Morphometric analyses of New Zealand giant vegetable sheep lead to the recognition of its two varieties as species: Haastia pulvinaris and Haastia minor (Senecioneae: Asteraceae)

ABSTRACT

Haastia pulvinaris (Asteraceae: Senecioneae) is endemic to alpine ecosystems in the north-eastern mountains of the Southern Alps of Te Waipounamu – South Island of Aotearoa – New Zealand. Two varieties are currently recognised: H. pulvinaris var. pulvinaris and H. pulvinaris var. minor. Univariate and multivariate analyses of 37 specimens resulted in the discovery of distinct differences between these varieties in, amongst others, branchlet diameter, leaf shape and the presence of an apical style tuft. Their consistent and distinct differences, even in sympatry, are evidence that the two varieties should be recognised as species: H. pulvinaris and H. minor comb. et stat. nov. Descriptions and distribution maps of each species are provided, together with illustrations and an identification key to all species and varieties of Haastia.

KEYWORDS:

• Haastia
• Haastia pulvinaris
• Compositae
• taxonomy
• morphology
• morphometrics
• New Zealand

Introduction

Haastia Hook.f. (Senecioneae: Asteraceae) is a small genus of low growing subshrubs or perennial herbs that are woody at the base, and have large and solitary heterogamous and disciform capitula with yellow florets (Breitwieser and Ward 2005). It is endemic to high elevations in the mountain ranges of Te Waipounamu – South Island of Aotearoa – New Zealand. Haastia was previously placed in Gnaphalieae (as ‘Gnaphalioid Compositae’; Hooker 1864) and Astereae (Bentham and Hooker 1873), but more recent molecular phylogenetic and morphological studies support its classification in Senecioneae (Wagstaff and Breitwieser 2002, 2004; Breitwieser and Ward 2005; Pelser et al. 2007, 2010). The species delimitation of Haastia has remained unchanged since Hooker (1864) described it. To this day, three species are recognised: H. pulvinaris Hook.f., H. recurva Hook.f. and H. sinclairii Hook.f. (Kirk 1899; Cheeseman 1906; Allan 1961; Figure 1).

Figure 1. A & B, Haastia sinclarii; C & D, H. recurva; E & F, H. pulvinaris.

Haastia pulvinaris is the most enigmatic of the three species. It is a cushion plant, forming large whitish mounds of tightly packed branchlets (Figure 1) and they can be confused for a flock of sheep from a distance. Hooker (1864, p. 156) noted that it is ‘one of the most extraordinary plants in the islands. Sinclair says the patches are so dense, that the finger cannot be thrust between the branches’. The mounds can be up to 2 metres in diameter and half a metre tall. This appearance gives the species its common name of ‘giant vegetable sheep’. It is found in the northern mountains of the Te Waipounamu – South Island (i.e. Canterbury, Marlborough, and Tasman), east of the main divide and above 1300 m.a.s.l., on rock fields and scree slopes (Figure 2A and B).

Figure 2. A & B, Habitat of Haastia pulvinaris; C, Habit of H. pulvinaris var. pulvinaris; D, Habit of H. pulvinaris var. minor; E, Surface of H. pulvinaris var. pulvinaris plant in flower; F, Surface of H. pulvinaris var. minor plant in flower.

Two varieties of Haastia pulvinaris are currently recognised: H. pulvinaris var. pulvinaris and H. pulvinaris var. minor Laing (Allan 1961). Laing (1912) described H. pulvinaris var. minor as being smaller than the typical variety in all parts, with a paler indumentum, along with fimbriate instead of truncate pappus hairs. Moreover, the two varieties occur in sympatry. They are commonly found along the same mountain ridges and even observed growing in their distinct forms side by side and flowering at the same times (Mark and Adams 1973; Todd 1996; Mark 2021).

In his unpublished MSc thesis, Todd (1996) recommended that Haastia pulvinaris var. minor should be elevated to species level because of a large number of morphological differences with H. pulvinaris var. pulvinaris. In addition, he discussed differences in phenology and biochemistry (flavonoid compounds). This taxonomic conclusion also appears to have been considered by Druce, because he included ‘Haastia minor’ in his unpublished checklist of indigenous vascular plants of New Zealand (Druce 1993, p. 18). Among the morphological differences mentioned by Todd (1996) are differences in the diameter of branchlets (including leaves), leaf colour and shape, shape of the involucral bracts, and apex of the style branches. Breitwieser and Ward (2005) also noted differences in the latter character within H. pulvinaris, without attributing them to the named varieties.

This study aimed to test the hypothesis that Haastia pulvinaris var. pulvinaris and H. pulvinaris var. minor should be recognised at the species level. Formally recognising taxa at an appropriate taxonomic level is, amongst others, crucial for biodiversity conservation and understanding of ecosystems (Dussex et al. 2018; Grace et al. 2021). Inaccurate assessments can result in over- or underestimates of species-level biodiversity (Chaitra et al. 2004; Iglésias et al. 2010). This can, for example, mask declines in biodiversity (Iglésias et al. 2010; Huang and Knowles 2015). In this study, we adopted the ‘unified species concept’ (de Queiroz 2007) and used morphological distinctiveness as well as the persistence of morphological differences in sympatry as a ‘secondary species criterion’ for identifying separately evolving metapopulation lineages, which are considered distinct species under this species concept (de Queiroz 2007). Specifically, we considered non-overlapping patterns of morphological character distributions as evidence that taxa are distinct at the species level (Pelser and Houchin 2004). To determine this for the two varieties of H. pulvinaris, we performed univariate and multivariate morphological analyses of data collected mostly from herbarium specimens. We also mapped their recorded collection locations to clarify their distribution areas.

