Morphological and genetic variation in the ... - CSIRO Publishing

CSIRO PUBLISHING

www.publish.csiro.au/journals/asb

Australian Systematic Botany 17, 29–48

Morphological and genetic variation in the Senecio pinnatifolius complex: are variants worthy of taxonomic recognition? I. J. RadfordA,D, R. D. CousensB and P. W. MichaelC A

Department of Crop Sciences, University of Sydney, NSW 2006, Australia. School of Resource Management, The University of Melbourne, Vic. 3010, Australia. C 5 George St, Epping, NSW 2121, Australia. D Current address: Botany Department, Otago University, PO Box 56, Dunedin, New Zealand. Corresponding author; email: [email protected] B

Abstract. The current taxonomy of the Senecio pinnatifolius complex (formerly Australian S. lautus) is inadequate in describing intra-specific variation. We present several putative taxa as alternatives to current subspecies, based on variants observed during both herbarium surveys and field studies. We sought to establish whether these taxa were objectively justified in terms of morphology and genetic relationships. This was done in three ways. First, a morphometric study of plants grown under standard conditions was undertaken. Second, isozymes were analysed to establish genetic relationships within the complex. Third, achene morphology was examined by scanning electron microscopy. Variants from central Queensland (BRIGALOW V.) and the deserts of central and Western Australia (DESERT V.) were clearly separated from all other variants based on the number of involucral bracts. This differentiation into two major groups may warrant subspecific recognition. Although variation within each of the proposed subspecies was continuous, separation of variants was possible based on statistical survey. This is consistent with their formal recognition as varieties. Further work is required to determine correct nomenclature of proposed subspecies and varieties, and to fully elucidate variation and provenance in inland forms. SB0313 .IeVJta.lriRatdiofnoirdnteh Seneicopni aitfolius com plex

Introduction The current intra-specific taxonomy of Senecio pinnatifolius (formerly Australian members of the S. lautus complex; Ali 1969) is inadequate in describing variation within this group. In a comprehensive examination of S. pinnatifolius specimens in major Australian herbaria (Radford 1997), one third of specimens fell outside recognised subspecies of S. pinnatifolius according to the key developed by Ali (1969). Several previous authors have also made the observation that S. pinnatifolius is in an unresolved state (Lawrence and Belcher 1986; Michael 1992; Belcher 1993, 1994; Marohasy 1993). Confusion in the S. pinnatifolius group arises for several reasons. First, there is confusion resulting from the inclusion of several different species in Ali’s (1969) paper describing subspecies. These included S. madagascariensis, S. spathulatus and S. anacampserotis specimens. This source of confusion has been partially clarified by taxonomic separation of the exotic S. madagascariensis from the S. pinnatifolius complex (Michael 1981; Radford et al. 1995, 2000; Scott et al. 1998). Second, confusion arises in S. pinnatifolius because many specimens in Ali’s (1969) © CSIRO

9 March 2004

classification are described merely as being similar to particular subspecies (e.g. aff. dissectifolius), as intraspecific hybrids, or simply as anomalous populations. This is despite the fact that some of these variants (e.g. the headland variant within ssp. maritimus) are apparently as numerous and widespread as described subspecies. There is no apparent justification for giving some variants taxonomic recognition while others are dismissed. Third, uncertainty within S. pinnatifolius arises because foreshadowed intra-specific treatment, following from the separation of S. pinnatifolius from New Zealand’s S. lautus sensu stricto (Belcher 1994), has not yet eventuated. During herbarium and field observations of S. pinnatifolius (Radford et al. 1995, 2000; Radford 1997) we have identified several additional groups or variants to those described by Ali (1969). In this paper we present these groups as putative taxa within S. pinnatifolius. We then seek to establish objectively whether our groups are justified in terms of observed patterns of morphological and genetic variation. Variation within S. pinnatifolius was quantified in several ways in this study. First, a morphometric analysis was undertaken of plants grown under standard conditions from 10.1071/SB03013

1030-1887/04/010029

30

Australian Systematic Botany

living material collected in the field. Second, genetic relationships among populations and putative taxa were investigated using isozyme analysis. Third, electron microscopy was used to compare achene surface morphology. Description of S. pinnatifolius This description of S. pinnatifolius, and the ways in which this species is differentiated from other Australian Senecio spp., is based on the key and description of S. lautus by Harden (1992). S. pinnatifolius is a radiate Senecio with marginal florets female, disc florets bisexual and spreading ligules. Ligules are yellow. Involucral bract number is generally recorded as 11–14 (Ornduff 1960; Ali 1964a, 1964b; Harden 1992), although records of 19 bracts in some ssp. dissectifolius specimens (Lawrence and Belcher 1986) have been made. Bracteoles (6–10) are present at the base of the involucre. The plant is glabrous and can be either a herb or a sub-shrub, either erect or prostrate, and can be 10–75 cm high. The leaves are variable, from terrete/linear to lanceolate, obovate or elliptical, and are 2–7 cm long and 3–10 mm wide. Leaf margins can be entire, toothed or lobed/pinnatisect and bipinnatisect. Upper leaves are often dilated or stem clasping at the base. Inflorescences are found in a loose terminal corymb with 10–25 heads, each 4–5 mm in diameter. Ray florets number 9–14 and are 8–12 mm long. Achenes are 1.9–3.0 mm long, grey, brown or green and are usually hairy. Putative taxa within S. pinnatifolius Our proposed intra-specific taxa are briefly introduced here with no indication of their taxonomic status. This includes a qualitative description and an explanation of how the putative taxa relate to previous taxonomic groups put forward by Ali (1969). Much of the comparison will relate to Ali’s (1969) key of leaf shape variation because his taxonomic classification relies primarily upon this character. Ali (1969) divided leaf shapes into three main types, described as α, β and γ. The α leaf shape was oblanceolate, elliptic to lanceolate, entire to dentate margins, with unbranched (α 1–3) to pinnatifid (α 4) and pinnatisect (α 5–6) types. The β leaf shape was filiform to linear, with entire (β 1) and pinniatsect (β 4–5) types. Finally the γ leaf type was oblanceolate with entire to lobed (γ 1–3), pinnatifid, pinnatisect (γ 4) and bipinnatisect (γ 5–6) branching patterns. Descriptions of putative taxa are based on examination of 2993 S. pinnatifolius specimens housed in all the major herbaria in Australia (University of Sydney Herbarium SYD, New South Wales Herbarium NSW, New England University Herbarium NE, Canberra Botanic Gardens Herbarium CBG, CSIRO Australian National Herbarium CANB, Queensland Herbarium BRI, South Australian Herbarium AD, National Herbarium of Victoria MEL, Western Australian Herbarium

