Evaluation of Lucerne Cultivars of Two Winter Activity Classes in Contrasting Pedo-Climatic Mediterranean Environments

Abstract

Lucerne (Medicago sativa L.) forage production may be shifting towards the autumn–spring season, or in cooler environments, due to warmer and drier summers in Mediterranean Europe. This study aimed to evaluate the performance of lucerne cultivars with varying winter activity ratings (WAR) and hardiness in divergent environments of Greece: a cool highland versus a warm lowland. Highlywinter-active (HWA) cultivars were compared to semi-winter-active (SWA) ones for dry matter (DM) yield, seasonal harvest ratios, quality, and agronomic parameters. The SWA cultivars exhibited twice as many overwintered plants and higher summer yields, and were more productive (by 10.4%) in the cooler site. Conversely, HWA cultivars showed enhanced autumn yields and greater productivity (by 10.5%) in the warmer site. Notably, stability analyses revealed trade-offs between productivity and stability, with the most stable cultivar showing lower productivity (by 5.3–5.5%). Additionally, SWA cultivars exhibited higher crude protein content. Plant height and node number correlated with yields in the cool site but not in the warm, underscoring their effectiveness as indirect indicators in marginal environments. Outstanding temperature-specific cultivar responses fully justify the negative relation between winter activity and hardiness regarding productivity. This emphasizes the importance of matching cultivar winter activity and hardiness with specific microenvironments. Region-specific early screening could target the decoupling of the association between low winter activity and increased hardiness, enabling the optimization of cultivars for both traits, while interregional advanced line testing could capitalize on stability and resilience to address the challenges of climate change.

1. Introduction

Lucerne or alfalfa (Medicago sativa L.) has grown in Greece since antiquity [1]. Currently, it is a major spring crop occupying 119,000 ha and yielding annually 18.0 t/ha under irrigation and 7.0 t/ha under dryland conditions [2]. Lucerne stands out as the prevailing forage crop in the Mediterranean region and many parts of the world due to its high protein yield per unit area, high water-use efficiency, low establishment and tilling costs, minimal pesticide and herbicide requirements, high N₂-fixation, and low NO₃- leaching [3,4]. The dominance of lucerne in animal husbandry is attributed to its high nutritive value, characterized by high crude protein concentration (CP), digestibility, and palatability, along with low neutral (NDF) and acid detergent fiber (ADF), and lignin. Forage quality in lucerne is determined by the leaf/stem ratio and the degree of stem lignification, which increases significantly with the onset of flowering [5].

Genetic improvement in lucerne forage yield and nutritive value has shown only modest progress over the last few decades: major reasons are the lack of breeding materials and genetic resources, difficulties in selecting real hybrids or pure lines, a lack of advanced breeding methods, low breeding investment, long selection cycles, and high non-additive genetic variance [4,6]. Recent reports suggest improvements in productivity, with newly released cultivars exhibiting higher winter activity and disease tolerances. However, challenges persist in enhancing cultivar stability and resilience to extreme climatic events such as drought and cold, with trials revealing significant genotype × environment interactions (GEI) [7].

Significant GEI can arise from changes in the magnitude of genotype differences among environments (non-crossover interaction), changes in their relative ranking (crossover interaction), or a combination of both [8]. To leverage GEI profitably, it is necessary to maximize crop yield potential by minimizing GEI and selecting specific adaptations to distinct sub-regions or mega-environments, which are defined according to prevailing climatic, soil, and crop management practices [9]. In lucerne, large GEI implies that the highest-yielding cultivars may not be stable across environments [10]. Data from various environments in Italy demonstrate the importance of breeding for specific adaptation to achieve yield progress over local, top-performing commercial varieties [11,12,13]. The fragmented terrain in the Mediterranean region shapes different pedo-climatic microenvironments, resulting in high GEI [14].

The ongoing climate change in Mediterranean Europe is shifting forage production towards cooler environments with higher precipitation during autumn, winter, and spring [15]. Under these changing conditions, the degree of lucerne winter activity becomes a critical determinant of dry matter (DM) yield, allowing for better utilization of the extended growth period. Winter activity rating (WAR), also known as fall dormancy, is characterized by the reduction or pause of shoot plant height after the final autumn harvest, and is assessed against a set of check cultivars representing different dormancy classes [16]. Cultivar dormancy classes span from 3 to 10: dormant (3–4), semi-dormant (5), semi-winter-active (SWA) (6–7), and highly-winter-active (HWA) (8–10) [17]. Dormancy onset is triggered by decreasing temperatures and daylengths, and is closely linked to winter hardiness, plant survival, persistence, and forage nutritive value [18,19,20]. Dormant cultivars typically exhibit reduced shoot elongation and a decumbent shoot orientation in autumn, making them more winter hardy. Conversely, winter-active cultivars display extensive shoot elongation with a vertical orientation in autumn and may exhibit lower winter survival rates [21]. Winter hardy cultivars are physiologically adapted to withstand subzero temperatures without cellular damage, contributing to improved plant persistence and stand longevity, traits highly valued by growers [22,23]. However, environmental and management stresses can weaken plants, rendering them more susceptible to biotic stresses such as resident insects and pathogens, thereby reducing their winter survival [24].

The changing climate and fragmented terrain in the Mediterranean region pose a significant challenge in matching lucerne cultivars of various winter activities to specific microenvironments. While Greek cultivars generally demonstrate acceptable yield potential and persistence, there is a growing interest, particularly among sheep farmers, in introducing high WAR cultivars to enhance biomass production during the cool, high-precipitation season [25]. Consequently, elite cultivars with high WAR from abroad (Australia, USA, Italy) have undergone testing under both irrigated and dryland conditions to decipher yield and quality determinants amidst the current shifting conditions [25,26]. In brief, HWA cultivars outperformed in autumn harvests but lagged in summer yields, ultimately yielding similarly to local SWA cultivars. Notably, these findings pertain to lowland, warm-zone lucerne crops, overlooking cool, high-elevated, semi-mountainous environments.