Material and methods

Plant specimens and data for morphometric study

Morphological data from 19 specimens of Haastia pulvinaris var. pulvinaris and 18 specimens of H. pulvinaris var. minor were used to identify and quantify the morphological differences between these two varieties. These specimens were selected across their geographic ranges and to represent the morphological variation of H. pulvinaris. We used only specimens with a sufficient number of capitula to allow for destructive sampling.

Four of these specimens, two of each variety, were collected over the summer of 2021/22 from Mt Robert in the Nelson Lakes region so that data could be obtained that cannot easily be recorded from herbarium specimens, such as dimensions of the whole plant, and fresh floral and vegetative characters.

The remaining specimens were accessed at the Allan Herbarium (CHR) and the University of Canterbury Herbarium (CANU; herbarium acronyms follow Thiers 2016). Many of these specimens were labelled only with the species name, Haastia pulvinaris, so specimens were provisionally attributed to one of either variety prior to data collecting using three diagnostic vegetative characters that can be studied without dissecting a specimen (i.e. branchlet diameter (incl. leaves), leaf apex prominence and colour). In addition to dried specimens, some associated material preserved in FAA (Formalin-Acetic Acid-Alcohol) was used.

Following a pilot study to identify potentially taxonomically informative morphological characters, 24 characters were selected for our study (Table 1). For many of these characters, character states were recorded from more than one homologous structure per specimen (e.g. branchlet diameter was often measured for several branches per specimen), and in these cases the average value of the measurements was used for the statistical analyses described below. For each qualitative character, the most common character state was recorded for the specimen.

Table 1. Morphological differences between Haastia pulvinaris var. pulvinaris and H. pulvinaris var. minor in 24 morphological characters.

		H. pulvinaris var. pulvinaris	H. pulvinaris var. minor
	Character	Mean ± SD	Mean ± SD	Test	P-value
	Branchlets
1.	Branchlet diameter (mm)	15.6 ± 2.4	10.4 ± 1.4	A	<0.001***
2.	Leaf apex prominence	Obscured (79%)	Distinguishable (83%)	F	<0.001***
	Leaves
3.	Leaf length (mm)	11.5 ± 1.2	7.9 ± 1.1	A	<0.001***
4.	Leaf width (mm)	8.0 ± 0.8	9.1 ± 1.3	A	0.005**
5.	Distance from base to widest point of leaf (mm)	9.2 ± 1.3	4.5 ± 0.8	K	<0.001***
6.	Ratio of (3) to (4)	1.46 ± 0.09	0.87 ± 0.16	K	<0.001***
7.	Ratio of (3) to (5)	0.80 ± 0.05	0.58 ± 0.08	K	<0.001***
8.	Leaf shape	Cuneate (100%)	Flabellate (100%)	F	<0.001***
9.	Leaf apex shape	Truncate (100%)	Rounded (100%)	F	<0.001***
10.	Leaf vein number	3.1 ± 0.2	5.1 ± 0.8	K	<0.001***
	Capitula
11.	Capitulum diameter (mm)	5.2 ± 1.0	4.6 ± 0.7	K	0.140
12.	Apparent capitulum position^a	Lateral (95%)	Terminal (67%)	F	0.001**
13.	Involucral bract length (mm)	6.7 ± 0.6	6.6 ± 0.5	A	0.673
14.	Distance from base to widest point of involucral bract (mm)	1.4 ± 0.6	3.5 ± 1.3	K	<0.001***
15.	Ratio of (13) to (14)	0.20 ± 0.09	0.53 ± 0.20	K	<0.001***
16.	Receptacle diameter (mm)	3.2 ± 0.8	2.6 ± 0.8	A	0.027*
17.	Receptacle shape	Flat (89%)	Convex (72%)	F	0.002**
	Florets
18.	Pistillate floret corolla length (mm)	1.9 ± 1.1	3.1 ± 0.9	A	0.001**
19.	Perfect floret corolla length (mm)	5.9 ± 0.8	5.7 ± 0.8	A	0.441
20.	Pistillate floret style apical tuft presence	Absent (100%)	Absent (78%)	F	0.046*
21.	Perfect floret style apical tuft presence	Absent (100%)	Present (100%)	F	<0.001***
22.	Cypsela length (mm)	2.0 ± 0.4	1.8 ± 0.4	A	0.079
23.	Pappus hair length (mm)	7.3 ± 1.1	6.8 ± 0.9	A	0.237
24.	Pappus hair apex	Truncate (89%)	Fimbriate (78%)	F	<0.001***

Note: Means and standard deviations (SD) are provided for quantitative characters. Percentage of specimens exhibiting the most common character state are listed for qualitative characters. P-values of ANOVA (A), Kruskal-Wallis (K) and Fisher’s Exact (F) tests are presented, along with their level of significance: (*) Significant, (**) Very significant and (***) Highly significant. Leaf characters were recorded for leaves c. 1 cm below the branchlet apices.

^a The position of the capitulum relative to the branch on which it is positioned as seen via visual inspection of the surface.

Multivariate analyses

Four multivariate analyses were performed: Non-metric Multidimensional Scaling (NMDS) analysis, Principal Coordinates Analysis (PCoA), Hierarchical Agglomerative Clustering (HAC) and Random Forest (RF) classification. For these analyses, two of the 24 characters studied, ‘leaf width’ and ‘distance from base to widest point of leaf’, were removed from the dataset as these characters are not independent if the characters ‘leaf length to width ratio’ and ‘leaf length to widest point ratio’ are also included. All data analyses were performed using R version 4.2.2 (R Core Team 2022) and RStudio (RStudio Team 2020).