I. J. Radford et al.

PERTH, Tasmanian Herbarium HO, Northern Territory Herbarium NT). Maps of distribution and habitat descriptions are compiled from location and habitat data held on herbarium databases and from observations by the authors at field sites (Radford 1997). Senecio pinnatifolius lanceolatus variant This variant conforms closely with Ali’s (1969) ssp. lanceolatus, or his moist gully genoecodeme (Ali 1964a, 1964b, 1966, 1968). Leaf shape is broad and lanceolate with coarsely and sometimes raggedly toothed margins (Fig. 1a). Within Ali’s (1969) key, leaf shape is described as α (1–3) type. Plants of this variant have been found to root from stem nodes and plants form perennial clumps, resprouting from basal stems in spring. Plants fitting this description are found in south-eastern Australia only (Fig. 2a) and are associated with moist high altitude habitats along the Great Dividing Range (Table 1). Senecio pinnatifolius RANGE variant Leaf shape is narrow lanceolate to oblanceolate and membranous with finely toothed margins and often with lobes or branches, particularly higher on the plant (Fig. 1b). Specimens of this variant were often determined by Ali (1969) as intra-specific hybrids (ssp. lanceolatus and ssp. dissectifolius) both in his taxonomic description and on herbarium specimens. Despite this applied hybrid status, this variant is widespread across eastern NSW (Fig. 2b) in relatively dry habitats along the Great Dividing Range (Table 1). Leaves would be described as α-type within Ali’s (1969) classification, but more dissected than lanceolatus. Leaves generally appear narrower and shorter than those of lanceolatus and broader than those of dissectifolius (Fig. 1b, f). These plants have similar leaf shape to some DUNE variant plants except that leaves tend to be narrower and are not succulent. Senecio pinnatifolius alpinus variant This variant is the same as Ali’s (1969) ssp. alpinus and Belcher’s (1996) var. pleiocephalus. Leaf shape was described as γ-type by Ali (1969). Leaves are many-lobed and lobes themselves have secondary branches or are subdivided (Fig. 1c). Plants of this variant have both a prostrate and an erect habit (at exposed sites ground hugging clumps of leaves form from which erect stems bearing inflorescences arise) and opportunistically roots arise from stem nodes leading to the formation of continuous clumps which are presumably a single genet. Some forms of the alpinus leaf could be difficult to distinguish from dissectifolius or some lobed forms of ssp. maritimus. The alpinus variant is found in a range of alpine habitats (Table 1) in NSW, Victoria and Tasmania (Fig. 2c).

Variation in the Senecio pinnatifolius complex

a


f

31

j

g b

c h

d

e

i

Fig. 1. Scanned images of leaf shape for variants of S. pinnatifolius. Variants shown are (a) lanceolatus, (b) RANGE V., (c) alpinus, (d) HEADLAND V., (e) DUNE V., (f) dissectifolius, (g) BRIGALOW V., (h) DESERT V., (i) W. NSW V. and (j) Bass Strait Island V. Scale bar = 10 mm.

Senecio pinnatifolius HEADLAND variant Leaves of the HEADLAND variant are oblanceolate to lanceolate, toothed or entire and sometimes lobed. Leaves

are distinguished from other variants by their small size and fleshiness (Fig. 1d). Leaves are consistent with Ali’s (1969) α (1,2) or γ (1–3) leaf forms (Fig. 1d). Plants have a prostrate

32



a

e

b

f

c

g

d

h

i

j

Fig. 2. Maps of distribution for intra-specific S. pinnatifolius variants. (a) lanceolatus, (b) RANGE V., (c) alpinus, (d) HEADLAND V., (e) DUNE V., (f) dissectifolius, (g) BRIGALOW V., (h) DESERT V., (i) W. NSW V. and (j) Bass Strait Island V. based on herbarium records. Each point on the map represents a single plant specimen.

or cushion habit distinguishing them from typical ssp. maritimus, which often has a sprawling habit (Ali 1969). Ali (1969) described plants of this variant as anomalous or simply as aff. ssp. maritimus (headland form). This variant occurs throughout the southern coastline of Australia (Fig. 2d) and is associated with rocky headlands and cliffs (Table 1).

Senecio pinnatifolius maritimus or DUNE variant The DUNE variant corresponds closely with Ali’s (1969) ssp. maritimus and has a similar range of leaf shapes to the HEADLAND V. (broad lanceolate to obovate, often with leaf serrations on margins, Fig. 1e), but leaves are generally larger. Leaf shape is described as


Table 1.


Habitat descriptions for S. pinnatifolius variants

S. pinnatifolius taxa/variant lanceolatus V.

RANGE V.

alpinus V.

HEADLAND V.

DUNE V. dissectifolius V.

BRIGALOW V.

DESERT V.

W. NSW V.

BASS STRAIT V.

33

Habitat description Wet subalpine tableland habitat on Great Dividing Range in NSW and south-east Queensland, often on basalt, dominated by Eucalyptus pauciflora woodland with Poa labillardieri understorey; also in moist gully habitats in Victoria Rocky creek banks and ridge lines along Great Dividing Range in NSW and Victoria, range of geologies, dry sclerophyll vegetation such as Eucalyptus eugenioides, Angophora floribunda and Casuarina torulosa High alpine meadows and heath in NSW, Victoria and Tasmania, particularly where vegetation is sparse; associated plants include Craspedia, Celmisia, Cyanthoides and Poa spp. on dolerite, granite or basalt Exposed coastal headlands and cliff tops throughout southern Australia, often on basalt, sandstone, mudstone, granite, limestone or dolerite; in NSW vegetation often dominated by Themeda triandra with scattered coastal heath Coastal beaches and sand dunes throughout southern Australia, particularly where vegetation is sparse; often associated with Spinifex sericeus, Acacia sophorae and Scaevola calendulacea Wide range of dry inland habitats in non-tropical parts of Australia, in NSW on poorly vegetated hills with skeletal soils, often on sedimentary rocks, often with Callitris endlicheri and Eucalyptus dealbata in inland NSW, or in mallee habitats in Victoria Associated with brigalow on low hills in central and southern Queensland, Acacia aneura, A. harpophylla and A. cambagei and nearby cleared alluvial pastures; often found in cattle and horse paddocks as a local weed Occurs in large areas of central and Western Australia on intermittent water courses including swamps, river and creek beds, flood-plains and alluvial depressions on a range of soil types. Associated vegetation includes chenopod grassland, spinifex, Eragrostis australasica, Eucalyptus microtheca and Acacia aneura Associated with alluvial areas in mallee and shrubland on a wide range of soil types in inland NSW and adjoining regions. Associated with Eucalyptus microtheca, E. camaldulensis and E. largiflorens Found on exposed southern Australian offshore islands. Associated vegetation includes tussock grassland and low heath or herbland including Atriplex spp. and Stipa spp.

α-form by Ali (1969). DUNE V. leaf shape appears to overlap with RANGE V., dissectifolius and lanceolatus in some instances, however, may be distinguished by succulence. This variant occurs throughout temperate and subtropical coastal Australia (Fig. 2e) on coastal sand dunes and beaches (Table 1). Senecio pinnatifolius dissectifolius variant The dissectifolius variant is recognisable on the basis of narrow to linear leaves which are often branched or dissected (Fig. 1f) corresponding to Ali’s (1969) subspecies of the same name. This variant is described by Ali (1969) as having a β leaf form. This form is quite variable and there may be several sub-forms differing in leaf width, number and pattern of dissections. Despite these apparently clear cut leaf shape characteristics dissectifolius appears to overlap morphologically with many of the other variants that also have branched or dissected leaf shapes. This variant is found throughout much of the southern half of Australia (Fig. 2f) in a range of predominantly inland habitats (Table 1). Senecio pinnatifolius BRIGALOW, DESERT and WESTERN DIVISION NSW variants These variants probably fall within what Ali (1964a, 1964b, 1966) described as the desert genoecodeme and are

