Variation in Leaf Morphology and Agronomic Attributes of a Naturalized Population of Medicago polymorpha L. (Burr Medic) from New South Wales, Australia, and Relationships with Climate and Soil Characteristics

Abstract

As one component of a study to improve Medicago spp. germplasm in eastern Australia, fifteen phenotypic and agronomic attributes were recorded for 4715 plants grown from the seed of 90 accessions of the widely naturalized pasture legume Medicago polymorpha from 90 sites in eight regions of inland New South Wales. The species expressed wide polymorphism. However, many leaflet attributes were associated with specific climate and soil characteristics, which varied from east to west across the collection zone. Discriminant analysis showed that accessions from the four most northern (summer dominant rainfall) and western (arid–semiarid) regions (Group A) differed from accessions from the most southern, temperate (winter dominant rainfall) and eastern (higher rainfall) regions (Group B). Group A flowered earlier and had shorter pod spines. Group B had lower plant vigor. Regions from which Group A accessions were collected had higher soil pH, lower winter rainfall, and higher minimum winter temperature than Group B regions. The diversity in the population, particularly the difference in flowering times among accessions collected from drier, warmer regions and those from more mesic, cooler regions, and the wide variation in flowering time measured among plants grown from accessions within all collection regions, is likely to ensure the long-term persistence of M. polymorpha in a changing climate. Elite lines were subsequently identified and lodged in National and International Genebanks for future research.

1. Introduction

Burr medic (Medicago polymorpha L.) is an annual self-regenerating legume native to countries in the Mediterranean basin and central Asia [1] which has naturalized throughout parts of Australia [2,3,4], and in countries with a Mediterranean-type climate, especially Chile, where the population has been studied extensively [5,6], and in parts of North America [7].

It was accidentally introduced into Australia, most likely through the importation of sheep and other animals from Mediterranean regions by early British and other settlers in the late 18th and throughout the 19th centuries. Its distribution was facilitated by spiny pods clinging to the coats of animals. Burr medic is widely naturalized in New South Wales and Queensland [2,8,9,10,11,12]. Its persistence is due to a number of factors, with a key underpinning factor being its ability to produce a high proportion of hard seed at seed set [13,14] that softens gradually over time, thus ensuring staggered germination and long-term survival.

On the western plains of New South Wales (~455,000 km²), M. polymorpha is found in a wide variety of vegetation communities ranging from open grasslands to shrublands and woodlands. It grows on a wide range of soils from sandy soils to clays but does best on alkaline soils, with the most prolific growth on heavy clays. It is an extremely valuable autumn–winter–spring growing legume providing palatable, nutritious fodder for grazing livestock [12]. It also occurs sporadically in the remainder of the state at elevations < 700 m [15,16], occurring on uncultivated land, in gardens and playing fields, and on roadsides and waste areas.

Annual medics have played a significant role in pastoral and crop-pasture systems in eastern Australia [17,18,19], with accessions from more than 20 species released for commercial use following introduction through the Commonwealth Plant Introduction Service that commenced in 1912 [2] which, subsequently, was enhanced by State Departments of Agriculture, particularly in New South Wales and South Australia [20,21]. Three cultivars of spineless burr medic, M. polymorpha var. brevispina, cvv. Serena [22], Circle Valley [23], and Santiago [24], were released for commercial use based on germplasm introduced into Australia prior to this study.

As part of a National Annual Medic Improvement Program established in the 1990s to improve annual Medicago spp. germplasm for commercial use Australia-wide (P.S. Cocks, pers. comm.), this study addressed the potential, in eastern Australia, for the naturalized population to provide germplasm able to contribute to new cultivars to improve medic productivity and adaptation, particularly in semi-arid environments. In this New South Wales component of the study, burr medic was collected from the temperate to summer rainfall dominant semi-arid environments of the inland, western plains of the State [25] to begin that investigation. In these low to medium rainfall areas, it is naturalized on neutral to alkaline clay soils and contributes significantly to livestock production systems, thus providing an opportunity to identify elite germplasm to incorporate into those systems. Burr medic is also naturalized to the east of the area chosen for study, but where Trifolium spp. (clovers) and Medicago sativa (lucerne/alfalfa) are the dominant and most productive pasture legumes contributing to livestock production systems.

Comparable projects in Sardinia aimed to identify elite germplasm for the release of new cultivars [26,27] and defined, for example, the separate occurrence of the M. polymorpha varieties vulgaris (no longer separated from var. polymorpha) and var. polymorpha, partially based on the longer pod spine length of the latter, and occurring at different altitudes [27]. Similar studies in Chile have shown that accessions from arid environments flowered earlier than those from mesic environments based on differing photoperiod requirements [6,28], and that phenological and agronomic attributes such as winter vigor and above-ground biomass production were also environmentally differentiated [5,6].

This paper describes our study, conducted to quantify the phenotypic variation in burr medic germplasm growing across inland regions in New South Wales where the species is naturalized, and define their relationships with climate and soil. Our hypothesis was that the observed, significant phenotypic variation across regions could be associated with those site characteristics, thus providing information to identify elite germplasm for subsequent cultivar development. It was conducted at the same time as, and in collaboration with, a study carried out to the immediate north in subtropical Queensland [11].

2. Materials and Methods

The study began with the collection of pods of naturalized germplasm of M. polymorpha from 90 sites in eight regions of inland New South Wales, followed by the characterization of leaf morphological and plant agronomic attributes of 4715 plants grown at a single location.