Gower’s general similarity coefficient using the ‘daisy’ function from the R ‘cluster’ package (Maechler et al. 2022) was used to prepare a dissimilarity matrix for the NMDS, PCoA and HAC analyses. Gower’s coefficient is favoured in morphometric analysis as it can be used with data sets that contain both qualitative and quantitative variables (Crisp and Weston 1993; Ward 1993). NMDS was performed using the ‘metaMDS’ function from the ‘vegan’ package (Oksanen et al. 2022) in R. PCoA was conducted using the ‘cmdscale’ function in the package ‘stats’ (of base R) in R. The Unweighted Pair Group Method with Arithmetic mean (UPGMA) was used to perform HAC using the ‘agnes’ function from the package ‘cluster’ in R (Maechler et al. 2022).

Random Forest (RF) classification (Breiman 2001) using the R package ‘randomForest’ (Liaw and Wiener 2002) was used to determine how well the primary clusters identified by the UPGMA analysis are morphologically differentiated. The optimal number of randomly preselected splitting variables was determined using the function ‘tuneRF’and the number of trees was set at 1000.

Univariate analyses

To quantify the differences between the groups identified in the UPGMA analysis for each of the 24 characters, univariate analyses were carried out, all at the 95% confidence level. Quantitative characters were first checked for normality using the Shapiro–Wilk test (Shapiro and Wilk 1965) with the ‘shapiro.test’ function in the package ‘stats’ (of base R) in R. Characters with Shapiro–Wilk p-values larger than 0.05 were assumed to have a normal distribution, and for those a one-way Analysis of Variance (ANOVA) was performed using the ‘aov’ function in the package ‘stats’ (of base R) in R. For characters with a p-value of less than 0.05, a Kruskal–Wallis test was performed, which is a non-parametric version of ANOVA (Kruskal and Wallis 1952). To analyse the qualitative characters, a Fisher exact test was considered suitable due to the comparison involving data in a 2 × 2 contingency table and a relatively small sample size (Fisher 1925). This test was performed using the ‘fisher.test’ function in the package ‘stats’ (of base R) in R.

Taxonomic descriptions and distribution map

Following the identification of taxonomically diagnostic differences between the two taxa, these characters were used to study all herbarium specimens of Haastia pulvinaris at the Allan Herbarium (CHR), University of Canterbury Herbarium (CANU), Museum of New Zealand Te Papa Tongarewa (WELT) and Auckland Museum Herbarium (AK). Their identifications were verified, corrected, or provided and specimens were annotated. Data from these specimens (including those used in the morphometric analyses) were subsequently used to prepare taxonomic descriptions. Location information recorded from 135 specimens in CANU and CHR was used to make a distribution map with the packages ‘rnaturalearth’ (South 2017) and ‘ggplot’ (Wickham 2016) in R.

Results and discussion

Morphometric analyses

The stress value of the NMDS ordination when using two dimensions (K = 2; Figure 3) to represent the data was 0.06, a value considered to represent a good fit between the Gower’s dissimilarities and ordination distance, with no real risk of drawing false inferences (Clarke 1993). The first two principal coordinate axes of the PCoA (Figure 4) accounted for 76.1% and 10.7% of the total variation in the morphological dataset, combining for a total of 86.8% of variation explained.

Figure 3. NMDS plot obtained from 22 morphological characters recorded from 37 Haastia pulvinaris specimens. Specimens are coloured by the variety that they were assigned to prior to statistical analysis. 95% confidence ellipses are drawn for both varieties.

Figure 4. PCoA plot obtained from 22 morphological characters recorded from 37 Haastia pulvinaris specimens. Specimens are coloured by the variety that they were assigned to prior to statistical analysis. 95% confidence ellipses are drawn for both varieties.

The NMDS (Figure 3), PCoA (Figure 4) and UPGMA hierarchical clustering analysis (Figure 5) performed of Gower’s general dissimilarity coefficient matrix all placed the specimens into two groups. The composition of these groups perfectly aligned with the taxonomic identity of the specimens as determined prior to the morphological study. In addition, the results of the RF analysis showed that all specimens identified as either Haastia pulvinaris var. minor (UPGMA Cluster 1) or H. pulvinaris var. pulvinaris (UPGMA Cluster 2) were correctly assigned to their respective cluster 100% of the time.

Figure 5. UPGMA dendrogram obtained from 22 morphological characters recorded from 37 Haastia pulvinaris specimens. Specimens are labelled with their CANU or CHR accession number. Specimens forming the cluster I are those identified as H. pulvinaris var. pulvinaris. Cluster II is composed of H. pulvinaris var. minor specimens.

There were statistically significant differences between the two varieties for 19 of the 24 characters (Table 1). For some of these characters, we found slightly more variation when additional specimens were studied for the purpose of preparing taxonomic descriptions (see Taxonomy). However, leaf length to width ratio did not show overlap in its values between the two varieties and they had unique character states for leaf shape and the presence of a style tuft.

Thus, both the multivariate and univariate analyses of the morphological data set divided the specimens into two distinct groups. The two groups aligned perfectly with the attributed varieties on the herbarium specimen labels where these were available, and to the allocation of specimens to variety that we made prior to our morphometric study on the basis of three vegetative characters that can be used without dissecting a specimen (i.e. branchlet diameter, leaf apex prominence and colour). Below, we discuss the taxonomic and diagnostic value of the characters used for this study.