represented by several broad dissected, lobed or ragged-toothed leaf forms (Fig. 1g–i). This genoecodeme was not formally described by Ali (1969) despite being included in previous analyses (Ali 1964a, 1964b, 1966). The BRIGALOW Variant (one form corresponding with the desert genoecodeme) has broad highly lobed to pinnatisect leaves (Fig. 1g), while the DESERT V. leaf form appears to be broader, more irregular in shape and toothed rather than branched or lobed (Fig. 1h). The leaf lamina appears distinctly membranous in dried specimens. Both the DESERT and BRIGALOW variants appear to have more involucral bracts around their capitula (approximately 19 compared with 13) than the WESTERN DIVISION NSW V. (W. NSW V.). The W. NSW V. is otherwise similar in leaf shape (Fig. 1i) to the DESERT V. The BRIGALOW variant has a central- and southernQueensland distribution (Fig. 2g), while the DESERT V. occurs in central and Western Australia (Fig. 2h). W. NSW V. occurs in the Western Division of NSW and may grade into the DESERT V. westwards (Fig. 2i). The BRIGALOW V. is naturally associated with Brigalow scrub vegetation (Acacia aneura) but also occurs as a weed in central Queensland pastures (Noble et al. 1994), whereas most S. pinnatifolius variants are not associated with agriculture (Sindel et al. 1998). This variant has been implicated in the death of cattle

34


after ingestion, owing to the presence of toxic pyrrolizidine alkaloids in S. pinnatifolius plant tissues (Noble et al. 1994). DESERT and W. DIVISION variants are found associated with ephemeral alluvial habitats in arid and desert areas of far inland Australia (Table 1). Living material was only able to be collected for the BRIGALOW V. Senecio pinnatifolius Bass Strait variant Typical leaf shape for this variant is fine pinnate to bipinnatisect, often with secondary branches, leading to an almost fern-like appearance (Fig. 1j). Most similar leaf shapes would be alpinus leaves, but leaves were generally smaller and finer than this type. We only observed this form in herbaria and therefore do not know what habit these plants have. Distribution is islands off the southern coast of Australia (e.g. Bass Strait and islands off South and Western Australia) (Table 1, Fig. 2j). Failure to collect living material from this variant has meant that we have been unable to include it in further analysis. Materials and methods Sampling strategy Owing to limitations of budget and time, collection of comprehensive samples of living material for further analysis was confined to eastern NSW and Victoria, south-eastern Queensland and eastern and central Tasmania (Table 2). A few additional samples were collected in South Australia, Western Australia and central Queensland. We did not collect any samples in the western Division of NSW, in central Australia or in the majority of Western or South Australia. Living material was obtained for lanceolatus, RANGE, alpinus, HEADLAND, DUNE/maritimus and dissectifolius variants within these regions. Geographically limited material was also obtained from the BRIGALOW variant close to Biloela in central Queensland. No living material was obtained for DESERT, Bass Strait or W. DIVISION variants. Members of the S. lautus complex within New Zealand (S. lautus sensu stricto) were also excluded from this analysis. Living material collection Achenes (‘seeds’) were collected from a minimum of 10 plants per site (and where possible many more) in order to sample a range of genotypic variation within S. pinnatifolius populations. Collections were made from well-spaced individuals to reduce the chance of repeat sampling of single genets; perennial lanceolatus and alpinus plants often formed continuous clumps which may have been made up by a single individual, so we made sure that we sampled from at least 10 of these. Single bulked collections were made for each population. Achenes were collected from fully opened capitula, where possible, to ensure maximum maturity and viability. Achenes were stored at a constant temperature (20°C) with silica gel to reduce humidity. These collections were used to provide living material for morphometric and isozyme studies. Common garden experiment Achenes from each bulked population sample (accession) were sown into five replicate pots. Three lanceolatus (Population no. 1–3, Table 2), two RANGE (Population no. 6–7, Table 2), three alpinus (Population no. 14–16, Table 2), 13 HEADLAND (Population no. 19, 21, 23–33, Table 2), eight DUNE (Population no. 33–37, 40–42, Table 2), seven dissectifolius (Population no. 43, 44, 46, 52, 54, 56, 57, Table 2) and


three BRIGALOW accessions (Population no. 58–60) were represented in this experiment. Sowing took place on 5 September 1994. Achenes were placed into 12-cm diameter pots into a 1:1 river gravel/peat moss mixture. Slow release fertiliser (Osmocote®) was applied to each pot. Plants were watered daily. Seedlings were grown in a 18/25°C controlled glasshouse under ambient light. Juvenile plants were thinned to one per pot in October 1994 and pots moved outside into a fully randomised design for the remainder of the experiment. Re-randomisation was done at two-weekly intervals throughout the experiment. Plants were harvested once they had started to produce mature capitula (4 December 1994–6 July 1995) and several morphological measurements were conducted (see below). High mortality among some populations due to white fly attack caused unbalanced population sets. Supplementary planting of DUNE, BRIGALOW and alpinus variants took place from 27 October 1994 for later harvest. Measurement of morphological attributes Each plant was assessed for habit, floral attributes and leaf thickness while alive and in the pot. Involucral bract number, number of ray florets and number of disc florets were counted on three capitula per plant. Total inflorescence diameter (across ray florets), individual ray floret length and breadth, capitula diameter at the base and capitulum length were measured. Two of each of these measurements were made, as sub-samples, for each individual plant. Use of two measurements from different randomly chosen capitula for each plant appeared to give an adequate estimate of each plant’s mean as capitula were not highly variable (individual plants were the replicates within the experimental design). Habit was quantified by measuring vertical height and horizontal spread of each individual plant at the widest point. Vertical height at the widest point was also measured. Ratios of plant width/height, height at the widest point/plant height and height at the widest point/plant width were also calculated. Leaf thickness was measured with callipers on two randomly chosen leaves per plant. Once habit, capitula and leaf thickness were assessed, sample leaves were removed from each plant to be pressed and dried for measurement of leaf attributes and shape. To sample a full range of variation in leaf shape relative to nodal position, the 3rd, 6th, 9th and 12th nodal leaves were sampled from the main stem. Secondary and tertiary branches were also sampled but owing to differences in branching structure among variants no standard set of leaves could be defined. Therefore one secondary and one tertiary branch were chosen, where applicable, and a leaf was sampled from every third node on these branches. Leaf shape was quantified in two ways. First, measurements were made of representative leaf traits to give a broad description of leaf shape. Measurements used were leaf length, leaf width, leaf span (width of external perimeter inclusive of lobes), number of lobes or branches per leaf and the number of serrations on the leaf margin (one side of the leaf only). Three leaves from at least five individuals per variant were measured where possible, for a total of 40 plants. Second, leaf shape was quantified by a Noran TN-8502 image analysis package (Noran Instruments, Middleton, WI). Image analysis had the advantage over hand measurements that many leaves and individual plants could be measured simultaneously for multiple traits. This allowed characterisation of plants based on leaves from a range of stem and branch positions. A total of 1208 leaves and 180 individual plants were characterised using 11 separate leaf shape measurements. Similar use of computer based image analysis for variable taxonomic complexes are described by Kincaid and Schneider (1983), West and Noble (1984) and White et al. (1988). Image analysis operated in the following way. First, a Soni® CCD DXC-3000P Camera was used to capture leaf images at a 1:1 scale. The image analysis package was then used to quantify leaf shape



Table 2. Achene (seed) collection sites (accessions) for S. pinnatifolius variants NT NSW, Northern Tablelands New South Wales; SE QLD, south-eastern Queensland; CT, Central Tablelands NSW; ST, Southern Tablelands NSW; Alps VIC, Victorian Alps; TAS, Tasmania; FNC, Far North Coast; CC, Central Coast NSW; NC, North Coast NSW; SC, South Coast NSW; CWS, Central Western Slopes NSW; NWS, North Western Slopes NSW; W VIC, Western Victoria; SA, South Australia; C QLD, Central Queensland; W QLD, Western Queensland Taxa/variant lanceolatus

RANGE V.

alpinus

HEADLAND V.