2.1. Germplasm Collection

Pods of naturalized burr medic were collected in bulk from 90 sites in eight regions across extensive east–west and north–south transects in inland New South Wales (Figure 1) in the summer of 1993/94. The regions chosen experience both the arid and mesic extremes of climate in which burr medic is found and were arbitrarily designated according to merged local government areas. Climate types of the collection sites based on the Köppen–Geiger classification [29] were mainly semi-arid (BSh/k) and warm temperate (Cfa) environments, with winter dominant rainfall to the south tending to summer dominant rainfall to the north, and with higher rainfall to the east and lower rainfall to the west. The climate types at each collection site are listed in Table S3. These collection sites were located (with median values in parenthesis) between latitudes 28.73–35.12° S (31.45° S) and longitudes 141.62–151.12° E (147.7° E) at altitudes between 61 and 627 m (167 m above sea level defined by the Australian Height Datum (AHD) [30]).

Pods produced in 1993 were collected from the soil surface in a single sampling at sites which, mainly, had never been cultivated, or had been cropped and subsequently returned to naturalized or sown pasture. Provenance data, including altitude, latitude, longitude, land use, soil type, vegetation, and slope were recorded at each of the 90 collection sites (Table S1). Soil samples (0–10 cm) were taken from 85 of the collection sites and analyzed for the following: clay content (%), pH (1:5 in CaCl₂ solution), electrical conductivity (mS cm⁻¹ in 1:5 soil suspension), organic carbon (C, %), total nitrogen (N, %), available phosphorus (P, Colwell bicarbonate extraction, μg g⁻¹), exchangeable cations: calcium (Ca), magnesium (Mg), sodium (Na), potassium (K), and cation exchange capacity (CEC) (meq 100 g⁻¹).

Average climate data were obtained from recording stations nearest to the collection sites using Australian Government Bureau of Meteorology data [31] and the Rainman program [32]. These data included mean annual rainfall (mm), growing season (April–September) rainfall (mm), mean daily maximum winter temperature (°C), and mean daily minimum winter temperature (°C).

2.2. Characterization of Germplasm

The accessions were characterized in 1994 at Condobolin, New South Wales (33.09° S, 147.15° E, elevation 220 m AHD, 424 mm average annual rainfall). The site was located on a Red Chromosol soil [33] of slightly acid pH and low fertility.

During preparation of the site, 130 kg ha⁻¹ single superphosphate (9% P, 10% S) with 2% molybdenum was applied with a fertilizer drill at ~50 mm depth. Glyphosate herbicide (450 g L⁻¹ a.i.) was applied at 1.2 L ha⁻¹ about two months before sowing for the control of all weeds. Fluazifop-P (128 g L⁻¹ a.i.) at 0.5 L ha⁻¹ was applied twice to selectively control grass weeds; the first application following sowing and the second following the establishment of the medic plants. Endosulfan insecticide (350 g L⁻¹ a.i.) was applied at 0.5 L ha⁻¹ after sowing but before the emergence of the sown burr medic, and again at 1.5 L ha⁻¹ 8 weeks after establishment, to control red-legged earth mite (Halotydeus destructor), which was endemic to the location.

Seed pods of the 90 accessions were threshed and the seed abraded to increase their germinability. That seed, together with seed of both cvv. Serena and Santiago sown in 5 randomly located plots as standards, was sown into plots consisting of 4 rows, each 2 m long and 1 m apart, on 16 May 1994. Seeds were sown at a depth of 10 mm and inoculated by watering with a slurry of Rhizobium WSM40. Growing season rainfall was low, with 51 mm received between April and September. Therefore, the site was irrigated five times during the growth period, firstly in May (47 mm) to promote establishment, and in June (25 mm), July (27 mm), August (38 mm), and September (25 mm) to ensure that the plants had the opportunity to express their potential growth.

Seedling emergence began 9 days after sowing (DAS) on 25 May and plants were thinned manually after 42 days to create a spaced population of about 50 plants per accession, resulting in a total population of 4715 plants representing the collection (comprising 217 plants from Far West region; 954 Riverina-MIA; 708 Central West Slopes & Plains; 1195 North West Plains Coonamble; 508 North West Plains Brewarrina; 321 North West Plains Walgett; 384 North West Plains Moree; 428 Liverpool Plains), together with 421 commercial cultivar check plants (220 of cv. Santiago; 201 of cv. Serena). Plants were assessed for leaflet morphological attributes using the International Board for Plant Genetic Resources (IBPGR) conventions [34] at 42 and 98 DAS. Plant vigor was rated at 107 DAS, days to first flower were recorded, and pod attributes (spine length and number of coils) assessed after pod harvest (

Table 1. Attributes recorded of individual M. polymorpha plants grown at Condobolin.