Vegetative characters

Branchlet diameter (incl. leaves), measured across the cushion surface (Figure 6A and C), was a diagnostic character used by Laing (1912) in his description of Haastia pulvinaris var. minor. Our morphometric study showed that branchlets of Haastia pulvinaris var. pulvinaris were indeed significantly wider than those of H. pulvinaris var. minor included in our morphometric analyses (i.e. average branchlet diameter c. 16 vs. 10 mm; Table 1, Figure 6A and C). However, some H. pulvinaris var. pulvinaris specimens had relatively thin branchlets which overlapped in diameter with those of H. pulvinaris var. minor (see descriptions below). This means that additional diagnostic characters may need to be observed for accurate field identification.

Figure 6. Comparison of branchlet characters of Haastia pulvinaris var. pulvinaris (A & B) and H. pulvinaris var. minor (C & D); A & C, Plant surface showing branchlet apices; B & D, Lateral view of single branchlet with capitulum present (FAA-preserved specimens). Scale bar: B & D = 10 mm.

Leaf apex prominence is another informative character (Table 1, Figure 6A and C). The branchlet tips of Haastia pulvinaris var. pulvinaris were typically covered in very dense indumentum, which obscured the leaf apices on the surface of the plant (Figure 6A). In contrast, the leaf apices of H. pulvinaris var. minor tended to be visible on the surface, with the clustered leaves of each branchlet forming a characteristic rosette shape (Figure 6C). However, leaf apex prominence did vary occasionally within the varieties and even within individual plants.

Although previous research (Allan 1961; Todd 1996) suggested that the two varieties differ in leaf colour, we did not include this character among those selected for our morphometric analyses, because colour could not be reliably assessed from many dried and FAA-preserved specimens selected for our study. Nonetheless, we were able to record some data from freshly collected specimens and those collected relatively recently, as well as from photographs on iNaturalist (2023) or taken by ourselves. The colour of Haastia pulvinaris mounds tends to be a product of the colour and density of the indumentum of the leaves, as no clear differences in leaf lamina colour were observed during the morphological investigation. Colour can assist with distinguishing the varieties in the field (particularly when sympatric) as well as when they are preserved as herbarium specimens. Haastia pulvinaris var. pulvinaris plants are gold or cream-gold, whereas H. pulvinaris var. minor specimens are pale green or cream-pale green (Figure 6A and C).

Several other leaf characters separated the specimens investigated into two groups (Table 1). Leaves sampled c. 1 cm from the branchlet apex differed between the two varieties in size and shape. Haastia pulvinaris var. pulvinaris leaves tended to be longer and are cuneate with a truncate apex (Figure 7A and C), whereas those of H. pulvinaris var. minor were usually shorter and flabellate with a rounded apex (Figure 7B and E). While leaf width alone is not a useful diagnostic feature due to a large amount of overlap between the taxa, leaf length to width ratio is informative, with H. pulvinaris var. pulvinaris having leaves longer than wide (Figure 7A) and H. pulvinaris var. minor most commonly wider than long (Figure 7B). The difference in length to width ratio is a particularly useful diagnostic character because its values did not overlap between the varieties.

Figure 7. Comparison of leaf characters of Haastia pulvinaris var. pulvinaris (A, C, D) and H. pulvinaris var. minor (B, E); A & B, Adaxial view of leaves from fresh samples. Scale bar = 10 mm; C, Adaxial view of cuneate shaped leaf with indumentum removed; D, Adaxial view of oblong leaf with indumentum removed; E, Adaxial view of flabellate shaped leaf with indumentum removed.

The number and branching of veins in the lower part of the leaf varies between the varieties, with Haastia pulvinaris var. pulvinaris almost always having 3, rarely 4, major veins running parallel to each other from the lower part of the leaf towards the leaf apex (Figure 7C and D). In contrast, H. pulvinaris var. minor specimens had (3–)4–7 major veins in the lower part of the leaf that anastomose close to the apex (Figure 7E).

Capitulum and floret characters

Receptacle diameter and shape show significant differences between the varieties (Table 1). Figure 8A and B show the differences in capitulum shape typical for the two varieties, and highlight that while the capitulum diameter may be similar, the receptacle of Haastia pulvinaris var. minor is smaller and more convex than the larger, flatter receptacle of the typical variety.

Figure 8. Comparison of capitulum and floret characters of Haastia pulvinaris var. pulvinaris (A, C, E, G) and H. pulvinaris var. minor (B, D, F, H); A & B, Capitulum with portion of involucral bracts removed to show receptacle; C & D, Involucral bract; E & F, Apical portion of flower showing style apices; G & H, Pappus hair apices.

The involucral bract shape also differs between the two varieties. Involucral bracts of Haastia pulvinaris var. pulvinaris tend to be wider towards the base (Table 1), tapering to a point at the apex, and giving the involucral bracts an overall lanceolate shape (Figure 8C). The involucral bracts of H. pulvinaris var. minor vary more in shape within an individual specimen than those of the typical variety. They are either lanceolate or narrowly oblong (having the widest point towards the middle/upper portion of the bract and quickly tapering to a point at the apex) (Figure 8D, Table 1).

The length of the corolla of the outer florets differs between the varieties. These are pistillate florets and form narrow tubes of even width that envelop the style, with corolla lobes rarely being distinguishable. In Haastia pulvinaris var. pulvinaris, the corolla tubes tend to be very short, only extending 1.0–2.0 mm up the style (Table 1). In H. pulvinaris var. minor however, the corolla tube extends much further up the style, generally between 2.5 and 4.0 mm (Table 1), around half the length of the style. Present in both varieties is the occasional specimen with a single long corolla lobe which can extend up to 2.0 mm from the corolla tube.

The morphology of the papillae found on the style apices of the central perfect florets shows distinct differences. The papillae of Haastia pulvinaris var. pulvinaris are short and uniformly distributed along the style branches, slightly lengthening towards the tip, forming a club shaped apex (see Figure 8E) whereas in H. pulvinaris var. minor, the papillae are uneven and protrude from the style apices in distinct tufts (see Figures 8F and 9A). This difference is consistent, making it a useful diagnostic character.