DUNE V.

dissectifolius

Locality A

Point Lookout, Dorrigo Cobark Lkt, Barrington TopsA Pole Blue, Barrington TopsA O’Reilly’s, Lamington NP Uni Rover Trail, KanangraA Kalang Falls, Kanangra Kowmung R, S Blue Mountains Snowgum walk, KosciuskoA Daners Ck, KosciuskoA Perisher Valley, Kosciusko Merrits Pass, Kosciusko Rawson’s Track, Kosciusko Mt Gingera, Brindabella’s ACT Falls CreekA Mt BogongA Mt NiggerheadA Mt Wellington, HobartA The Temple, Central Highlands White HeadA Norries HeadA Point Lookout, Stradbroke IslandA Garie, Royal National ParkA Cape Byron Lennox Head Evans Head Coffs Harbour, South Headland Hat Head Tacking Point, Port Macquarie Kiama Blowhole Green Cape Remarkable Cave, Port ArthurA Tessellated Pavement, Pt Arthur Shelly Beach, BallinaA Evans Head beachA Lennox Head beachA Amity Point, Stradbroke Island South West Rocks Coffs Harbour beachA Mungo Brush, Myall LakesA Green Patch, Jervis Bay Moruya airport Roaring Beach, Bruny IslandA Callatoota Vineyard, DenmanA MurgaA Euraldrie Trig, GrenfellA Tamworth Yarraman Vineyard, Denman Sandy Hollow Mt Canobolas Mittagong NSW Cox’s River, Blue Mountains Red Cliffs, Murray RiverA

Region

Latitude, longitude

NT NSW NT NSW NT NSW SE QLD CT NSW CT NSW CT NSW ST NSW ST NSW ST NSW ST NSW ST NSW ST NSW Alps VIC Alps VIC Alps VIC TAS TAS FNC NSW FNC NSW SE QLD CC NSW FNC NSW FNC NSW FNC NSW NC NSW NC NSW NC NSW SC NSW SC NSW TAS TAS FNC NSW FNC NSW FNC NSW SE QLD NC NSW NC NSW CC NSW SC NSW SC NSW TAS CWS NSW CWS NSW CWS NSW NWS NSW CWS NSW CWS NSW CT NSW CT NSW CT NSW W VIC

30°29′S, 152°24′E 31°53′S, 151°35′E 31°57′S, 151°25′E 28°13′S, 153°07′E 34°05′S, 150°05′E 33°59′S, 150°07′E 34°03′S, 150°09′E 36°25′S, 148°25′E 36°22′S, 148°27′E 36°25′S, 148°25′E 36°30′S, 148°17′E 36°22′S, 148°28′E 35°35′S, 148°42′E 36°29′S, 152°24′E 36°34′S, 147°19′E 36°55′S, 147°14′E 42°53′S, 147°14′E 41°49′S, 146°18′E 28°51′S, 153°35′E 28°19′S, 153°35′E 27°26′S, 153°32′E 34°09′S, 151°04′E 28°39′S, 153°39′E 28°49′S, 153°37′E 29°09′S, 153°28′E 30°19′S, 153°09′E 31°04′S, 153°04′E 31°29′S, 152°56′E 34°40′S, 150°51′E 37°16′S, 150°03′E 43°12′S, 147°50′E 43°01′S, 147°58′E 28°52′S, 153°34′E 29°04′S, 153°27′E 28°48′S, 153°37′E 27°24′S, 153°28′E 30°54′S, 153°03′E 30°19′S, 153°09′E 32°32′S, 152°19′E 35°07′S, 150°45′E 35°54′S, 150°09′E 43°27′S, 147°15′E 32°18′S, 150°38′E 33°22′S, 148°33′E 33°59′S, 148°02′E 31°04′S, 150°07′E 32°16′S, 150°38′E 32°19′S, 150°34′E 33°19′S, 148°59′E 34°29′S, 150°30′E 33°44′S, 150°11′E 34°18′S, 142°11′E

35

36



Table 2. Taxa/variant

BRIGALOW V.

DESERT V.

(continued)

Locality Renmark, Murray River Port GermeinA Wilpena Pound, Finders Ranges Coorong Grass Tree Hill, Nth HobartA BiloelaA Nth Biloela ‘Elsewhere’, Jambin Windorah

Region

Latitude, longitude

SA SA SA SA TAS C QLD C QLD C QLD W QLD

34°10′S, 140°44′E 33°01′S, 138°01′E 31°37′S, 138°36′E 35°40′S, 139°08′E 42°42′S, 147°08′E 24°24′S, 150°30′E 24°20′S, 150°28′E 24°13′S, 150°21′E 25°25′S, 142°39′E

A

Populations for which isozyme analysis was undertaken.

parameters. Each leaf image was digitised and grey-scale thresholds adjusted to distinguish leaf from shadows or anomalous particles. Non-leaf material was edited from images. A binary image of each leaf was then created upon which multiple measurements were performed. Leaf measurements were as follows. (i)

Leaf area and external perimeter (microns) were measured to give an indication of leaf size. (ii) Circularity (defined by the equation: circularity = Perimeter2/4π × area) was used to give a measurement of a leaf ’s similarity in shape to a circle. (iii) A measure of the ‘roughness’ of the leaf margin caused by lobes, dissections or teeth was calculated using coefficients of the 20-harmonic Fourier description of the perimeter as described by Kincaid and Schneider (1983), calculated using the following equation: Roughness = √[Σ(Ri2+Ii2)/2], where Ri refers to the real part and I to the imaginary part of the ith harmonic. (iv) Convex area, and convex perimeter measured the area and perimeter of each leaf outline. These measurements draw a line between leaf dissections and therefore fill in gaps between leaf divisions. (v) Convex circularity measured the difference of the convex shape described above from a circle by the equation: convex circularity = (convex perimeter)2/4π(convex area). (vi) Maximum and minimum leaf projections were measured using the convex area of a leaf (see above) and were the longest and the shortest line drawn through the centre of a leaf from one end to the other. (vii) Mean and standard deviation of projections were statistical measures based on all the possible straight lines drawn across a leaf’s convex area. (viii) Width, which was the measure of distance in a straight line across the leaf at right angles to the maximum projection and may give a different measurement of span depending on how symmetrical the leaf is. (ix) Aspect ratio was calculated by the maximum projection/width and may be used to give an approximation of leaf elongation. In order to quantify additional leaf shape attributes, independent of leaf size, the following ratios were calculated: maximum projection/leaf perimeter, leaf surface area/convex area, and leaf perimeter/convex perimeter. Analyses of morphometric data Analysis of variance was used to test for differences between variants for all morphological measurements (leaf dimension, leaf image, habit

and capitula attributes). Homogeneity of variance was tested before ANOVA using Levene’s and Bartlett’s tests. Where variance was heterogenous a non-parametric statistic, the Kruskal–Wallis test, was used to test for differences in traits. Principal component analysis (PCA) was used to examine variation among hand-measured attributes. These measurements included leaf dimensions, habit and capitula attributes. Pearson’s correlations were used to test for the strength of individual attribute contribution to differentiation in scatter plots. Canonical variates analysis (GENSTAT-5) was used to analyse variation in leaf shape among individual plants. Unlike PCA, Canonical variates analysis used information on all leaves and attributes from each individual plant to summarise variation to a single point. This allowed comparisons between many individuals comprised of multiple leaf variables, rather than between single leaf attributes. Canonical variates analysis allowed us to compare larger datasets than PCA.