Attribute Designation	Description
Leaflet attributes * recorded as LEAF42 and LEAF98, at 42 and 98 days after sowing (DAS)
1a	No mark on leaflets. Leaflet margins slightly serrate.
1b	Leaflets with faint to obvious brown-purple proximal midrib coloration 0.5–2 mm long. Leaflet margins slightly serrate.
1c	Leaflets with two distinct proximal small squarish smudges either side of the mid-rib, sometimes merging. Leaflet margins slightly serrate.
1d	Leaflets with proximal dark inverted Y-like marks. Leaflet margins slightly serrate.
2	No mark on leaflets. Leaflet margins distinctly dentate (usually occurring with 1a leaflets on the one plant).
3	Distinct proximal blotch with dark band across the base of leaflets with lighter color near leaflet bases. Leaflets usually grey. Leaflet margins slightly serrate.
4	Leaflets a larger darker form of 3.
Additional LEAF attributes assessed 98 DAS
1bd	1b types with larger marks.
GL	Grey leaflets.
WO	Light grey leaflets.
Other plant attributes
VIGOR	Plant vigor rated visually, 1 (minimum) to 6 (maximum), 107 DAS.
FLOWERDAY	DAS to appearance of first flower.
POD COILS	Number of coils on mature pods.
POD SPINES	Rating of mature pod spine length: (1) <1 mm; (2) 1–2 mm; (3) 2–3 mm; (4) >3 mm.

* The default leaflet color for attributes other than GL and WO is green

). Pod coil and spine data were not recorded for the commercial cvv. standards, Santiago and Serena. The data matrix of plants and recorded attributes is given in Table S2.

2.3. Statistical Analysis

All statistical analyses were conducted in Genstat, 23rd edition [35]. Plant agronomic attributes were analyzed by analysis of variance (ANOVA) for unbalanced designs with treatments being accessions nested in regions. Comparisons were made where the F test was significant by l.s.d. designated Fl.s.d. at p < 0.05 and by the Bonferroni multiple range test. Correlation analyses of the mean plant attributes of the M. polymorpha accessions with the climate and soil properties of their respective collection sites were conducted. Linear discriminant analysis of regions based on plant attributes of the collection was conducted for 3809 plants, the number with complete attribute data, using the data matrix given in Table S2 and excluding the two cultivars. Correlation analyses of the discriminant scores for regions with site characteristics of climate and soil were conducted. A hierarchical cluster analysis, based on the inter-regional distance matrix (Mahalanobis D-squared) from the discriminant analysis, and using average link as the sorting strategy, was produced and displayed as a dendrogram.

3. Results

Climate and soil data from all collection sites are given in Table S3.

The sites received total annual rainfall (range with median in parenthesis) from 225 to 660 mm (486 mm) and April–September (growing season syn. winter) rainfall from 100 to 275 mm (202 mm). They encountered average maximum winter temperatures of 17.8–22.8 °C (20.2 °C) and minimum winter temperatures of 4.4–8.0 °C (6.7 °C).

The soil characteristic measures (range with median in parenthesis) were clay content 25–60% (38%), pH 5.72–8.08 (7.02), electrical conductivity 0.06–0.41 mS cm⁻¹ (0.14 mS cm⁻¹), organic C 0.35–2.48% (0.95%), total N 0.037–0.290% (0.109%), extractable P 10.5–144.6 μg g⁻¹ (26.2 μg g⁻¹), exchangeable cations Ca 5.1–37.0 meq 100 g⁻¹ (14 meq 100 g⁻¹), Mg 2–32 meq 100 g⁻¹ (6.2 meq 100 g⁻¹), Na 0.07–7.54 meq 100 g⁻¹ (0.88 meq 100 g⁻¹), K 0.16–4.10 meq 100 g⁻¹ (1.27 meq 100 g⁻¹), and CEC 10–80 meq 100 g⁻¹ (28 meq 100 g⁻¹). Soils on which burr medic accessions had naturalized varied from brown and red-brown loams to brown clay loams, to clay soils, all of neutral to alkaline pH with generally moderate to occasionally high fertility. Twenty-four of the 90 soils from which accessions were collected were sodic (% exchangeable sodium > 7, derived from Table S3), particularly in the Riverina MIA region, where 63% of collection site soils were sodic.

3.1. Plant Attributes—Days to Flower, Plant Vigor, and Pod Types

3.1.1. Days to Flower

There were highly significant differences among accessions and regions in the number of days to flower (FLOWERDAY) (p < 0.001, Table S4). The distribution of values for days to flower of plants within regions and cultivars, with mean and median values, are shown in Figure 2.

Among regions, there were distinct differences in the times to flower, with plants derived from the four most western and north-western regions (Far West and North West Plains Brewarrina, Walgett, and Moree) flowering significantly earlier (mean values 91–94 DAS) than those collected from the Riverina-MIA, the Central West Slopes & Plains and the North West Plains Coonamble (100–101 DAS). Plants from the Liverpool Plains were the latest to flower (106 DAS). Notably, the commercial cultivars Santiago (84 days) and Serena (70 days) flowered, on average, substantially earlier than naturalized accessions from the regions (p < 0.05, Figure 2).

In addition, there was a wide variation within regions among the times to flower of individual plants grown from the accessions. The widest ranges occurred in the three North West Plains regions of Brewarrina, Walgett, and Moree (up to 52 days), with narrower ranges in regions to the south (down to 30 days). The earliest flowering individual plants were from those northern regions, and the latest flowering individual was from the more mesic Liverpool Plains. The commercial cultivars Serena (64–114 DTF) and Santiago (77–109 DTF) also expressed a range of intra-cultivar flowering times. Summary statistics including mean, median, minimum, maximum, range, and standard error of days to flower of plants from the various regions are given in Table S5.