Figure 9. Haastia pulvinaris var. minor. A, Style apex tufts of perfect floret; B, Uncommon form of pistillate floret style apices with singular tuft of cells present.

While the majority of style apices of the pistillate florets lack papillae, four Haastia pulvinaris var. minor specimens were observed with a single tuft of cells protruding from the style apices (Figure 9B). These tufts are visually very similar to those seen in the perfect floret styles (Figure 9A), but are lacking the uneven papillae tufts protruding laterally. While this may not always be a helpful diagnostic character due to the uncommon presence of these tufts in pistillate florets, if these tufts are present, they can provide a compelling identification diagnosis for H. pulvinaris var. minor.

The two varieties display differences in the shape of the pappus hair apices. In Haastia pulvinaris var. pulvinaris, pappus hair apices tend to be obtuse and truncate (see Figure 8G) whereas those in H. pulvinaris var. minor are usually more variable and fimbriate, with cells fanning out at the apex (see Figure 8H). This observation agrees with the information in Laing’s (1912) protologue of this variety.

Taxonomic recognition

In this study, we tested the hypothesis that Haastia pulvinaris var. pulvinaris and H. pulvinaris var. minor should be recognised at the species level under a unified species concept (de Queiroz 2007). We tested this using morphological data and evidence of sympatry. The NMDS and PCoA ordination of 22 morphological characters placed the investigated specimens in two distinct, non-overlapping clusters in multivariate morphometric space. In addition, 19 out of 24 vegetative and reproductive characters investigated were found to differ statistically significantly between the two groups, with five of these showing non-overlapping character states. These findings offer conclusive evidence that the two groups of specimens are morphologically distinct from each other, indicating that they constitute separately evolving metapopulations and can therefore be considered different species (de Queiroz 2007). The identified groupings of specimens corresponded exactly with the two varieties. This was further supported by the UPGMA cluster analysis and subsequent Random Forest classification analysis (accuracy score of 100%).

Mapping of the collecting locations of specimens held at CANU and CHR (Figure 10) showed a high degree of overlap of the distributions of the two varieties of Haastia pulvinaris, as was observed previously (e.g. Mark and Adams 1973; Todd 1996; Mark 2021), although H. pulvinaris var. pulvinaris has a distribution area that extends further north and south. Todd (1996) noted that the two varieties of H. pulvinaris were sometimes even observed growing adjacent to one another, for example on Mt St Patrick. During our fieldwork, plants of both varieties were found growing within 20 m of each other on Mt Robert. Information on herbarium labels as well as our own observations further showed that plants with similar or identical collecting coordinates also had comparable habitat information. We did not observe specimens with an intermediate morphology at locations where both varieties are found. There is therefore evidence that the two varieties remain morphologically distinct from each other in sympatry and are therefore probably reproductively isolated. In conclusion, our data show that H. pulvinaris var. pulvinaris and H. pulvinaris var. minor are morphologically distinct, even in sympatry, and that their taxonomic rank therefore should be elevated to species level: H. pulvinaris and H. minor comb. et stat. nov.

Figure 10. Distribution map of the two Haastia pulvinaris varieties in the northern Te Waipounamu – South Island of Aotearoa – New Zealand based on recorded location data from all CANU and CHR specimens with sufficiently detailed location data.

Taxonomy

Key to species and varieties of Haastia

This identification key to all six Haastia species and varieties recognised in this study was prepared using H. pulvinaris and H. minor data collected during this study. For H. recurva and H. sinclairii, we relied on data provided by Allan (1961) and Todd (1996). The latter two species display considerable morphological variation (Todd 1996) and further study is required to determine their infraspecific delimitation. The parts of this key pertaining to the differences between the presently-recognised varieties of H. recurva and H. sinclairii should be used with caution and with this in mind.

1.	Compact cushion shrub with branchlets tightly packed to form a closed surface; leaves very densely imbricate, (5–)7–12(–16) mm long ……… 2.
–	Laxly branched subshrub or herb woody at base, branchlets clearly separated; leaves loosely imbricate, 13–35 mm long ……… 3
2.	Plant pale green to cream-pale green; branchlets (incl. leaves) 8–14 mm in diameter, leaf apices usually visible; leaves flabellate, wider than long, major veins 3–7; style branches of central florets with apical tuft of papillae; pappus hair apices fimbriate ……… H. minor
–	Plant gold or cream-gold; branchlets (incl. leaves) 9–23 mm in diameter, leaf apices usually obscured by indumentum; leaves usually cuneate, longer than wide, major veins 3–4; style branches of central florets without apical tuft of papillae; pappus hair apices slightly obtuse or truncate ……… H. pulvinaris
3.	Leaves 13–20 mm long, obovate, recurved, indumentum floccose; internodes c. 1 mm long; pappus hairs connate at base ……… 4
–	Leaves 18–35 mm long, obovate to ovate, not or slightly recurved, indumentum appressed to subappressed; internodes 1–5 mm long; pappus hairs free at base ……… 5
4.	Leaves grey to greyish yellow, 13–20 mm long, recurved up to 75 degrees, indumentum fulvous to rufous; capitula 8–10 mm in diameter; apex of involucral bracts acute ……… H. recurva var. recurva
–	Leaves yellow-green, 14–15 mm long, recurved up to 90 degrees, indumentum mostly white; capitula 6–7 mm in diameter; apex of involucral bracts apiculate ……… H. recurva var. wallii
5.	Leaves 18–25 mm long, patent, silver or grey, indumentum dense, greyish white; involucral bracts 10–12 mm long; pappus hairs 9–10 mm long ……… H. sinclairii var. sinclairii
–	Leaves 25–35 mm long, slightly recurved, yellow-green to green, indumentum sparse, cream to fulvous; involucral bracts 8–10 mm long; pappus hairs 7–9 mm long ……… H. sinclairii var. fulvida

Haastia pulvinaris Hook.f., Handb. New Zealand Fl.: 156. 1864.