Procedures for isozyme electrophoresis Bulked achenes from each population (achenes sampled from multiple maternal plants) were placed on moist filter paper (Whatman® No. 1, Whatman, Kent, UK) and imbibed with 1.63 mM gibberellic acid (GA3) to overcome dormancy. Seedlings were transplanted into pots in a standard 1:1 gravel/peat mixture and fertilised with slow release Osmocote® pellets. Seedlings were grown to the four-leaf stage (excluding cotyledons) to provide sufficient leaf material for protein extraction and to increase levels of enzyme activity to allow good staining. A minimum of 10 seedlings was assayed (a maximum of five per run) for each population by starch gel electrophoresis. The electrophoretic procedures used were mostly as described by Soltis et al. (1983) and Moran and Hopper (1987). Actively growing young leaf material was ground on ice in two drops of Tris–HCl buffer (0.01 M EDTA, KCl, MgCl.4H20, pH = 7.5) containing 0.1% (v/v) 2-mercaptoethanol and 10% (w/v) PVP 40000 (polyvinylpyrrolidone) to inhibit the breakdown of proteins by secondary metabolites (Kephart 1990). Protein extracts were then absorbed on to 10 × 3 mm filter paper wicks (Whatman No. 1) and loaded onto 13% starch gels. Gels were made according to procedures described by Kephart (1990). Bio-Rad (Hercules, CA) power packs were used to apply a constant current of 50–60 mA through gels via platinum electrodes in filled buffer trays and electrolyte bridges made of Wettex® cloth. Electrophoresis was run for 6 h at 4°C to allow sufficient protein separation with minimum dispersion of enzyme material. When each electrophoretic run was completed, gels were cut into four horizontal slices (using fine metal wire) to be stained for four enzyme systems. Two gels were run at a time to stain for a total of eight separate enzyme systems. A total of 17 separate enzyme systems were trialled on S. pinnatifolius material during the study. Stain recipes were taken from Soltis et al. (1983).



37

Table 3. Enzyme systems used for isozyme analysis and number of loci consistently scored in S. pinnatifolius plants Boric acid/tris-citrate: electrode buffer 0.100 M NaOH, 0.300 M boric acid; gel buffer 0.015 M tris, 0.004 M citric acid. Tris/citric acid: electrode buffer 0.223 M tris, 0.086 M citric acid; gel buffer 0.008 M tris, 0.003 M citric acid Enzyme system Acid phosphatase (AP) Aspartate amino transferase (AAT) Phosphoglucoisomerase (PGI) Triosephosphate isomerase (TPI) Isocitrate dehydrogenase (IDH) 6-Phosphogluconate dehydrogenase (6-PGD) Shikimate dehydrogenase (SkDH) Total

Seven out of the total 17 enzyme systems trialled produced consistently scorable bands (Table 3). From these seven enzyme systems 13 loci were identified. Genetic interpretation of stained isozyme bands followed Pasteur et al. (1988). Interpretation of enzyme systems was straightforward in all except phosphoglucoisomerase (PGI). This system had a complex banding pattern whereby two loci apparently overlapped in space (allele B of PGI-2 occurred on top of allele A of PGI-3). For most individuals PGI-2 and PGI-3 were independent (the occurrence of homozygote and heterozygote bands in individuals for the two systems were unrelated). However, in some individuals some alleles from each of these loci (for instance allele A from PGI-2 and allele B of PGI-3) would form typical heterozygote banding patterns between loci (three bands). To allow this variation to be included within the analysis we interpreted this banding pattern as a fourth independent loci (PGI-4), though no crossing experiments were conducted to test whether this was the case. Failure of staining for triosephosphate isomerase (TPI) loci in many gel runs necessitated removal of some populations and individuals from the analysis. Seedling material used for isozyme analysis was of equivalent age and grown under standard glasshouse conditions throughout the study, however, daylength and quality of light is likely to have differed for plants grown at different times of the year. Staining for TPI differed between gel runs during the year, not among individuals or populations, indicating that differences in expression of this allele were not genotype related. Failure of isozymes to stain for this system may have been due to changes in enzyme activity with season. For this reason isozyme data was only available for five BRIGALOW and 20 RANGE variants individuals. Isozyme data analysis Analysis of isozyme genotype arrays was performed using BIOSYS-1 (Swofford and Selander 1981, 1989). Individual genotype data were entered and used to calculate allele frequencies. These were then used to calculate hierarchical genetic distance statistics and genetic relatedness of populations. Calculations were made of the percentage of alleles within two allelic distribution classes, similar to those described by Moran and Hopper (1987). Alleles were divided into widespread (in more than one variant) and localised (found in one population only). The unweighted pair group method with arithmetic averaging (UPGMA), using Nei’s (1978) unbiased genetic distance between pairs of populations was used to construct clusters of related populations. An UPGMA cluster was also constructed to directly assess relationships between identified variants, pooling accessions into single populations for the analysis. The robustness of groups thus formed was tested by comparing dendrograms constructed from different distance measures between populations. Roger’s genetic similarity, Nei’s genetic identity and modified Roger’s distance measures (all cited by Swofford and

Buffer system

No. loci

Boric acid/tris-citrate Boric acid/tris-citrate Boric acid/tris-citrate Boric acid/tris-citrate Tris/citric acid Tris/citric acid Tris/citric acid

01 01 05 02 01 02 01 13

Selander 1989) were used for comparison with UPGMA to test for the robustness of relationships. Achene measurements Achene length was measured using an eyepiece micrometer under a dissecting microscope. Randomly chosen achenes (10) were measured from bulked population samples (see above) for 8 lanceolatus, 1 RANGE V., 16 alpinus, 6 HEADLAND V., 6 DUNE V., 20 dissectifolius and 3 BRIGALOW V. population accessions. Detailed observations of achene surface morphology was made by scanning electron microscopy (s.e.m.). Achenes were directly mounted on s.e.m. stubs with double-sided adhesive tape. The s.e.m. stubs were gold coated in a Dynavac sputter coater and examined in a Cambridge S360 Stereoscan electron microscope operated at 15 kV. Achenes of lanceolatus (four), alpinus (four), HEADLAND V. (eight), DUNE V. (four), dissectifolius (two) and BRIGALOW V. (two) were examined in this way. Achenes examined were taken from bulked population collections made in the field (see above) and so it was not possible to determine whether individual achenes were from ray or disc florets. Detailed observations of these achenes were used to characterise and describe several morphological achene types. Observations included measurement of achene hair length, the width of longitudinal grooves and raised ridges, and groove to width ratio. More general observations of bulked achene collections were then made to determine how common different achene morphological types were within and among populations. At least 30 achenes from each of 4 lanceolatus, 1 RANGE V., 11 alpinus, 4 HEADLAND V., 5 DUNE V., 9 dissectifolius and 1 BRIGALOW V. accession were examined and the most common type identified for each. Statistical analysis of achene morphology was not attempted owing to the small number of achenes that were measured in detail.