3.1.2. Plant Vigor, Pod Coils, and Pod Spines

There were highly significant differences among regions and accessions for plant vigor assessed at 107 DAS (VIGOR) when plants were in a late vegetative phase of growth (p < 0.001, Table S6). There were also highly significant differences among regions and accessions for the number of coils per pod (POD COILS) (p < 0.001, Table S7) and length of pod spines (POD SPINES) (p < 0.001, Table S8). The mean values of these plant attributes for regions are shown in

Table 2. Regional mean values with standard errors (SEM) from ANOVA for plant vigor (and including values for cvv. Santiago and Serena), number of pod coils, and pod spininess of Medicago polymorpha plants. Different letters indicate significant differences among means by the post hoc Bonferroni multiple range test. No data for pod attributes were obtained for the spineless commercial cvv. Santiago and Serena.

Region	Plant Vigor		Pod Coils		Pod Spines
	Mean	SEM	Mean	SEM	Mean	SEM
1 Far West	3.94 d	0.036	3.49 bc	0.013	2.24 a	0.015
2 Riverina MIA	3.39 a	0.017	3.52 c	0.006	2.85 d	0.007
3 Central West Slopes & Plains	3.57 c	0.020	3.39 a	0.007	2.82 c	0.009
4 North West Plains Coonamble	3.53 c	0.015	3.56 d	0.006	2.86 d	0.007
5 North West Plains Brewarrina	3.93 d	0.023	3.48 b	0.009	2.65 b	0.010
6 North West Plains Walgett	3.84 d	0.029	3.88 f	0.011	2.79 c	0.013
7 North West Plains Moree	3.52 bc	0.027	3.63 e	0.010	2.84 cd	0.012
8 Liverpool Plains	3.41 ab	0.025	3.57 d	0.009	3.01 e	0.011
9 cv. SANTIAGO	4.26 e	0.032	-	-	-	-
10 cv. SERENA	4.23 e	0.033	-	-	-	-

.

Among regions, there were significant differences in the mean values for the vigor of mature plants. The commercial cvv. Santiago and Serena had the highest mean vigor scores (p < 0.05), followed by plants from the most western (Far West) and north-western regions (North West Plains Brewarrina and Walgett). Plants from the Riverina MIA and Liverpool Plains had the least vigor (Table 2). However, there was wide variation in the vigor of all plants grown from accessions across all regions, with a small number of plants from multiple regions expressing greater vigor than the mean and median values for the commercial cultivars (Table S2).

The mean number of pod coils of accessions in the different regions ranged mainly between 3 and 4 coils per pod (Table 2). Accessions from the Central Western Slopes & Plains region had the smallest number of coils (p < 0.001). Accessions collected from the North West Plains Walgett had the greatest number of coils (p < 0.001), and also the smallest variation among accessions. The greatest range in number of pod coils of plants grown from accessions was measured in those collected from the Riverina MIA region.

None of the plants grown from the accessions collected were spineless. There was considerable variation in length of plant spines within most regions. Plants with the shortest spines on average were from the Far West region, with the longest from the Liverpool Plains (Table 2).

3.2. Correlation Analysis of Plant Attributes of Accessions with Site Characteristics

The mean plant attributes of accessions are given in Table S9, and their correlation coefficients with collection site characters of climate (n = 90) and soil (n = 85) are given in Table S10. Statistically significant correlations are summarized in

Table 3. Site characters significantly correlated with mean plant attributes of accessions. Lat = Latitude (°S); Long = Longitude (°E); Alt = Altitude (m); WR= Winter rainfall (mm); AAR = Average annual rainfall (mm); Min T = Minimum winter temperature (°C); Max T = Maximum winter temperature (°C); Clay = soil clay (%); pH = soil pH (CaCl₂); EC = soil solution electrical conductivity (mS cm⁻¹); C = soil organic carbon (%); N = total soil nitrogen (%); P = soil phosphorus (Colwell, µg g⁻¹); Ca = soil calcium (meq 100 g⁻¹); Mg = soil magnesium (meq 100 g⁻¹); Na = soil sodium (meq 100 g⁻¹); CEC = soil cation exchange capacity (meq 100 g⁻¹). Asterisks indicate statistical probability of the correlation coefficient with *** < 0.001, ** < 0.01, and * < 0.05. Descriptions of the plant attributes are given in Table 1.

Plant Attribute	Site Characters Significantly Correlated with Plant Attributes
LEAF42_1a	Lat 0.412 ***	Min T −0.332 **	Na 0.295 **	Ca −0.292 **
LEAF42_1c	Lat 0.250 *
LEAF42_1d	Ca 0.394 ***
LEAF42_2	Ca 0.326 **	Long 0.321 **	WR 0.307 **	Lat −0.272 *	Mg 0.252 *	Min T 0.244 *
LEAF42_3_4	AAR −0.284 *
LEAF98_1a	Max T 0.295 **	Lat −0.294 **	EC 0.237 *
LEAF98_1c	Lat 0.246 *	pH −0.253 *	Clay 0.230 *
LEAF98_1bd	Alt 0.429 ***	CEC 0.305 **	AAR 0.301 **	Max T −0.334 **	C 0.290 *	WR 0.282 *	Long 0.235 *	N 0.228 *
LEAF98_2	Ca 0.315 **	WR 0.279 *	Long 0.277 *	AAR 0.245 *
LEAF98_3_4	AAR −0.237 *
LEAF98_GL	Alt 0.281 *
LEAF98_WO	AAR −0.254 *	C −0.244 *	Na 0.231 *
FLOWERDAY	Max T −0.512 ***	AAR 0.462 ***	Alt 0.394 ***	C 0.377 ***	Min T −0.367 **	N 0.364 **	Lat 0.350 **	pH −0.315 **	WR 0.295 *	Long 0.216 *
POD COILS	Ca 0.346 **
POD SPINES	AAR 0.273 *	WR 0.239 *	Max T −0.231 *

.