Therefore, this study aimed to evaluate 16 lucerne cultivars sourced from various origins and classified into two winter-activity categories (HWA and SWA) across two distinct pedo-climatic environments to identify any adaptability advantages among the two classes.

2. Materials and Methods

2.1. Experimental Sites and Set Up

The study was conducted at two locations in central Greece: the semi-mountainous site of Dolichi (40°04′ N, 22°10′ E, 606 m a.s.l.) and the lowland site of Fanari (39°41′ N, 21°83′ E, 93 m a.s.l.) over four growing years (2017–2020, Figure 1). The soil at Dolichi is characterized as alkaline (pH 8.3), Cambisol sandy loam, while at Fanari, it is slightly alkaline (pH 7.9), Vertisol clay. Additional soil properties are detailed in

Table 1. Soil properties at 0–30 cm depth before the establishment of the experiments.

Variable	Method	Dolichi	Fanari
Sand (%)	Hydrometer	66	18
Silt (%)	“	14	25
Clay (%)	“	20	57
Texture		Sandy loam	Clay
pH (1:1 in H2O)	pH meter	8.3	7.9
Organic matter (g/kg)	Wet oxidation	21.1	16.0
N-NO3 (mg/kg)	2M KCl	12	15
P-Olsen (mg/kg)	Olsen	8	18
K (mg/kg)	NH4-acetate	250	170
CaCO3 (g/kg)	Volumetrically	88	40
EC (mS/cm)	Conductance meter	230	578

. The two locations are typical of those sown to lucerne in Greece. However, Fanari is more fertile, with a pH closer to the optimum (6.6–7.5) for lucerne [4].

The climate in central Greece is characterized as Mediterranean. However, based on the Köppen–Geiger classification, Fanari is categorized as Csa (temperate, dry hot summer), but it also exhibits characteristics close to BSk (arid steppe, cold), whereas Dolichi is classified as Cfa (temperate without a dry season, hot summer) [27]. Climatic parameters during the four growing seasons are summarized in

Table 2. Precipitation (Prec., mm) and mean monthly temperature (T, °C) during the four years as compared to the 55 yr average (1955–2010).

Month	2017		2018		2019		2020		55 yr Average
	Dolichi	Fanari	Dolichi	Fanari	Dolichi	Fanari	Dolichi	Fanari	Dolichi	Fanari
	Prec. (mm)
Jan.	35	32	38	43	33	32	35	24	35	35
Feb.	31	38	34	37	30	42	32	20	31	32
Mar.	37	38	40	38	35	25	37	29	36	38
Apr.	43	38	47	37	40	31	43	49	42	32
May	55	39	60	39	52	32	56	45	55	38
Jun.	37	22	40	19	35	12	37	36	37	24
Jul.	34	20	37	41	32	10	34	36	33	20
Aug.	28	19	31	39	26	8	28	39	28	15
Sep.	34	25	38	43	32	21	35	24	34	33
Oct.	50	49	55	47	47	45	51	26	50	53
Nov.	57	64	62	64	53	68	57	61	56	54
Dec.	54	47	59	46	51	82	54	49	53	52
Total	495	431	539	493	466	408	500	438	490	426
Mar.–Nov.	375	314	408	367	353	252	379	345	371	307
	T (°C)
Jan.	5.1	9.8	5.9	10.0	5.6	10.1	5.3	6.4	5.2	9.9
Feb.	8.1	12.0	7.8	12.2	8.8	12.3	8.4	12.0	8.2	12.1
Mar.	11.7	14.9	11.2	15.3	12.6	15.4	12.0	14.1	11.8	15.1
Apr.	16.3	19.6	15.7	20.0	17.7	20.2	16.8	19.8	16.5	19.8
May	21.7	25.6	20.8	26.2	23.4	26.4	22.3	22.8	21.9	25.9
Jun.	26.5	30.9	25.5	31.5	28.7	31.8	27.3	30.1	26.8	31.2
Jul.	29.3	33.0	28.1	33.6	31.7	34.0	30.2	32.4	29.6	33.3
Aug.	29.3	32.5	28.1	33.1	31.7	33.5	30.2	31.2	29.6	32.8
Sep.	24.7	28.0	23.7	28.6	26.6	28.9	25.4	28.4	24.9	28.3
Oct.	18.6	22.1	17.9	22.5	20.1	22.7	19.2	22.1	18.8	22.3
Nov.	12.4	15.7	11.9	16.1	13.4	16.2	12.8	14.9	12.5	15.9
Dec.	7.4	10.9	7.1	11.1	8.0	11.2	7.7	11.5	7.5	11.0
Τ in Mar.–Nov.	21.2	24.7	20.3	25.2	22.9	25.5	21.8	24.0	21.4	25.0
Min. Τ in Jan.	−12.1	−7.1	0.0	2.0	−9.2	−4.1	−3.1	2.9	−1.6	0.67
Record min T in Jan.	−16.7	−12.3	−4.1	−2.2	−14.4	−9.3	−5.9	−1.0	−16.2	−12.1
Τ Jun.–Aug.	28.4	31.1	26.3	31.7	29.7	32.0	28.3	31.2	28.7	31.4

Min. T, minimum temperature.

. Historical data (55-year average) indicates that Dolichi receives higher total precipitation (490 mm) compared to Fanari (426 mm), but experiences lower mean daily temperatures (21.4 °C vs. 25.0 °C) during the growing season (March–November). January typically exhibits the lowest temperatures, with Dolichi recording colder temperatures than Fanari, as evidenced by the 55-year mean low temperatures (−1.6 °C vs. 0.67 °C). Dolichi also holds the record low temperature of −16.2 °C (Table 2).