Results Single morphological attributes Separation was possible among putative taxa for several morphological attributes (Table 4). Involucral bract number separated BRIGALOW V. plants from all other groups [19 compared with 13–14 with a maximum least significant difference (l.s.d.) for comparisons of 2.3]. Of the other measurements, 16 of 18 also showed significant differences among various putative taxa. However, the range of values for these measurements overlapped between groups. For instance, dissectifolius plants had the narrowest leaves but could not be separated from DUNE plants with narrow

n = 60 2.14 (0.03)

Achene length (no. measurements) n = 80 n = 10 n = 160 Length (mm) 2.71 (0.03) 2.84 (0.09) 2.62 (0.03)

Loge transformed for ANOVA; the means presented here were back-transformed.

A

18.6 (1.1) 3.5 (0.3) 4.9 (0.5) 1.7 (0.3) 3.0 (0.4) n = 48 0.71 (0.02)

35.1 (4.4) 35.5 (6.0) 43.4 (5.9) 9.3 (1.2) 4.3 (1.5) 4.3 (0.6) 9.3 (1.2) 10.7 (1.6) 14.0 (1.9) 1.0 (0.0) 5.3 (1.3) 6.0 (0.8) 10.1 (0.8) 7.2 (1.9) 10.0 (3.2) n=5 n=2 n = 32 Succulence (leaf thickness) (mm) 0.63 (0.03) 0.38 (0.03) 0.56 (0.02) n = 60 2.89 (0.05)

34.1 (2.6) 3.1 (0.4) 3.5 (0.6) 1.1 (0.1) 3.8 (0.6) n = 31 0.64 (0.02)

n = 31 13.2 (0.2) 11.6 (0.3) 64.4 (2.6) n = 32 21.2 (0.8) 7.7 (0.5) 4.5 (0.1) 7.5 (0.2) 2.6 (0.1) n = 29 20.8 (1.2) 33.8 (2.6) 12.1 (1.4) 1.7 (0.2) n = 17

HEADLAND DUNE V. V. n = 31 13.4 (0.2) 12.4 (0.2) 64.9 (2.6) n = 32 18.6 (0.7) 6.7 (0.3) 4.5 (0.1) 7.5 (0.2) 2.4 (0.1) n = 24 9.6 (0.4) 28.3 (1.7) 0.5 (1.6) 3.0 (0.2) n = 23

alpinus – – – – n=2 18.7 (0.5) 7.2 (0.1) 4.2 (0.0) 6.6 (0.2) 2.8 (0.2) n=5 8.8 (2.0) 24.4 (3.8) 1.6 (2.4) 0.4 (0.1) n=7

n = 10 13.0 (0.1) 12.4 (0.5) 53.6 (3.7) n = 16 23.2 (0.8) 8.9 (0.5) 4.1 (0.2) 6.9 (0.3) 2.8 (0.1) n = 11 25.0 (2.7) 27.6 (2.7) 15.4 (3.0) 1.2 (0.2) n=8

Flowers (no. heads examined) No. involucral bracts No. ray florets No. flowers Capitula (no. heads examined) Total width incl. ray (mm) Ray floret length (mm) Diameter of head (mm) Length of head (mm) Ray floret width (mm) Habit (no. plants examined) Height (cm) Width (cm) Height at widest point (cm) Width/height ratio Leaf dimensions (no. measurements) Length (mm) Width (mm) Span (mm) No. dissections No. serrations

n=5 13.8 (0.8) 12.0 (0.3) 52.2 (4.0) n = 10 18.6 (0.9) 7.1 (0.5) 4.1 (0.2) 7.4 (0.2) 2.2 (0.1) n=5 19.6 (2.2) 31.6 (2.3) 10.2 (5.2) 1.7 (0.3) n=6

lanceolatus RANGE V.

Morphological attribute

n = 200 2.42 (0.02)

29.5 (2.3) 1.1 (0.1) 11.8 (2.0) 4.9 (0.6) 1.5 (0.3) n = 28 0.63 (0.03)

n = 18 13.9 (0.4) 11.3 (0.4) 57.3 (2.4) n = 12 18.3 (0.9) 7.5 (0.5) 4.0 (0.1) 7.0 (0.3) 2.5 (0.2) n = 15 21.3 (1.1) 25.7 (2.1) 13.4 (0.8) 1.3 (0.1) n = 22

n = 20 n = 21 17.0 (0.6) 12.8 (0.1) – – – – – – – – – – – – – – – – – – – – – – – – – – n = 20 n = 21

n = 30 2.16 (0.03)

– –

– –

53.5 (5.9) 57.6 (2.2) 42.2 (3.0) 5.0 (1.0) 6.1 (0.8) 2.5 (0.3) 27.2 (3.1) 19.0 (2.0) 14.2 (1.9) 13.1 (0.6) 3.8 (0.6) 3.6 (0.4) 5.4 (1.1) 5.8 (0.9) 2.1 (0.3) n=4 – – 0.49 (0.01) – –

n=9 19.1 (0.9) 12.7 (0.3) 82.2 (4.7) n = 12 16.7 (1.1) 5.7 (0.6) 4.9 (0.1) 6.9 (0.3) 2.4 (0.2) n = 15 28.4 (2.5) 21.0 (1.7) 20.7 (2.2) 2.0 (1.3) n=8

dissectifolius BRIGALOW DESERT W. NSW V. V. V. h h

H = 321.74

< 0.001

< 0.001

H = 29.50

< 0.001 9 on each side). Achenes in specimens of this variant often exhibited the glabrous (Type 1) achene form. HEADLAND

Fig. 7. Scanning electron micrograph of achene surfaces. Three achene surface morphological types shown are (a) Type 1, which is completely glabrous, (b) Type 2, which has hairs only in longitudinal grooves and (c) Type 3, which is almost completely covered by hairs. These micrographs show achenes of (a) lanceolatus (accession number 4, Table 2), (b) alpinus (acc. no. 9) and (c) DESERT V. (acc. no. 61). Scale bar = 100 µm.