Most leaflet attributes, determined when plants were in both juvenile (42 DAS) and adult (98 DAS) stages of growth, were correlated with one or more collection site characteristics:

• Leaflets with no marks and slightly serrate margins on both juvenile (attribute LEAF42_1a) and adult (LEAF98_1a) plants were correlated with latitude (juvenile plants positively and adult plants negatively) and negatively correlated with minimum temperature, thus indicating a strong occurrence as adult (but not juvenile) plants among accessions collected from warmer, northern locations. This leaflet attribute was also correlated positively with soil exchangeable sodium (juvenile plants) and soil electrical conductivity (adult plants), but negatively with calcium (juvenile plants), suggesting naturalization on some sodic and slightly saline soils.
• Leaflets without marks but with distinctly dentate margins were positively correlated with winter rainfall, longitude, and soil exchangeable calcium as both juvenile (LEAF42_2) and adult plants (LEAF98_2). In juvenile plants, this attribute was also negatively correlated with latitude, and positively with minimum growing season temperature and soil exchangeable magnesium. In adult plants, this attribute was also positively correlated with annual rainfall. Plants with this attribute thus occurred more frequently in accessions from more mesic areas in the east of the collection zone.
• Leaflets with faint to obvious brown-purple proximal midrib coloration with markings more than 2 mm long with mainly serrate margins on adult plants (LEAF98_1bd) were positively correlated with altitude, longitude, and winter rainfall. These leaflet attributes were positively correlated with cation exchange capacity, soil organic C and total N. They were negatively correlated with maximum growing season temperature. Thus, plants with these markings occur mainly in fertile, mesic locations at higher altitude with lower growing season maximum temperatures.
• Leaflets with two proximal squarish smudges on either side of the midrib and slightly serrate margins on both juvenile (LEAF42_1c) and adult plants (LEAF98_1c), were positively correlated with longitude and, as adult plants, were also correlated positively with soil clay content and negatively with soil pH, indicating a strong occurrence in eastern collection regions on soils of higher clay content but with lower pH.
• Leaflets with a dark proximal inverted Y mark with slightly serrate margins as juvenile plants only (LEAF42_1d) were positively correlated with soil exchangeable calcium and were widely distributed across the collection zone.
• Leaflets with a dark or light proximal blotch with a dark band across the lower leaf on both juvenile (LEAF42_3_4) and adult (LEAF98_3_4) plants, which were mainly grey in color and with slightly serrate margins, were negatively correlated with annual rainfall, thus tending to occur in drier environments.
• Adult plants with grey colored leaflets (LEAF98 GL) were positively correlated with altitude, and those with lighter grey leaflets (LEAF98 WO), were positively correlated with soil exchangeable sodium, and negatively with annual average rainfall and soil organic C. This implies a separation in which plants with grey leaflets were strongly represented in higher altitude locations such as the Liverpool Plains, while those with light grey leaflets occurred on more sodic soils of lower fertility in drier environments.

Plant vigor (VIGOR) was not correlated with any site attributes. However, days to flower (FLOWERDAY) was positively correlated with the environmental characteristics latitude, longitude, altitude, and annual and winter rainfall, and the soil characteristics soil N and soil organic C. This attribute was also negatively correlated with growing season maximum and minimum temperatures and soil pH. Thus, accessions from southern, eastern, and wetter locations, at higher altitudes, and on soils of higher organic matter, tended to flower later. Accessions from warmer locations that receive lower annual and winter rainfall and on soils of higher pH tended to flower earlier.

Accessions collected from soils with higher exchangeable calcium tended to produce pods with more coils. Accessions from wetter (both higher annual and winter rainfall), cooler environments produced longer spines than those from warmer environments.

3.3. Discriminant Analysis

The latent roots and trace from the discriminant analysis of regions based on plant attributes are given in Table S11. The percentage variation accounted for by the latent roots is shown as a scree plot in Figure 3. The first two latent roots accounted for 42.8 and 26.6% of the variation, respectively, for a total of 69.4% of the total variation of the population.

The mean discriminant scores for the eight regions based on plant attributes are given for the seven significant axes in Table S12, with inter-regional distances shown in Table S13. The distribution of scores for plants along discriminant axis 1 grouped by geographical region is shown in Figure 4. The distribution of the regional means of discriminant scores on the plane of discriminant axes 1 and 2 is shown in Figure 5.

The overlapping extremes in the distribution of burr medic plants both within and between regions highlight a wide polymorphism in the phenotypic attributes of the naturalized population (Figure 4). A separation into two regional groups is evident when discriminant scores 1 and 2 are considered (Figure 5). The first group, Group A, has DF1 scores > 0.5 and includes accessions from the summer dominant rainfall regions. These include the North West Plains Brewarrina and Walgett regions, both having a hot, semi-arid climate (Köppen–Geiger: BSh), the North West Plains Moree region having a semi-arid-influenced humid subtropical climate (Cfa), and the arid Far West region with a winter cold/summer hot, semi-arid climate (BSk/h). The second group, Group B, had DF1 scores < 0.5 and includes accessions collected from the warm temperate (Cfa/BSk) regions, the Liverpool Plains, Riverina-MIA, North West Plains Coonamble, and Central West Slopes & Plains regions. These groupings are confirmed from the cluster analysis of regions based on inter-regional distances of Mahalanobis D-squared (Table S13), presented as a dendrogram in Figure 6.