The experiments were conducted using randomized complete block designs (RCBD) with 16 treatments (lucerne cultivars) replicated three times. Each plot measured 15 m in length, comprising six rows spaced 0.25 m apart (totaling 22.5 m²). A 1.5 m buffer zone separated neighboring plots, while a 2.5 m buffer zone separated blocks. Sowing occurred in April 2017 at a rate of 15 kg/ha.

In Dolichi, the preceding crop was bread wheat (Triticum aestivum L.), while in Fanari, it was maize (Zea mays L.). Lucerne seeds were not inoculated with rhizobia, but natural root nodulation was confirmed through visual examination. Basal fertilization, based on soil analysis recommendations, included 50 kg N/ha at both locations and 70 kg P₂O₅/ha in Fanari, and 150 kg P₂O₅/ha in Dolichi, broadcasted before seeding and incorporated into the soil. Additionally, 90 kg P₂O₅/ha was applied in early February at both sites in the subsequent years.

Weed control was achieved using Pulsar^® 4 SL (imazamox 40 g/kg, BASF Hellas SA, Maroussi, Athens, Greece) through post-emergence application at the trifoliate stage and yearly application in mid-February in the subsequent years. The experiments were adequately watered to prevent drought stress, with a total of nine irrigations per year at Fanari (630 mm) and six at Dolichi (420 mm).

2.2. Genetic Materials

The trial comprised sixteen lucerne cultivars sourced from various origins, including both local and imported varieties and representing a mix of commercial and experimental development statuses. These cultivars were chosen to encompass two primary winter-activity classifications: SWA (WAR 6–7) and HWA (WAR 8–9). Each WAR rating was represented by four cultivars, totaling eight cultivars per class (

Table 3. Name, reported winter activity rating (WAR), origin and line development status (Commercial, Com.; Experimental, Exp.) of the 16 lucerne cultivars.

Cultivar	WAR A	Origin	Status
Dolichi	6	Greece	Com.
Florina	6	Greece	Com.
Yliki	6	Greece	Com.
Ypati 84	6	Greece	Com.
Icon	7	Australia	Com.
Lamia	7	Greece	Exp.
Pella	7	Greece	Exp.
Vasiliki	7	Greece	Exp.
Talia	8	Australia	Exp.
Chaironia	8	Greece	Com.
Kalliopi	8	Greece	Exp.
57Q53	8	USA	Com.
59N59	9	USA	Com.
Almasa	9	Australia	Exp.
Blue Ace	9	Australia	Com.
Evergreen	9	Australia	Exp.

^A Rating 6 and 7 correspond to semi-winter-active, 8 and 9 to highly winter-active.

).

2.3. Yield and Agronomic Measurements

Within each plot, three randomly selected quadrats measuring 0.25 m² were cut at 5 cm above the soil surface using hand shears at the beginning of the flowering stage (BBCH 60) [28]. Subsequently, the plots were mechanically mown, and all biomass was removed [29]. The fresh biomass samples from each plot were collected and weighed to determine the mean fresh biomass yield per plot. A subsample (~400 g) from each plot was dried in a forced-air oven at 75 °C until a constant weight was achieved. The dry matter yield (DM_H) was then calculated in t/ha for each harvest occasion (DM_H). The sum of DM over the season’s harvests represented the annual DM (DM_A) for each cultivar. The sum of the DM_A constituted the total production (DM_T) over the four growing years. Additionally, for each harvest occasion, the ratio of DM to DM_A (R_H, g/kg) was computed as

$${\mathrm{R}}_{\mathrm{H}}=\frac{{\mathrm{D}\mathrm{M}}_{\mathrm{H}}}{{\mathrm{D}\mathrm{M}}_{\mathrm{A}}}$$

In the establishment year (2017), three harvests were conducted, while in the following years (2018–2020), plots were harvested six times annually. The harvests were grouped to correspond temporally (every 35 days) and physiologically to the same seasons, such as mid-spring (H₁, 10–20 April), late spring (H₂, 15–25 May), early summer (H₃, 20–30 June), late summer (H₄, 25 July–5 August), early autumn (H₅, 30 August–10 October), and late autumn (H₆, 5–15 November). The mean R_H of each harvest was used to compare the seasonal yield distribution.

The plant height to the top of the highest stem (PH) and the number of nodes on the main stem (NN) were measured on 10 randomly selected shoots (one per crown) in each plot, before the second harvest each season, and averaged. Additionally, approximately 25 days after the last harvest each year, the natural plant height (NPH) was measured on 10 plants per plot. This measurement used a scale from 1 to n with 5 cm increments and served as an indicator of autumn growth habit [16].

To monitor plant persistence, measurements were taken on initial plant density (IPD), final plant density (FPD), and plant survival (Survival). These measurements involved counting plants within a 1.0 m row length in the third and fourth rows of each plot. In the first year, the IPD count was performed at the trifoliate stage (BBCH 12) after establishment. In the subsequent years, the FPD count was conducted in mid-February on overwintering plants, with the FPD count of the previous year serving as the IPD for the following year. Plant survival was then determined annually according to the method outlined by Ventroni et al. (2010) [21] as

$$\mathrm{S}\mathrm{u}\mathrm{r}\mathrm{v}\mathrm{i}\mathrm{v}\mathrm{a}\mathrm{l}=\left(\frac{\mathrm{F}\mathrm{P}\mathrm{D}}{\mathrm{I}\mathrm{P}\mathrm{D}}\right)\times 100$$