Lectotype. NEW ZEALAND, South Island, Kaikoura Mtns, A. Sinclair s.n., 1860 (K000844044 [web!])

Description. Perennial pulvinate or convex shrubs, stout, gold to cream-gold, 0.1–0.4 m tall, individual mounds 0.5–2.0 m in diameter. Rootstock woody. Branchlets (incl. leaves) densely packed forming a closed surface, at surface (9–)14–18(–23) mm in diameter, with leaf apices usually obscured by dense indumentum. Leaves cuneate or sometimes oblong, (7–)10–12(–16) x (4–)6–9(–10) mm, length/width ratio 1.2–1.8, length/distance from base to widest point ratio 0.7–0.9, pale green, darkening towards apex; lamina below apex thickened and crenulate, both lamina surfaces densely clad in long, tangled, slightly fulvous hairs extending to 3–4 mm beyond apex; sheath white to pale green, sometimes purple at base; veins 3–4 at base, anastomosing near the apex; apex truncate, sometimes rounded or obtuse, crenulate. Leaves subtending capitula linear 6–8(–11) × 1–4 mm, apex truncate, rounded or obtuse. Capitula usually lateral, occasionally terminal, heterogamous and disciform, 3.5–7.0 mm in diameter. Involucral bracts (15–)18–21, valvate, in single row with 1–5 individual bracts outside the row, lanceolate or ovate, rarely narrowly oblong, 5.0–8.0 × 0.5–1.2 mm, length/distance from base to widest point ratio 0.10–0.20(–0.70), hairs present on abaxial side, apex acute with slight fringe of short hairs; veins 1–2, simple and unbranched. Receptacles flat, 2.0–4.0 mm in diameter. Outer florets 20–24, pistillate; corolla narrowly tubular around style, sometimes uneven at apex, 1.0–2.0(–5.0) mm long, yellow; style bifurcating, (2.5–)5.0–7.5 mm long, yellow, papillae absent. Central florets 12–20, perfect; corolla tubular, 5-lobed, (3.5–)4.5–7.5 mm long, yellow; style bifurcating, (2.0–)4.0–7.0 mm long, yellow, branches club-shaped, papillae present along abaxial side of style branch, short and evenly distributed, lengthening towards apex; stamen filament collars slender; anthers ecaudate and ecalcarate. Cypselae oblong, glabrous, 1.5–2.5 mm long. Pappus hairs slender, barbellate, up to 9.0 mm long, always extending beyond corolla and style, flattened at base, apex slightly obtuse or truncate.

Distribution. Aotearoa – New Zealand, endemic. Te Waipounamu – South Island: Eastern Tasman, Marlborough, North Canterbury.

Habitat and ecology. Haastia pulvinaris grows on rocky alpine fellfields and stable scree slopes, usually at elevations exceeding 1500 m. The densely packed woolly branchlets and secure root system of H. pulvinaris allow the plant to withstand the harsh conditions prevalent in alpine environments. Although it does not consistently co-occur with other cushion or mat forming plants, it can occasionally be found in their vicinity, sometimes in direct contact.

Phenology. Due to the alpine habitat Haastia pulvinaris grows in, reproduction is limited to the period of time between the spring thaw and the first snow of autumn. It flowers from late December to mid-March and fruits from late February to mid-April.

Conservation status and notes. This species (as Haastia pulvinaris Hook.f. var. pulvinaris) has been assessed as ‘Not Threatened’ [Population 20000–100000 mature individuals, with a stable population (±10%)] by the New Zealand Indigenous Vascular Plant Threat Listing Panel (de Lange et al. 2018) using the New Zealand Threat Classification System (Townsend et al. 2008).

Etymology. The specific epithet pulvinaris is derived from Latin, meaning ‘of or belonging to a cushion’, referring to the cushion-forming growth habit of this species.