All correlation coefficients between plant attributes and significant discriminant axes 1 to 7 of regions are given in Table S14, with major correlations with discriminant functions 1 and 2 shown in

Table 4. Major correlations (r > 0.1) between attributes of 3809 M. polymorpha plants and the first two discriminant functions from a discriminant analysis of eight regions.

Attribute	DF1	Attribute	DF2
FLOWERDAY	−0.7388	LEAF98_3_4	−0.4096
LEAF42_1a	−0.4176	LEAF42_1a	−0.3498
POD SPINES	−0.3453	LEAF98_1a	−0.3421
LEAF98_1a	−0.2624	LEAF42_3_4	−0.3080
LEAF98_GL	−0.1388	VIGOR	−0.2682
LEAF42_1c	−0.1160	LEAF98_WO	−0.1457
LEAF98_WO	0.1826	LEAF42_1c	0.1004
POD COILS	0.1845	LEAF42_1d	0.1052
LEAF98_3_4	0.2715	FLOWERDAY	0.1256
VIGOR	0.2973	LEAF98_1c	0.1796
LEAF42_3_4	0.3240	LEAF42_1b	0.2332
		POD COILS	0.2413
		LEAF98_1b	0.2448
		POD SPINES	0.3536
		LEAF42_2	0.4484
		LEAF98_2	0.4761

.

The strongly negative correlation of days to flower (FLOWERDAY) with DF1 confirms that plants from the more temperate regions (BSk, Cfa) (Group B) were later maturing than those from the more arid and/or summer dominant rainfall regions (BSh, Cfa) and the Far West region (BWh, BSk) (Group A). Plants from Group B regions, having no leaflet marks (LEAF42_1a; LEAF98_1a), longer pod spines (POD SPINES), and grey leaflets (LEAF98_GL), or leaflets with two distinct proximal small squarish smudges either side of the mid-rib (LEAF42_1c), differed from those from Group A regions. Those from Group A, in contrast, had leaflets with a distinct proximal blotch, with a dark band across the base of leaflets, which were usually grey (LEAF42_3_4; LEAF98_3_4), and many with washed-out light grey leaflets (WO). They also expressed greater plant growth (VIGOR) and produced pods with more coils (POD COILS) in comparison with plants from regions in Group B.

Plants negatively associated with DF2 from the Central West Slopes & Plains and Far West regions differed from those from other regions by exhibiting a distinct proximal blotch and a dark band across the base of leaflets, which were usually grey in color (LEAF98_3_4, LEAF42_3_4, LEAF98_WO) or leaflets with no marks and slightly serrate margins (LEAF42_1a, LEAF98_1a). Plants also had greater vigor (VIGOR).

Plants positively associated with DF2 from the Liverpool Plains and North West Plains Moree differed from those from other regions, particularly Central West Slopes & Plains and Far West, by exhibiting leaflets without marks and distinctly dentate leaf margins (LEAF98_2, LEAF42_2), leaflets with faint to obvious brown-purple proximal midrib color 0.5–2 mm long (LEAF98_1b, LEAF42_1b), leaflets with a proximal dark inverted Y-like mark (LEAF42_1d), or leaflets with two distinct proximal small squarish smudges either side of the mid-rib, sometimes merging (LEAF98_1c, (LEAF42_1c). Plants from these regions also tended to produce pods with longer spines and more coils (POD SPINES, POD COILS), with a tendency to flower later (FLOWERDAY).

The mean climate and soil characteristics of regions are given in Table S15. Correlation coefficients between discriminant scores based on plant attributes for the eight regions of New South Wales and mean regional site characters are given in Table S16, with statistically significant results shown in

Table 5. Major correlation coefficients (r with 7 d.f.) between the first two mean discriminant scores of eight regions of New South Wales based on the attributes of M. polymorpha accessions and mean climate and soil characteristics of regions.

Discriminant Score	Site Characteristics	r	Probability
DS1	pH	0.951	<0.001
	Winter rainfall	−0.742	0.035
	Temperature min winter	0.663	0.073
DS2	Cation exchange capacity	0.750	0.032
	Clay content	0.723	0.043
	Longitude	0.701	0.053
	Total nitrogen	0.681	0.063
	Annual rainfall	0.668	0.070
	Exchangeable magnesium	0.641	0.087

.

Regional discriminant score DS1, based on the phenotypic attributes of M. polymorpha accessions collected in each region, was positively correlated with soil pH and winter minimum temperature, and negatively with winter rainfall. These indicate that those accessions were from collection sites that have higher soil pH, lower winter rainfall, and higher winter minimum temperatures. These sites are identified with the Far West and the three more northerly North West Plains regions that constitute Group A (Figure 5 and Figure 6), in contrast with the other more southerly regions and the higher Liverpool Plains (Group B).

Regional discriminant score DS2 was positively correlated with the soil characteristics of clay content, total N, CEC, and exchangeable Mg, and average annual rainfall and longitude, identifying with regions with higher soil fertility in the east of the collection zone. These align with the occurrence of accessions that flower later and produce longer spines, thus with those from the Liverpool Plains and North West Plains Moree regions (Figure 5).