2.4. Quality Traits

In the second harvest in each year, all the dried subsamples were ground to pass a 1 mm screen and then analyzed for quality traits. Total nitrogen concentration (%N) was determined using the Kjeldahl method and crude protein concentration (CP) was calculated as %N × 6.25 [30]. Neutral detergent fiber (NDF) and acid detergent fiber (ADF) were measured according to Van Soest et al. [31] using an ANKOM 220 Fiber Analyzer (ANKOM Technology Corporation, 2052 O’Neil Road, Macedon, NY, USA). The relative feed value (RFV) was calculated according to Jeranyama and Garcia [32] as

$$\mathrm{R}\mathrm{F}\mathrm{V}=\frac{120/\mathrm{N}\mathrm{D}\mathrm{F}\times (88.9-0.779\times \mathrm{A}\mathrm{D}\mathrm{F})}{1.29}$$

2.5. Statistical Analyses

The sources of variation and the appropriate F-ratios were estimated through repeated measurement analysis of variance (ANOVA) for combined randomized complete block designs (RCBD) over the years (Y) and locations (L). Genotypes (G) and L were considered as fixed factors and Y as random effects for DM_A, quality traits, and agronomic parameters to compare G between Y and L [33]. Two-way ANOVA was performed to compare G across Y within L for R_H and one-way ANOVA compared G within each Y and L combination for DM_A and DM_T.

The Shapiro–Wilk test for normality on standardized conditional residuals indicated that the variables were normally distributed. Additionally, the ANOVA assumptions for the equality of the error variances and residual normality were satisfied based on Levene’s test and the normal quantile-quantile (QQ) plot method, respectively [34].

Means were compared using the LSD test at p < 0.05. Pearson correlation coefficients (r) between the agronomic traits, yield and quality were calculated using the statistical software IBM SPSS package v. 26 (IBM Corp., New York, NY, USA).

The multivariate model G plus G × E interaction (GGE) biplot was utilized to visualize the “which-won-where” pattern for analyzing the GEI within the two-way data matrix (where the environment corresponds to combinations L × Y). This analysis was conducted using the GGE-biplots package in R version 1.0-8 [35], following the methodology outlined by Yan [36].

3. Results

3.1. Growing Conditions

Over the four years of experimentation (2017–2020), the mean daily temperature during the growing season (March–November) in Dolichi averaged 21.5 °C, which was 3.3 °C lower compared to Fanari, where it averaged 24.8 °C. In Fanari, the summer months (June–August) experienced hot temperatures, with a mean daily temperature exceeding 31.0 °C. The coldest winters occurred in 2017 and 2019 for both locations, with Dolichi recording record-low temperatures in January 2017 (−16.7 °C) and 2019 (−14.4 °C). In January, the mean daily temperature was 5.5 °C in Dolichi and 9.1 °C in Fanari (Table 2).

3.2. Dry Matter Yield and Yield Ratios (R_H)

The annual dry matter yield (DM_A) was significantly influenced by years (Y), locations (L), genotypes (G), and their interactions, including two- and three-way interactions (

Table 4. F-criterion and the statistical probabilities of the traits determined in 16 lucerne genotypes tested over four years (2017–2020) in two locations.

Source of Variation	d.f.	DMA	PH	NN	NPH	FPD	Sur	CP	NDF	ADF	RFV
	F-values
Years (Y)	3	525.2 **	421.1 **	20.3 **	14.3 **	250.3 **	45.2 **	25.6 **	36.4 **	19.3 **	518.3 **
Locations (L)	1	234.6 **	24.6 **	12.3 **	6.8 *	17.4 **	13.2 **	18.2 **	18.2 **	13.7 *	429.6 **
Y × L	3	110.3 *	1.7	3.7 *	2.3	3.1	4.5 *	1.2	0.9	2.7 **	135.6 **
Genotypes (G)	15	25.1 **	15.6 **	2.7 **	2.7 **	3.8 **	6.4 **	3.5 **	2.5 **	2.8 **	125.6 **
G × Y	45	61.7 **	23.4 **	3.6 **	5.8 **	12.3 **	14.2 **	2.3 **	3.5 **	6.7 **	30.9 **
G × L	15	24.4 *	12.3 **	2.8 **	4.6 **	7.8 **	6.7 **	5.6 **	2.7 **	2.1 *	23.2 **
G × Y × L	45	5.8 **	2.6 **	2.7 **	2.7 **	6.4 **	5.6 **	1.1	2.2 *	3.5 **	8.3 **

d.f., degree of freedom; DM_A, annual dry matter yield; PH, plant height; NN, number of nodes; NPH, natural plant height; FPD, final plant density; Sur, plant survival; CP, crude protein concentration; NDF, neutral detergent fiber; ADF, acid detergent fiber; RFV, relative feed value. * p < 0.05; ** p < 0.01.

). The significant G × Y × L effect suggests that G performance varied across different Y and L. Therefore, evaluation was conducted within each combination of Y × L (

Table 5. Comparisons for annual and total dry matter yield (DM_A, DM_T) of the 16 lucerne cultivars tested over four years (2017–2020) in Dolichi (Dol) and Fanari (Fan).