1966, 1968, 1969). Additional traits that differentiate the BRIGALOW V. from other variants were high flower number per inflorescence, high number of leaf branches or dissections and small achenes (Table 5). Although BRIGALOW plants formed an outgroup in the variant level isozyme analysis (Fig. 6) this should not be seen as strong support for separation in itself because only five individuals

This variant is separable on the basis of small leaf size (< 23 mm long), and from all variants except alpinus on the basis of prostrate or cushion forming habit (plants < 10 cm high). HEADLAND plants also generally has small achenes (< 23 mm long) compared with other members of the south-eastern Australian group. This variant can have either entire or lobed leaves. dissectifolius Although leaf width varies and may be broader in many southern populations, generally leaves less than 1.5 mm wide belong to specimens of the dissectifolius variant. Achenes from this variant are also more hairy than other variants, with the majority of achenes examined in many populations of Type 3 almost completely covered in hairs. Leaf branching pattern within this widespread group is quite variable and several names have been used to describe different parts of

46


the complex based on branching pattern (pinnatifolius, dissectifolius and tripartitus; Belcher 1994). Our sampling of live material was confined to eastern Australia and we have not conducted a detailed examination of branching pattern in these plants. We are therefore unable to differentiate plants within this large group further. Members of this group show some evidence of partial genetic divergence from DUNE, HEADLAND and alpinus variants according to isozyme analysis. This lends further support to the recognition of dissectifolius from other variants. RANGE, DUNE and alpinus variants These three variants are more difficult to differentiate based on single quantitative traits and did not separate from one another in multivariate analyses. However, separation of most specimens is possible. RANGE and DUNE variants (the latter corresponding to Ali’s 1969 ssp. maritimus) are separable based on the number of leaf divisions (the former usually has more than three leaf branches; although it should be noted that some populations of DUNE are more dissected, e.g. Moruya, SC NSW). Leaves of the RANGE V. also tend to be narrower and more membranous than those of DUNE (the latter are often succulent and lanceolate in shape). Although alpinus (or var. pleiocephalus as per Belcher 1996) does not clearly separate from DUNE or RANGE variants by quantitative traits, it may be separable by a qualitative trait. The alpine variant is differentiated from both variants within this group by secondary leaf branching and rounded lobe outline. This gives alpinus the distinctive appearance described by Ali (1969) as the γ leaf type.


W. NSW and Bass Strait variants Observations of these variants were confined to specimens held in herbaria and therefore detailed analysis of morphological variation in living specimens was not possible. However, distinctive fine (narrow) and highly branched almost fern like leaves with secondary branching found in Bass Strait plants may allow separation of this from other variants. Though herbarium specimens of the W. NSW variant are similar to RANGE plants, they may be distinguishable on the basis of broader leaves and fewer dissections and teeth on leaf margins. Separation of the W. NSW variant from DUNE plants may also be problematic and may rely on a membranous rather than a succulent leaf form. Implications for taxonomic treatment of S. pinnatifolius Do the morphometric and genetic data presented justify taxonomic recognition of variants? It would appear that all variants described in this paper can be differentiated according to one or several objective traits. This suggests that taxonomic recognition is appropriate. But at what level? Subspecific ranking within S. pinnatifolius for BRIGALOW and DESERT variants on the one hand and all other groups in south-eastern Australia on the other may be taxonomically justified. Clear separation of BRIGALOW and DESERT variants from all others on the basis of a single character (e.g. involucral bract number), within an otherwise intergrading morphological group (see univariate and multivariate data above), with regional separation into geographic races (e.g. Queensland/central Australian) is consistent with the definitions of subspecies (Du Rietz 1930;

Proposed key defining putative variants in S. pinnatifolius Differences used to define putative taxa within S. pinnatifolius have been summarised in the proposed key following. Values are based on interquartile range of plants raised under standard conditions. Owing to variability in S. pinnatifolius, differentiation of plants needs to be based on repeated-measures. The key was tested against herbarium specimens as an initial step towards verifying its usefulness (Radford 1997). The Bass Strait variant has been excluded from the key. 1. Involucral bract number 15–20 2. Leaf lobes/dissections ≤ 6 per leaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DESERT V. 2: Leaf lobes/dissections > 6 per leaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . BRIGALOW V. 1: Involucral bract number 12–14; all south-eastern Australian variants 3. Leaves ≥ 5 mm wide, entire, coarsely dentate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . lanceolatus 3: Leaves < 5 mm wide, entire or pinnatisect 4. Leaves ≤ 1.5 mm wide, filiform, linear to pinnatisect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . dissectifolius 4: Leaves > 1.5 mm wide 5. Leaves ≤ 23 mm long, prostrate growth habit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . HEADLAND V. 5: Leaves > 23 mm long 6. Leaves broad lobed, pinnatisect to bipinnatisect with rounded ends to branchlets, ≥ 6 dissections, ≥ 2.5 mm wide, rooting from stem nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . alpinus 6: Leaves narrow linear, entire or pinnatisect, < 6 dissections 7. Leaves succulent (≥ 0.5 mm thick), obovate, rarely pinnatisect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DUNE V. 7: Leaves membranous (< 0.5 mm thick), linear to lanceolate, pinnatisect 8. Teeth on leaf margins ≥ 10 in total . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . RANGE V. 8: Teeth on leaf margins < 10 in total . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . W. NSW V.


Stace 1980; Mackay and Morrison 1989). Genetic divergence based on isozyme evidence, though limited for this variant, lends support for recognition as separate subspecies. Species would be an inappropriate taxon for recognition of these two variants because of the intergrading nature of most of the variation between them (Du Rietz 1930; Stace 1980). Variants within each of these subspecies (the BRIGALOW/DESERT subspecies and another subspecies incorporating all other variants) are more problematic because variation was continuous. Although any imposed classification of these variants is somewhat arbitrary, some form of taxonomic recognition for habitat-based groups with some morphological differences may be appropriate in such a widespread and common species (e.g. Morrison and Rupp 1995). The necessity of using statistical survey (taking several measurements) for separation of variants into groups (Du Rietz 1930), the low precision of morphological measurements to define them and the implicit identification of variants with particular habitat types (as with localities, Du Rietz 1930) is consistent with the taxonomic definition of variety (Du Rietz 1930; Stace 1980). Partial segregation by leaf attributes would appear to justify recognition of BRIGALOW and DESERT variants as separate varieties within the central Australian subspecies. Morphological separation of the other variants (see key) within the south-eastern Australian subspecies would also seem to justify recognition as separate varieties. Genetic separation of dissectifolius and lanceolatus on the basis of isozyme data, particularly lends support for taxonomic recognition of these variants. Caveats to taxonomic treatment In such a variable and widespread complex, difficulties are likely to arise with any taxonomic treatment. As we have shown, variation does not separate neatly into discrete and clearly defined groups, but rather is continuous across the boundaries between habitat-linked variants. Along with this variation there appears to be a whole range of additional and almost randomly distributed variation in morphology. An example of this is the expression of a lobed leaf form at various HEADLAND and DUNE localities along the NSW coast (unpubl. data). Another example of extraneous variation is the expression of glabrous v. densely hairy achenes in single populations or within single variants (e.g. lanceolatus and alpinus). The latter was also noted by Ali (1969) in S. pinnatifolius taxa. It is likely, therefore, that there will always be some specimens that fall outside proposed taxa using keys provided. Proposed taxonomic classifications of the S. pinnatifolius complex must be judged on whether the keys they engender enable us to identify the majority of specimens.