4. Discussion

The polymorphism of this naturalized collection of M. polymorpha was demonstrated by the variation in the observed plant attributes and the overlap of discriminant scores based on plant attributes across regions. Plants flowered across a wide time frame, 70 to 120 days after sowing in May, but with few flowering as early as the commercial cvv. Santiago and Serena. The vigor of mature plants, an indicator of potential plant dry matter production, varied significantly, with a small number of plants only, located widely across the collection zone, expressing the vigor of the commercial cultivars. Pod spines varied from long (>3 mm) to short (<1 mm), with no plants producing spineless pods, and the number of pod coils varied mainly between 3 and 4. In general, leaves were mainly green in color, with a small proportion that were grey or light grey. Leaf marks were not expressed as strongly as those of accessions from subtropical Queensland to the north of the collection regions of New South Wales [11]. Nevertheless, most plants expressed either blotched, smudged, or brown midrib coloration, or proximal Y shaped markings. Only a small proportion expressed no leaf mark. The phenotypic and agronomic polymorphism of this naturalized population of burr medic contrasts with the absence of variation in the morphological or phenological attributes of the naturalized cut-leaf medic (M. laciniata) populations in New South Wales [36].

The relationships among these attributes of M. polymorpha plants grown from accessions, and the regional climates and soils of their collection sites, provide insights into the naturalization of the germplasm as follows:

4.1. Time (Days) to Flower

The time to flower and, by inference, the ability to set seed, is vital to the naturalization and survival of M. polymorpha in any environment. The time to flower of naturalized M. polymorpha in New South Wales was, on average, longer in accessions from more mesic, higher altitude locations in the south and east of the collection zone. Conversely, accessions naturalized in semi-arid areas at lower latitude in the north and west of the State flowered earlier. The four regions in which accessions flowered earlier were located in an arc from the northern to the western extremities of the collection zone, namely, the North West Plains regions of Moree, Walgett, and Brewarrina, and the Far West region. No one climate characteristic contributed solely to this relationship between the regions and the flowering time of the accessions, with the Far West being the most arid region and the Northern Plains regions the warmest. The pattern of early flowering occurring in these semi-arid and summer dominant rainfall, warm temperate climates was comparable with the patterns in Chile [5,6,28] and Sicily [37], in which accessions from drier and warmer environments flowered earlier than accessions from cooler and wetter environments.

Early flowering may relate to temperature and photoperiod, which control flowering in M. polymorpha [38], and to the potential for water stress [39]. It may also relate to the occurrence of frost, as in Chilean studies [5,6,28], where plants flowered earlier in locations where the growing season minimum temperatures were higher, and the risk of frost damage to the early development of flowers was lower. Plants from the North West Plains that flowered earlier were also associated with soils of higher pH. Conversely, in Queensland [11], M. polymorpha accessions that flowered earlier were associated with lower pH soils, which occurred more sporadically across the varying subtropical climates of those collection regions.

The broad group of accessions that flowered, on average, later were located in regions further to the south, with one region (Liverpool Plains) further east and disconnected from the others. The growing seasons in these regions were cooler than in the far western and northern regions. Late flowering accessions were also generally associated with more fertile soils and in regions of higher altitude, as occur commonly in the Liverpool Plains region.

These inter-regional differences in time to flower were not extreme, as nearly all accessions flowered, on average, 7–22 days later than cv. Santiago and 21–36 days later than cv. Serena. However, and similar to the Queensland study [11], there was a wider range in the time to flower of individual plants derived from accessions within each of the collection regions (between 30 and 52 days), with the widest variation between accessions in the northern (earlier flowering) regions.

Both the variation in time to flower between regions and between plants from within any region are therefore likely to be important contributors to the naturalization of M. polymorpha to the semi-arid, warm temperate, and summer dominant rainfall environments in New South Wales. Variation was also displayed within the commercial cvv. Santiago (77–107 days) and Serena (64–114 days).

4.2. Plant Vigor

With the exception of a small number of plants, the average vigor (a surrogate for dry matter production) of the naturalized accessions in all regions was lower than for the commercial cvv. Serena and Santiago. There was a trend towards greater vigor in accessions with grey, blotched leaves or with lighter grey leaves from drier and warmer regions. This was comparable with findings in Chile that associated high plant vigor in winter in plants from areas with a shorter growing season [5,6] and provides an opportunity for the selection of more productive germplasm.

4.3. Phenotypic Attributes—Leaflet and Pod Attributes

Leaflet attributes have not been a prime focus of studies on populations of M. polymorpha, other than that conducted in subtropical Queensland, Australia [11]. In this study in New South Wales, some leaflet attributes were associated broadly with specific climate and soil characteristics in both more mesic, eastern locations, and others associated with drier, western locations in the collection zone.

Across the whole collection zone, plants with:

• Green leaflets with slightly dentate (serrate) margins and a dark, proximal inverted Y mark occurred widely on soils with high exchangeable Ca.

In the east of the collection zone, plants with:

• Green leaflets with no leaf marks and strongly dentate margins occurred most frequently in wetter environments.
• Green leaflets with a smudge-type leaflet mark and serrate margins occurred on high clay soils of lower pH in the mesic, relatively cool, eastern Liverpool Plains region.
• Green leaflets with a brown-purple proximal midrib coloration and serrate margins occurred in cooler, higher altitude, mesic locations on fertile soils.
• Grey leaflets with serrate margins were strongly represented in higher altitude locations, particularly the Liverpool Plains.