Cultivar	WAR	DMA (t/ha)								DMT (t/ha)
		2017		2018		2019		2020
		Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan
Kalliopi	8	6.1	7.6	17.6	24.5	18.5	23.5	18.2	22.1	60.4	77.7
Yliki	6	6.2	7.2	17.8	22.4	21.3	21.5	20.2	20.2	65.5	71.3
59N59	9	5.9	7.4	17.6	24.5	18.2	24.5	18.4	19.1	60.1	75.5
Icon	7	5.1	6.4	16.9	22.3	19.9	21.4	20.1	22.3	62.0	72.4
57Q53	8	5.4	6.8	17.2	22.4	18	21.5	17.2	20.2	57.8	70.8
Blue Ace	9	5.1	6.4	17.4	22.1	17.9	21.2	17.4	19.9	57.8	69.6
Almasa	9	4.3	5.4	15.6	22.1	17.4	21.2	17.4	19.9	54.7	68.6
Ypati 84	6	4.8	6.0	16.3	19.5	19.8	18.7	19.5	17.6	60.4	61.8
Florina	6	4.9	6.1	16.8	18.7	19.9	18.0	20.3	17.4	61.9	60.2
Vasiliki	7	4.7	5.9	16.2	18.4	19.4	17.7	19.5	16.6	59.8	58.5
Dolichi	6	4.4	5.5	16.2	18.1	19.1	17.4	18.9	16.3	58.6	57.3
Lamia	7	4.4	5.5	15.7	17.5	19	16.8	18.7	15.8	57.8	55.6
Talia	8	4.1	5.1	15.6	19.5	17.4	18.4	15.2	17.6	52.3	60.6
Chaironia	8	3.9	4.9	15.7	18.7	17.3	19.2	15.1	16.8	52.0	59.6
Evergreen	9	3.8	4.8	15.5	19.4	17.1	18.6	14.9	17.5	51.3	60.2
Pella	7	4.2	5.3	15.7	17.2	18.9	16.5	16.2	15.5	55.0	54.4
Mean		4.7	5.9	16.4	20.2	18.7	19.5	17.9	18.2	57.8	63.7
LSD0.05		ns	ns	ns	3.20	ns	3.79	ns	2.19	2.18	2.88
CV (%)		15.6	17.1	16.5	10.2	15.4	14.7	18.7	12.4	9.8	14.1
SWA	6–7	4.8	6.0	16.5	19.3	19.7	18.5	19.2	17.7	60.1	61.4
HWA	8–9	4.8	6.0	16.5	21.7	17.7	21.0	16.7	19.1	55.8	67.8
LSD0.05		ns	ns	ns	ns	1.63	1.35	1.99	1.11	2.78	3.21
WAR6		5.1	6.2	16.8	19.7	20.0	18.9	19.7	17.9	61.6	62.6
WAR7		4.6	5.8	16.1	18.9	19.3	18.1	18.6	17.5	58.7	60.2
WAR8		4.9	6.1	16.5	21.3	17.8	20.7	16.4	19.1	55.6	67.2
WAR9		4.8	6.0	16.5	22.0	17.7	21.4	17.0	19.1	56.0	68.5
LSD0.05		ns	ns	ns	2.10	ns	2.29	2.41	1.10	2.19	4.59
CV (%)		12.3	14.5	12.5	9.4	14.6	13.2	17.9	11.6	9.8	13.4

WAR, winter activity rating; S_WA, semi-winter-active; H_WA, highly winter-active; ns, not significant. Note: the cultivars are listed in descending order from the highest- to the lowest-yielding (DM_T).

).

In the establishment year (2017), the lowest DM_A was achieved, with values of 4.7 t/ha in Dolichi and 5.9 t/ha in Fanari, attributed to only three harvests being conducted. Total dry matter yield (DM_T) was 10.2% higher in Fanari (63.7 t/ha) compared to Dolichi (57.8 t/ha). Notably, the highest DM_A was recorded in Fanari during the second growing season (2018), reaching 20.2 t/ha. However, this yield decreased by 3.6% in 2019 (19.5 t/ha) and by 10.0% in 2020 (18.2 t/ha). Conversely, in Dolichi, the highest DM_A was observed in 2019 (18.7 t/ha), declining by 4.5% in 2020.

Comparisons for DM_T revealed that the HWA cvs. ‘Kalliopi’ and ‘59N59’ performed significantly better in Fanari, yielding 77.7 t/ha and 75.5 t/ha, respectively. In contrast, the SWA ‘Yliki’ was the highest-yielding in Dolichi, producing 65.5 t/ha (Table 5). The GGE biplot analysis for ranking the cultivars against the “ideal” for DM_A and stability showed, in descending order, that the best cultivars across locations were ‘Icon’ > ‘Kalliopi’, ‘59N59’, ‘Blue Ace’, ‘57Q53’, ‘Yliki’, with cv. ‘Icon’ being the most stable (Figure 2). The DM_T of cv. ‘Icon’ was inferior by 5.3% and 5.5% compared to the top-performing. In Fanari, the HWA cultivars exhibited a DM_T higher by 10.4% (67.8 t/ha) compared to the SWA ones (61.4 t/ha). Conversely, in Dolichi, the DM_T of the SWA cultivars was higher by 7.7% (Table 5). The most productive cultivars in terms of DM_T, categorized by winter-activity rating, were those with a WAR of 6 in Dolichi, yielding 61.6 t/ha. Conversely, in Fanari, the cultivars classified as WAR 9 and WAR 8 were the best performing, with yields of 68.5 t/ha and 67.2 t/ha, respectively (Table 5). Specifically, the WAR 9 and WAR 8 cultivars yielded 10.5% more than the WAR 6 and WAR 7 cultivars in Fanari. Conversely, in Dolichi, the WAR 6 cultivars had the highest DM_A, outperforming the WAR 8 and WAR 9 cultivars by 10.4%.

Significant differences were observed between cultivars regarding the seasonal yield distribution, as indicated by R_H values (

Table 6. Comparisons of harvest ratios (R_H) for the six harvests in 16 lucerne cultivars tested over four years (2017–2020) in Dolichi (Dol) and Fanari (Fan).