47

The level of uncertainty within this treatment of S. pinnatifolius is greatest in central and western areas of Australia. It is clear that detailed studies on morphological variation in live plants from those areas are necessary to supplement our knowledge of this very complex group. Several possible approaches could be used to further elucidate taxonomy and variation in S. pinnatifolius. Detailed analysis of branching patterns in divided leaves, analysis of differences between upper and lower leaves and stem v. branch leaves, and assessment of whether leaves are widest proximally, mid-leaf or distally, have been suggested as promoting a full understanding of variation within S. pinnatifolius (I. Thompson, pers. comm.). Numerical analysis of the frequency of achene types within capitula (ray v. disc florets), individuals, populations, may allow further understanding relationships among variants. Finally, investigation of potential names for proposed subspecies and varieties (as per Ali 1969; Belcher 1994, 1996) is necessary before they can be correctly applied to members of the complex. Acknowledgments We thank the Dairy Research and Development Corporation for funding this research, Carolyn Porter and the National Botanic Gardens, Sydney for scanning electron microscopy, the National Parks and Wildlife Service of NSW for allowing achene collections in protected areas, Kathryn Radford, Pip Walsh, Anthony Whalen and Leon Scott for field assistance, Dennis Dwarte for setting up leaf image analysis, Gary Chapple for compiling herbarium database data from around Australia, Adrienne Kirby for statistical and multivariate advise, and Brett Abbott for help with preparing figures. Comments by Ian Thompson on a draft of this paper have improved the final version. References Ali SI (1964a) Senecio lautus complex in Australia. I. Taxonomic considerations and discussion of some of the related taxa from New Zealand. Australian Journal of Botany 12, 282–291. Ali SI (1964b) Senecio lautus complex in Australia. II. Cultural studies of populations. Australian Journal of Botany 12, 292–316. Ali SI (1966) Senecio lautus complex in Australia. III. The genetic system. Australian Journal of Botany 14, 317–327. Ali SI (1968) Senecio lautus complex in Australia. IV. The biology of the complex. Phyton (Horn, Austria) 13, 53–62. Ali SI (1969) Senecio lautus complex in Australia. V. Taxonomic interpretations. Australian Journal of Botany 17, 161–176. Belcher RO (1993) The ‘Senecio aff. lautus’ complex (Asteraceae) in Australia. I. Criteria for exclusion of lautusoid Senecio of Australia from S. lautus sensu stricto of New Zealand. Australian Systematic Botany 6, 359–363. Belcher RO (1994) The ‘Senecio aff. lautus’ complex (Asteraceae) in Australia. II. Clarification of names given to pseudolautusoid Australian specimens of Senecio by Richard and by Candolle. Australian Systematic Botany 7, 71–85. Belcher RO (1996) Australian alpine scapose radiate taxa of Senecio (Asteraceae). Muelleria 9, 115–131.

48



Clausen J, Keck DD, Hiesey WM (1939) The concept of species based on experiment. American Journal of Botany 26, 103–106. Du Rietz GE (1930) The fundamental units of biological taxonomy. Svensk Botanisk Tidskrift 24, 333–428. Harden GJ (1992) ‘Flora of New South Wales, volume 3.’ (New South Wales University Press: Kensington, NSW) Kephart SR (1990) Starch gel electrophoresis of plant isozymes: a comparative analysis of techniques. American Journal of Botany 77, 693–712. Kincaid DT, Schneider RB (1983) Quantification of leaf shape with a microcomputer and Fourier transformation. Canadian Journal of Botany 61, 2333–2342. Lawrence ME (1985) Senecio L. (Asteraceae) in Australia: reproductive biology of a genus found primarily in unstable environments. Australian Journal of Botany 33, 197–208. Lawrence ME, Belcher RO (1986) Senecio. In ‘Flora of South Australia. Part III. 4th edn’. (Eds JP Jessop, HR Toelken) pp. 1591–1605. (South Australian Government Printer: Adelaide, SA) Mackay D, Morrison DA (1989) Multivariate morphometric and allozymic analysis of the Conospermum taxifolium (Proteaceae) species complex. Plant Systematics and Evolution 166, 141–158. Marohasy JJ (1993) Are we justified in considering fireweed (Senecio madagascariensis) an exotic? In ‘Proceedings II. 10th Australian and 14th Asian-Pacific Weed Conference’. Brisbane. (Eds JT Swarbrick, CWL Henderson, RJ Jettner, L Streit, SR Walker) pp. 122–127. (Weed Society of Queensland Inc.: Brisbane, Qld) Michael PW (1981) Alien plants. In ‘Australian vegetation’. (Ed. RH Groves) pp. 44–46. (Cambridge University Press: Cambridge, UK) Michael PW (1992) Senecio lautus sensu lato—an unresolved complex. Australian Systematic Botanical Society Newsletter 72, 10–11. Moran GF, Hopper SD (1987) Conservation of the genetic resources of rare and widespread eucalypts in remnant vegetation. In ‘Nature conservation: the role of remnants of native vegetation’. (Eds DA Saunders, GW Arnold, AA Burbidge, AJM Hopkins) pp. 151–162. (Surrey Beatty and Sons Pty Ltd: NSW) Morrison DA, Rupp AJ (1995) Patterns of morphological variation with Acacia suaveolens (Mimosaceae). Australian Systematic Botany 8, 1013–1027. Nei M (1978) Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89, 583–590. Noble JW, Crossley J de B, Hill BD, Pierce RJ, McKenzie RA, Debritz M, Morely AA (1994) Pyrrolizidine alkaloidosis of cattle associated with Senecio lautus. Australian Veterinary Journal 71, 196–200. Ornduff R (1960) An interpretation of the Senecio lautus complex in New Zealand. Transactions of the Royal Society of New Zealand 88, 63–77. Ornduff R (1964) Evolutionary pathways of the Senecio lautus alliance in New Zealand and Australia. Evolution 18, 349–360.

Pasteur N, Pasteur G, Bonhomme F, Catalan J, Britton-Davidian J (1988) ‘Practical isozyme genetics.’ (Translator: M. Cobb) (Ellis Horwood Limited: Chichester, UK) Radford IJ (1997) Impact assessment for the biological control of Senecio madagascariensis Poir. (fireweed). PhD Thesis, The University of Sydney, Australia. Radford IJ, Liu Q, Michael PW (1995) Chromosome counts for the Australian weed known as Senecio madagascariensis (Asteraceae). Australian Systematic Botany 8, 1029–1033. Radford IJ, Muller P, Fiffer S, Michael PW (2000) Genetic relationships between Australian fireweed and South African and Madagascan populations of Senecio madagascariensis Poir. and closely related Senecio spp. Australian Systematic Botany 13, 409–423. doi:10.1071/SB98029 Scott LJ, Congdon BC, Playford J (1998) Molecular evidence that fireweed is of South African Origin. Plant Systematics and Evolution 213, 251–257. Sindel BM, Radford IJ, Holtkamp RH, Michael PW (1998) The biology of Australian weeds. 33. Senecio madagascariensis Poir. Plant Protection Quarterly 13, 2–15. Soltis DE, Haufler CH, Darrow DC, Gastony GJ (1983) Starch gel electrophoresis of ferns: a compilation of grinding buffers, gel and electrode buffers, and staining schedules. American Fern Journal 73, 9–27. Stace CA (1980) ‘Plant taxonomy and biosystematics.’ (Edward Arnold: London) Swofford DL, Selander RB (1981) BIOSYS-1: a FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. The Journal of Heredity 72, 281–283. Swofford DL, Selander R (1989) ‘BIOSYS-1, a computer program for the analysis of allelic variation in population genetic and biochemical systematics. Release 1.7. David Swofford Natural History Survey.’ (University of Illinios: Champaign, IL) West JG, Noble IR (1984) Analysis of digitized leaf images of Dodenea viscose complex in Australia. Taxon 33, 595–613. White RJ, Prentice HC, Verwijst T (1988) Automated image acquisition and morphometric description. Canadian Journal of Botany 66, 450–459.

Manuscript received 23 May 2003, accepted 14 November 2003

http://www.publish.csiro.au/journals/asb