In the drier west of the collection zone, plants with:

• Green leaflets with no mark and serrate margins on adult plants occurred widely across warmer, northern regions on slightly saline soils
• Mainly grey leaflets, with proximal blotches and serrate margins occurred more frequently in drier environments.
• Light grey leaflets and serrate margins occurred on more sodic soils of lower N and P fertility in drier environments.

The wide variety of phenotypes within accessions and the considerable overlap of phenotypes among regions suggests many adventitious depositions of M. polymopha burrs during the rapid population of grazing lands with cattle, sheep, horses, and camels during the nineteenth century (there were 1 million sheep in NSW by 1834 [40]). However, some plants observed in this study showed a degree of fitness to their host environments, for example, the early flowering plants from warmer northern regions with generally lower winter rainfall. Genotypes most suited to a particular environment would have survived and flourished at the expense of those less fit. Alternatively, though less likely due to the short time span of about 200 years, there might have been recent adaptation with genetic change in some naturalized M. polymorpha.

Although this study was carried out in the 1990s and the current relevance of the phenotypic and agronomic makeup of the population could be questioned, there is strong evidence that the time span is too short for changes in genotype (adaptation) to have occurred. There is evidence of rapid adaptation to drought of Brassica rapa, an obligate outcrossing annual species [41]. For self-pollinating M. polymorpha, there is evidence of adaptation over 2–4 centuries in North America to new environmental stresses [42] but not over a shorter time frame. A low level only of genetic divergence was evident in polymorphic populations of burr medic from both semi-arid and mesic environments in Chile [43]. In Australia, M. laciniata accessions from western NSW showed no genotypic differences, although they were phenotypically and genotypically different to accessions from northern Africa [44]. Rapid genetic characterization, as in studies with annual Medicago [45], would assist future selection, definition, and commercialization of cultivars of M. polymorpha adapted to Australian environments.

Some plants from accessions sourced from more arid and warmer environments were early flowering and more vigorous; expressed both blotched, grey, and lighter grey leaflets; and produced pods with short spines. This association of attributes may provide an advantage for their naturalization in those environments.

No plants from the accessions collected in New South Wales were spineless. Apart from the transportability of spiny pods, this may be associated with accessions with spiny pods being better able to establish in soils with high clay content [46], which occur widely in parts of the collection zone. Accessions producing short spines occurred mainly in warm, dry environments, especially in the semi-arid regions of the North West Plains Brewarrina and Walgett, and in the Far West region, while those from the most mesic and coolest region, the Liverpool Plains, produced predominantly long spines. The relationship between pod spininess and site characteristic mirrors Chilean studies [5], where accessions with spiny pods were from more mesic areas, and is analogous with Sardinian studies [27] in which accessions with shorter spines occurred at higher altitudes than those with longer spines.

Accessions collected from the drier regions of New South Wales, particularly from soils with a higher exchangeable calcium content, also tended to produce pods with a greater number of pod coils. Although not strongly expressed, this may enhance seed production by those accessions on those soils in drier regions [39], thus aiding the survival and naturalization of the species.

The naturalization of spiny burr medic to the collection zone in New South Wales was likely to be underpinned by the ability of the species to set a high proportion of hard seed, an attribute that varies but nevertheless occurs at a high level [47,48].

This study has identified germplasm that can, potentially, contribute to the development of new cultivars. Lines that produce high levels of dry matter, that flower at a range of times, and that maintain a range of morphological attributes were identified, and seeds were placed with the Australian Pastures Genebank and the Svalbard Global Seed Vault to preserve germplasm for future use, as were elite lines identified in subtropical Queensland [11] (author D.L. Lloyd, personal information). All lines identified produced pods with spines, an attribute controlled by a single gene [49] and deemed undesirable in Australian livestock production systems but readily removed by breeding. In addition, collections such as this may contribute not only to varieties to feed livestock but also to human nutrition and medicine [50].

As climate warms, the diversity of attributes in this naturalized population of burr medic is likely to enable its ongoing persistence, at least into the mid-term, across the collection zone.

5. Conclusions

This study has enhanced the understanding of factors in the survival of the naturalized population of M. polymorpha in the semi-arid, summer dominant rainfall, and warm temperate environments of New South Wales, Australia. The strength of this study was the geographic and climatic range of the environments from which seed was collected, the comprehensive analysis of soils from the collection sites, and the data analysis demonstrating the relationships between phenotypes and the climatic and edaphic characteristics of their source environments. However, the study was limited to the evaluation and comparison of phenotypes, and the determination of genetic differences and similarities among phenotypes would be essential in future studies.

Accessions from warmer and drier locations flowered, on average, earlier than accessions from cooler and more mesic environments, thus enhancing the opportunity for those accessions to set seed in environments in which flowering and the amount of seed set is often limited by soil water availability. This follows the recognized pattern in Mediterranean environments. In addition, individual plants in all climatic regions expressed a wide range of flowering times, with the widest range in the warmer and drier regions, further extending the probability of individual plants flowering and setting seed.

A number of strong locational, mainly east to west associations of climates and soils with attributes of leaflet morphology have also been identified.

All accessions produced pods with spines, the longest by plants from cooler, mesic environments, and the shorter spines by plants from warm and dry environments. While not strongly expressed, plants produced pods with more coils in these latter locations, thus increasing the opportunity to set a larger number of seeds. Allied with the flowering attributes described, this assists the naturalization of burr medic in more challenging environments.

This study has identified plants within the population with agronomic attributes of value to a cultivar improvement program. This, subsequently, enabled a short listing of lines that have been lodged in both National and International Genebanks for future research.