Cultivar	RH1		RH2		RH3		RH4		RH5		RH6
	g/kg
	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan
Kalliopi	280	302	220	200	126	117	120	100	136	141	118	140
Yliki	300	295	182	192	122	112	151	136	155	150	90	115
59N59	285	306	194	175	110	110	125	113	141	141	145	155
Icon	290	300	176	180	145	124	150	127	145	145	94	124
57Q53	291	301	207	201	122	118	135	119	130	134	115	127
Blue Ace	289	298	201	192	112	112	130	110	133	137	135	151
Almasa	288	297	200	191	114	114	139	119	128	128	131	151
Ypati 84	285	295	172	159	115	117	150	150	158	168	120	111
Florina	284	294	169	159	135	135	173	153	150	150	89	109
Vasiliki	282	292	158	148	147	147	173	153	149	149	91	111
Dolichi	279	289	162	152	152	152	173	153	150	150	84	104
Lamia	277	287	178	168	129	139	163	153	153	133	100	120
Talia	244	284	195	175	145	155	140	120	141	115	135	151
Chaironia	244	261	184	163	146	92	145	109	140	234	141	141
Evergreen	271	281	206	196	143	143	140	120	115	115	125	145
Pella	250	260	169	159	167	147	173	173	150	150	91	111
Mean	277	290	188	185	136	128	148	131	143	147	108	118
LSD0.05	14.0	12.1	22.1	15.1	12.1	13.5	13.1	18.5	19.5	23.1	27.4	6.4
CV (%)	14.2	14.2	22.4	14.3	24.5	9.8	14.5	14.5	16.4	16.4	22.1	11.3
SWA	279	289	175	184	145	136	163	148	150	151	88	92
HWA	276	291	202	187	126	120	133	114	137	143	127	145
LSD0.05	ns	ns	18.4	14.5	ns	ns	19.4	14.2	10.3	ns	21.4	18.8
WAR6	283	293	174	193	146	134	161	145	151	153	86	83
WAR7	275	284	175	176	145	139	165	152	149	149	91	100
WAR8	267	287	199	185	132	121	136	112	138	156	127	140
WAR9	285	296	204	189	120	120	130	116	135	130	127	151
LSD0.05	16.5	6.1	5.4	12.3	14.3	10.2	15.4	22.1	7.8	8.6	14.5	18.5
CV (%)	9.5	10.2	13.5	14.5	16.2	8.9	14.5	14.3	12.1	13.2	14.5	16.4

WAR, winter activity rating; SWA, semi-winter-active; HWA, highly winter-active; ns, not significant; R_H1, mid-spring; R_H2, late spring; R_H3, early summer; R_H4, late summer; R_H5, early autumn; RH6, late autumn. Note: the cultivars are listed in descending order from the highest- to the lowest-yielding (DM_T).

). Most cultivars exhibited high values during the spring harvests (R_H1, R_H2), with mean R_H1 values of 277 g/kg in Dolichi and 290 g/kg in Fanari, considerably higher than the respective values for summer (R_H3, R_H4) and autumn harvests (R_H5, R_H6). Specifically, the combined spring harvests (R_H1 + R_H2) contributed the most to DMA, with a value of 470 g/kg, surpassing both summer (271 g/kg) and autumn harvests (259 g/kg). Notably, the top-yielding cultivars ‘Kalliopi’, ‘Yliki’, and ‘59N59’ exhibited higher R_H1 or R_H2 values compared to lower-yielding cultivars.

The first spring (R_H1) and late-autumn (R_H6) harvests were 4.7% higher (290 g/kg vs. 277 g/kg) and 9.3% higher (118 g/kg vs. 108 g/kg), respectively, in Fanari compared to Dolichi. Conversely, the summer harvests (R_H3 + R_H4) were 9.6% higher in Dolichi.

In Fanari, the top-yielding HWA cv. ‘Kalliopi’ exhibited higher R_H1 and R_H6 values (by 7.8% and 18.6%, respectively) compared to Dolichi. Similarly, the HWA cv. ‘59N59’ had higher R_H1 and R_H6 values (by 7.4% and 6.9%, respectively) in Fanari compared to Dolichi. Conversely, the SWA cv. ‘Yliki’ demonstrated 10.1% higher summer yields (R_H3 + R_H4) in Dolichi compared to Fanari (Table 6).

Regarding the seasonal yield distributions, the two WAR classes, SWA and HWA, exhibited a similar declining trend, with high spring yields (R_H1 + R_H2) followed by lower summer (R_H3 + R_H4) and autumn (R_H5 + R_H6) yields.

However, two significant crossover interactions were observed: the first in summer harvests (R_H3 and R_H4), where the SWA cultivars demonstrated significantly higher summer yields, and the second in late autumn (R_H6), where the HWA cultivars outperformed the SWA cultivars (Figure 3).

3.3. Agronomic Traits

Significant interactions between cultivars L and Y indicated the variable performance in plant height (PH), number of nodes (NN), natural plant height (NPH), and final plant density (FPD) (Table 4).

In the warmer climate of Fanari, the plants exhibited higher PH, NN, and NPH (

Table 7. Comparisons of plant height (PH), number of nodes (NN), and natural plant height (NPH) for the 16 lucerne cultivars tested over four years (2017–2020) in Dolichi (Dol) and Fanari (Fan).

Cultivar	PH cm		NN		NPH
	Dol	Fan	Dol	Fan	Dol	Fan
Kalliopi	84.1	89.1	17.4	18.9	7.0	8.7
Yliki	85.2	87.3	19.2	21.3	5.2	7.4
59N59	82.1	87.1	17.2	18.2	7.9	8.8
Icon	78.5	84.2	21.2	21.3	5.6	6.9
57Q53	80.1	85.8	18.1	18.9	7.8	9.4
Blue Ace	79.8	84.8	17.4	18.4	8.3	9.6
Almasa	78.9	83.6	16.2	17.4	7.2	8.8
Ypati 84	78.3	83.3	18.9	19.9	5.0	6.5
Florina	80.1	88.2	18.2	18.4	4.7	6.2
Vasiliki	79.5	84.5	17.9	17.9	5.2	6.6
Dolichi	76.3	81.3	17.4	18.4	4.7	6.2
Lamia	78.5	85.3	19.1	17.4	4.7	6.2
Talia	75.2	80.2	16.5	16.5	6.6	9.2
Chaironia	76.5	80.1	16.2	15.4	6.9	8.1
Evergreen	77.1	90.0	16.2	16.5	8.7	9.9
Pella	77.0	79.5	16.2	18.1	5.3	6.4
Mean	79.2	84.6	17.7	18.3	6.3	7.8
LSD0.05	4.3	5.1	1.5	1.6	1.2	1.4
CV (%)	9.5	14.7	21.4	15.6	18.9	14.2
SWA	79.2	84.2	18.5	19.1	5.1	6.6
HWA	79.2	85.1	16.9	17.5	7.6	9.1
LSD0.05	ns	ns	1.22	0.58	1.45	0.87
WAR 6	80.0	85.0	18.4	19.5	4.9	6.6
WAR 7	78.4	83.4	18.6	18.7	5.2	6.5
WAR 8	79.0	83.8	17.1	17.4	7.1	8.9
WAR 9	79.5	86.4	16.8	17.6	8.0	9.3
LSD0.05	ns	1.2	2.3	1.8	1.5	1.6
CV (%)	15.6	18.4	18.5	12.3	19.5	22.5

