4 stamens occur, and for this reason this species may key out in two areas depending on the number of stamens present. 5 Lepidium peregrinum is regarded here as naturalised to New Zealand (see Heenan and de Lange 2011). It is as an uncommon naturalised plant of glauconite and glauconitic limestone rock stacks and islands along the southern shore line of the Kawhia Harbour. Lepidium peregrinum is a threatened species in Australia (Scarlett 1999). 6 Lepidium oblitum may have 2-4 stamens. For this reason this species may key out in two areas depending on the number of stamens present. 7 Occasional plants of Lepidium naufragorum with simple or weakly pinnatifi d leaves are also known, and these could be confused with L. oleraceum from which they diff er by the seasonal growth habit (with plants dying back to the rootstock over winter), and by the distinctly emarginate silicles 8 Lepidium seditiosum is known only from a single, imperfect gathering which lacks mature silicles. However, the shape of the ovary apex of immature silicles indicates that, in this species, they are notched.