Specimens examined. Specimens marked with an asterisk were not used in our morphometric study, but were selected from the three major New Zealand herbaria as additional representative specimens. Nelson, Julius Rocks, Travers Range, M.J.A. Simpson 5052, 26 Jan 1966 (CHR 171376); Nelson, Julius Rocks, Travers Range, M.J.A. Simpson 5054, 26 Jan 1966 (CHR 171377); Nelson, Ridge to Julius Rocks, Travers Range, M.J.A. Simpson s.n., 26 Jan 1966 (CHR 172077); Nelson, Ridge above Third Basin, Travers Range, M.J.A. Simpson s.n., 26 Jan 1966 (CHR 172083); Nelson, Waiau Pass, Spenser Mountains, M.J.A. Simpson 5729, 12 Jan 1970 (CHR 204985); Nelson, Waiau Pass, Spenser Mountains, M.J.A. Simpson 5724, 12 Jan 1970 (CHR 204988); Nelson, Waiau Pass, Spenser Mountains, M.J.A. Simpson 5731, 12 Jan 1970 (CHR 204997); Canterbury, Mt St Patrick, Amuri Ski Basin, Clarence Valley, M.J.A. Simpson 6507 (CHR 225904); Canterbury, Mt St Patrick, Amuri Ski Basin, Clarence Valley, M.J.A. Simpson 6527, 8 Jan 1972 (CHR 225924); Canterbury, Mt St Patrick, Amuri Ski Basin, Clarence Valley, M.J.A. Simpson 6529, 8 Jan 1972 (CHR 225926); Canterbury, Mt Princess, St James Range, M.J.A. Simpson 6417, 17 Dec 1971 (CHR 531030); Unknown origin and collector (CHR 571417); Canterbury, Mt Princess, St James Range, Matt Todd s.n., 13 Jan 1996 (CHR 669242); Nelson, Mount Robert/Pourangahau, Travers Range, C.C. Nicholls CN10, 15 Feb 2022 (CHR 684561); Nelson, Mount Robert/Pourangahau, Travers Range, C.C. Nicholls CN13, 15 Feb 2022 (CHR 684565); Canterbury, Mt St Patrick, St James Range, J.M. Ward 05040, 22 Feb 2005 (CHR 684536); Marlborough, Mt Schiza, Matt Todd 610, 4 Feb 1996 (CHR 684568); Marlborough, Altimarlock, Black Birch Range, J.M. Ward 05048, 24 Feb 2005 (CHR 684532); Nelson, Thompson Pass, J. McSweeney s.n., 20 Feb 1973 (CANU 29301); Nelson, Red Hills, Matt Todd MT39 (CHR 684566); Marlborough, Mt Altimarlock, Black Birch Range, N.C. Lambrechtsen s.n., 18 Nov 1969 (CHR 197901)*; Marlborough, Mt Richmond, Richmond Range, A.P Druce s.n., Dec 1985 (CHR 401550)*; Marlborough, Tapuae-o-Uenuku, Inland Kaikoura Range, B.C. Aston s.n., Dec 1915 (WELT 53541/A)*; Marlborough, Bounds Range, Pinnacle, W. Martin s.n. (WELT 53592)*; Canterbury, Mt St Patrick, St James Range, M. Rixon s.n., 14 Feb 1998 (AK 366081)*; Marlborough, Raglan Range, W. Martin s.n., 1933 (AK 235917)*.

Cytology. No information. The specimen counted (CHR 102512; Beuzenberg and Hair 1984) was identified by us to be Haastia minor.

Haastia minor (Laing) C.C.Nicholls, Breitw., J.M.Ward & Pelser, comb. et stat. nov.

≡ Haastia pulvinaris Hook.f. var. minor Laing, T.N.Z.I. 44: 67. 1912.

Holotype. NEW ZEALAND, South Island, Mt Princess, Laing s.n., Dec 1910 (CHR 333878!)

Description. Perennial pulvinate or convex shrubs, stout, pale green to cream-pale green, 0.1–0.2 m tall, individual mounds 0.5–3.0 m in diameter. Rootstock woody. Branchlets (incl. leaves) densely packed forming a closed surface, at surface 8–12(–14) mm in diameter, with leaf apices usually visible. Leaves flabellate, (5–)7–9(–11) x (6–)8–12 mm, length/width ratio 0.75–1.00, length/distance from base to widest point ratio 0.5–0.7, pale green, darkening towards apex; lamina below apex thickened and crenulate, both lamina surfaces densely clad in long, tangled, sometimes slightly fulvous hairs extending 1–2 mm beyond apex; sheath white to pale green, sometimes purple at base; veins (3–)4–7 at base, anastomosing near the apex; apex rounded, forming wide arc, crenulate. Leaves subtending capitula linear 6–7(–10) × 1–4 mm, apex truncate or obtuse. Capitula lateral or terminal, heterogamous and disciform, 3.5–6.5 mm in diameter. Involucral bracts (15–)19–23, valvate, in single row with 1–5 individual bracts outside the row, narrowly oblong or ovate, 5.0–7.5 × 0.5–1.2 mm, length/distance from base to widest point ratio (0.15–)0.40–0.80, hairs present on abaxial side, apex acute with slight fringe of short hairs; veins 1–2, simple and unbranched. Receptacles slightly convex or flat, 1.5–4.0 mm in diameter. Outer florets 18–22, pistillate; corolla narrowly tubular around style, sometimes uneven at apex, (1.5–)2.0–3.0(–5.5) mm long, yellow; style bifurcating, (2.5–)5.0–7.5 mm long, yellow, papillae usually absent, rarely present as tuft of uneven cells protruding from apex. Central florets 12–23, perfect; corolla tubular, 5-lobed, (4.0–)5.5–7.5 mm long, yellow; style bifurcating, (3.0–)5.0–8.0 mm long, yellow, papillae present along abaxial side of style branch, short and evenly distributed up to the style apex where papillae become long and uneven, forming distinct tufts protruding from the style apex; stamen filament collars slender; anthers ecaudate and ecalcarate. Cypselae oblong, glabrous, 1.0–2.5 mm long. Pappus hairs slender, barbellate, up to 8.5 mm long, always extending beyond corolla and style, flattened at base, apex obtuse and fimbriate with cells spreading out towards the tip and protruding unevenly.

Distribution. Aotearoa – New Zealand, endemic. Te Waipounamu – South Island: Eastern Tasman, Marlborough, North Canterbury.

Habitat and ecology. Haastia minor grows on rocky alpine fellfields and stable scree slopes, usually at elevations exceeding 1500 m. The densely packed woolly branchlets and secure root system of H. minor enables the plant to withstand the harsh conditions prevalent in alpine environments. Although it does not consistently co-occur with other cushion- or mat-forming plants, it can occasionally be found in their vicinity, sometimes in direct contact.

Phenology. Due to the alpine habitat Haastia minor grows in, reproduction is limited to the period of time between the spring thaw and the first snow of autumn. It flowers from late December to mid-March and fruits from late February to mid-April.