WAR, winter activity rating; SWA, semi-winter-active; HWA, highly-winter-active; ns, not significant. Note: the cultivars are listed in descending order from the highest- to the lowest-yielding (DM_T).

). In Dolichi, the top-yielding cultivars were taller (>82.1 cm) compared to low-yielding cultivars, but this trend was not confirmed in Fanari. For example, the low-yielding cultivar ‘Evergreen’ reached a height of 90.0 cm. Similarly, regarding NN, the top-yielding cultivars had more NN in Dolichi, but this trend was not consistent in Fanari. The SWA cultivars had higher NN compared to the HWA ones in both sites. The two locations showed a strong correlation (r = 0.96, p < 0.01) in terms of autumn growth, as evidenced by NPH (Table 7).

Plant survival was lower in Dolichi (especially after the first cold winter in 2017), where the mean survival was 38.9% compared to 53.0% in Fanari. After the second, colder winter in 2019, the survival dropped to 24.9% in Dolichi and 36.7% in Fanari (

Table 8. Comparisons of initial plant density (IPD) in the establishment (2017), final plant density (FPD) after the last year (2020), and plant survival (Survival) in each year for the 16 lucerne genotypes tested over four years (2017–2020) in Dolichi (Dol) and Fanari (Fan).

Cultivar	IPD Plants/m2		FPD Plants/m2		Survival %
	2017		2020		2017		2018		2019		2020
	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan
Kalliopi	289.1	304.1	56.7	98.6	39.5	55.4	34.6	53.9	22.6	39.8	19.6	32.4
Yliki	290.2	305.2	99.9	105.2	55.4	62.1	49.4	55.4	44.5	41.5	34.4	34.5
59N59	292.3	307.3	55.1	105.2	34.6	56.7	33.9	52.3	21.9	40.1	18.9	34.3
Icon	280.3	295.3	98.4	103.2	53.1	60.0	50.1	53.0	42.3	42.9	35.1	35.0
57Q53	315.1	330.1	52.3	88.7	29.9	51.9	28.9	44.9	19.6	29.9	16.6	26.9
Blue Ace	287.6	302.6	39.5	87.4	31.8	54.3	28.8	46.9	16.8	33.6	13.8	28.9
Almasa	365.3	380.3	40.2	88.9	29.0	48.4	26.0	41.4	12.3	27.8	11.0	23.4
Ypati 84	302.4	317.4	97.3	98.5	50.2	56.1	47.2	49.1	35.2	38.4	32.2	31.1
Florina	298.4	313.4	88.4	98.5	45.2	56.5	42.3	49.5	32.7	39.6	29.7	31.5
Vasiliki	297.2	312.2	65.2	85.2	42.3	54.1	37.0	45.3	24.1	39.5	22.0	27.3
Dolichi	304.1	319.1	84.8	78.9	44.5	49.7	42.9	42.7	30.9	41.5	27.9	24.7
Lamia	310.3	325.3	83.4	84.5	44.9	51.0	41.9	44.0	29.9	40.1	26.9	26.0
Talia	270.3	285.3	39.2	75.4	29.8	45.2	28.9	44.2	17.5	29.5	14.5	26.5
Chaironia	265.7	280.7	27.1	78.9	28.2	45.6	25.2	43.5	13.2	36.2	10.2	28.2
Evergreen	312.3	327.3	28.4	74.5	27.1	47.3	24.1	44.3	12.1	28.9	9.1	22.8
Pella	289.1	304.1	55.2	87.5	37.1	53.8	34.1	46.8	22.1	38.4	19.1	28.8
Mean	298.1	313.1	63.2	89.9	38.9	53.0	36.0	47.3	24.9	36.7	21.3	28.9
LSD0.05	ns	ns	16.5	18.2	4.3	4.5	3.2	5.6	6.4	6.2	4.3	5.1
CV (%)	24.5	25.4	23.6	19.8	21.5	22.1	18.5	21.8	19.4	21.3	32.1	28.5
SWA	296.5	311.5	84.1	92.7	46.6	55.4	43.1	48.2	32.7	40.2	28.4	29.9
HWA	299.7	314.7	42.3	87.2	31.2	50.6	28.8	46.4	17.0	33.2	14.2	27.9
LSD0.05	ns	ns	4.3	ns	3.8	4.2	6.5	ns	3.6	2.8	3.2	ns
WAR 6	298.8	313.8	92.6	95.3	48.8	56.1	45.5	49.2	35.8	40.3	31.1	30.4
WAR 7	294.2	309.2	75.6	90.1	44.4	54.7	40.8	47.3	29.6	40.2	25.8	29.3
WAR 8	285.1	300.1	43.8	85.4	31.9	49.5	29.4	46.6	18.2	33.8	15.2	28.5
WAR9	314.4	329.4	40.8	89.0	30.6	51.7	28.2	46.2	15.8	32.6	13.2	27.3
LSD0.05	ns	ns	22.3	ns	4.2	4.2	7.4	ns	7.8	5.6	7.5	ns
CV (%)	24.5	33.2	18.9	22.4	22.6	18.5	17.6	23.1	24.5	18.5	18.9	22.5

WAR, Winter-activity rating; SWA, semi-winter-active; HWA; highly winter-active; ns, not significant. Note: cultivars are listed in descending order from the highest-yielding to the lowest-yielding (DM_T).