Conservation status and notes. This species (as Haastia pulvinaris var. minor Laing) has been assessed as ‘Not Threatened’ [Population 20000–100000 mature individuals, with a stable population (±10%)] by the New Zealand Indigenous Vascular Plant Threat Listing Panel (de Lange et al. 2018) using the New Zealand Threat Classification System (Townsend et al. 2008).

Etymology. The specific epithet minor is derived from Latin, meaning ‘smaller, less’, referring to the smaller branchlet diameter compared to H. pulvinaris.

Specimens examined. Specimens marked with an asterisk were not used in our morphometric study, but were selected from the three major New Zealand herbaria as additional representative specimens. Nelson, Julius Rocks, Travers Range, M.J.A. Simpson 3119, 22 Mar 1961 (CHR 149101); Canterbury, Mt St Patrick, St James Range, J.M. Ward 05039, 22 Feb 2005 (CHR 684541); Marlborough, Balaclava Ridge, Crimea Range, Matt Todd 121, 12 Mar 1994 (CHR 684571); Canterbury, Mt Terako, Matt Todd MT52, 10 Apr 1995 (CHR 684567); Marlborough, Balaclava Ridge, Crimea Range, Matt Todd 120, 12 Mar 1994 (CHR 684570); Marlborough, Mt Barefell, Molesworth, I. Breitwieser 905, 11 Feb 1989 (CANU 33562); Marlborough, Mt Barefell, Molesworth, J.M. Ward 91093, 18 Jan 1991 (CHR 684528); Marlborough, Balaclava Ridge, Crimea Range, J.M. Ward 91176, 14 Feb 1991 (CHR 684529); Marlborough, Mt Schiza, Matt Todd 614, 4 Feb 1996 (CHR 684569); Nelson, Mount Robert/Pourangahau, Travers Range, C.C. Nicholls CN11, 15 Feb 2022 (CHR 684562); Canterbury, Mt Terako, J.M. Ward 96081, 14 Feb 1991 (CHR 684531); Nelson, Third Basin, Travers Range, L.B. Moore s.n., 27 April 1962 (CHR 149103); Nelson, Ridge to Julius Rocks, Travers Range, M.J.A. Simpson s.n., 26 Jan 1966 (CHR 172082); Nelson, Waiau Pass, Spenser Mountains, M.J.A. Simpson 5732, 12 Jan 1970 (CHR 204986); Canterbury, Miromiro, Hanmer Range, P.B. Heenan s.n., 16 Mar 2006 (CHR 639534); Marlborough, Balaclava Ridge, A.D. Wilton 119, 12 Mar 1994 (CHR 669235); Canterbury, Mt Princess, St James Range, Matt Todd s.n., 13 Jan 1996 (CHR 669236); Canterbury, Mt St Patrick, St. James Range, D.G. Lloyd, 29 Dec 1970 (CANU 70032); Marlborough, Happy Valley, Seaward Kaikōura Range, J. Anderson s.n., Dec 1977 (CHR 285563)*; Nelson, Gordons Knob, D. Petrie s.n., Feb 1910 (CHR 10173)*; Marlborough, Mt Barefell, Molesworth, I. Breitwieser 959, 11 Feb 1989 (WELT 96333)*; Nelson, Gordon’s Knob, H.H. Allan s.n., Jan 1922 (WELT 96334)*; Marlborough, Balaclava Ridge, Crimea Range, A. E. Wright 12821, 8 Jan 2002 (AK 257351)*; Marlborough, Crimea Range, Mt Maling, P.J. de Lange 6524, 20 Dec 2003 (AK 290577)*.

Cytology. 2n = 60 (CHR 102512; Beuzenberg and Hair 1984).

Notes. Unfortunately, Todd (1996) incorrectly applied the names of the varieties of Haastia pulvinaris in his thesis. He reported that the leaves of the type specimen of H. pulvinaris var. minor (CHR 333878) were long and narrow, a leaf shape associated with the form with wider and branchlets of which the apex is usually obscured by indumentum. However, the type specimen for H. pulvinaris var. minor appears to be a young plant with very short branchlets which may account for the long narrow leaves that Todd assumed were representative of the variety. He appears to have missed the flabellate leaf in the type specimen which matches the typical form of mature leaves of H. pulvinaris var. minor. It seems that, on the basis of his observations of leaf form alone, he assumed the name of H. pulvinaris var. minor was applied to the ‘wider and hairier branchlet form’. He consequently concluded that the type specimen of H. pulvinaris var. pulvinaris (K000844044) was of the ‘greener form with slender branchlets and short, flared leaves’ (Todd 1996, p. 109), despite images of the type specimen clearly showing a plant that belongs to the variety with wider branchlets. As a result, Todd (1996) incorrectly applied the name H. pulvinaris var. minor to the ‘wider and hairier branchlet, narrow leaved form’ and the name H. pulvinaris var. pulvinaris to the ‘greener, more slender branchlet, flabellate leaved form’.

Author contributions

CN: Study design, fieldwork, data collection, data analysis, writing – original draft.

IB: Study design, supervision, writing – review and editing, provision of supplementary information and notes.

JW: Provision of supplementary material (specimens, notes), critically reviewed manuscript.

PP: Study design, supervision, writing – review and editing.

Acknowledgements

The authors thank Ines Schönberger and the staff at CHR for assistance with the examination of specimens, access to databases and the accessioning of collected specimens; curators of AK for facilitating the viewing of specimens in person and curators of WELT for the loaning of specimens; Callum’s mother, Maree, for proofreading and providing suggestions for the manuscript, and Callum’s father, Jeff, for accompanying him in the field.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.