). The final plant density (FPD), after the last harvest of the SWA cultivars in Dolichi, was double compared to that of the HWA (84.1 vs. 42.3 plants/m²), while no significant differences were found in Fanari. The SWA cvs. ‘Yliki’ and ‘Icon’ had almost double the FPD (99.9 and 98.4 plants/m², respectively) compared to the HWA cvs. ‘Kalliopi’ and ‘59N59’ (56.7 and 55.1 plants/m², respectively) in Dolichi.

Moderate to strong correlations were found for DM_T with R_H1, PH, NN, NPH, and survival (

Table 9. Correlation coefficients for total dry matter yield (DM_T), dry matter ratio at first harvest (R_H1; mid-spring), plant height (PH), number of nodes (NN), natural plant height (NPH), and plant survival (Survival) for each location (Dolichi, Fanari) for the four growing seasons.

	Dolichi
	DMT	RH1	PH	NN	NPH
RH1	0.68 **
PH	0.73 **	0.57 **
NN	0.76 **	0.56 *	0.33
NPH	−0.50 *	0.00	0.06	−0.50 *
Survival	0.83 **	0.48	0.33	0.84 **	−0.70 **
	Fanari
RH1	0.72 **
PH	0.50 *	0.59 *
NN	0.49 *	0.55 *	0.27
NPH	0.50 *	0.19	0.27	−0.30
Survival	0.57 *	0.1	0.11	0.14	0.38

* p < 0.05; ** p < 0.01 for correlation coefficients; n = 16.

). The strongest correlations were realized in Dolichi, where DM_T was positively correlated with R_H1, PH, NN, and survival (r = 0.68, 0.73, 0.76 and 0.83, p < 0.01, respectively). In Fanari, DM_T was moderately correlated with R_H1 and weakly with PH and survival. The DM_T was weakly correlated with NPH, negatively in Dolichi (r = −0.50, p < 0.05), and positively in Fanari (r = 0.50, p < 0.05). There were no significant correlations between any other harvest ratio with DM_T.

3.4. Forage Quality

Significant differences between cultivars were found for crude protein concentration (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF) and relative feed value (RFV). Also, the two-way interactions (G × Y and G × L) were significant, while the three-way interaction (G × Y × L) was not significant only for CP (Table 4).

Cultivars ‘Icon’ and ‘Yliki’ exhibited the highest CP in both locations, ranging from 232 to 246 g/kg (

Table 10. Comparisons of protein concentration (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF), and relative feed value (RFV) for the 16 lucerne cultivars tested over four years (2017–2020) in Dolichi (Dol) and Fanari (Fan).

Cultivar	CP g/kg		NDF g/kg		ADF g/kg		RFV
	Dol	Fan	Dol	Fan	Dol	Fan	Dol	Fan
Kalliopi	194	191	368	398	297	327	166.2	148.2
Yliki	235	232	378	408	263	293	168.4	150.7
59N59	221	213	358	388	254	284	179.6	160.1
Icon	246	243	364	394	278	308	171.8	153.2
57Q53	189	186	371	401	246	276	174.9	156.4
Blue Ace	185	182	334	364	243	273	194.9	172.8
Almasa	179	176	339	369	287	317	182.6	161.9
Ypati 84	215	212	372	402	284	314	167.0	149.1
Florina	195	192	386	416	309	339	156.2	139.7
Vasiliki	204	201	369	399	299	329	165.4	147.5
Dolichi	228	225	378	408	296	326	162.0	144.8
Lamia	210	207	396	426	286	316	156.5	140.4
Talia	184	181	352	382	273	303	178.7	159.0
Chaironia	175	172	361	391	303	333	168.3	149.8
Evergreen	165	162	352	382	286	316	176.1	156.5
Pella	204	201	367	397	325	355	161.2	143.5
Mean	202	199	365	395	283	313	170.6	152.1
LSD0.05	20.9	11.9	ns	ns	31.9	ns	10.6	ns
CV (%)	14.2	9.8	18.4	15.2	12.3	16.5	18.9	16.3
SWA	217	214	376	406	293	323	163.6	146.1
HWA	187	183	354	384	274	304	177.7	158.1
LSD0.05	18.7	19.2	ns	ns	ns	ns
WAR 6	218	215	379	409	288	318	163.4	146.1
WAR 7	216	213	374	404	297	327	163.7	146.2
WAR 8	186	183	363	393	280	310	172.0	153.3
WAR 9	188	183	346	376	268	298	183.3	162.8
LSD0.05	12.2	23.0	ns	ns	ns	ns	12.3	14.3
CV (%)	12.3	14.2	18.5	16.5	12.3	14.5	9.6	14.2

WAR, winter activity rating; SWA, semi-winter-active; HWA, highly winter-active; ns, not significant. Note: the cultivars are listed in descending order from the highest-yielding to the lowest-yielding (DM_T).

). Additionally, CP was consistently higher in SWA cultivars in both locations. There were no significant differences among cultivars for NDF in either location, while for ADF significant but small differences were observed among cultivars in Dolichi. The RFV ranged from 152.1 in Fanari to 170.6 in Dolichi. Among the cultivars with high DM_T, cv. ‘59N59’ exhibited high RFV in Dolichi, while minor differences were observed in Fanari.

Moderate to strong correlations were found between quality and agronomic traits, with the most important relations regarding CP and NN. Crude protein concentration was positively correlated with NN (r = 0.75 in Dolichi and r = 0.81 in Fanari).

4. Discussion

4.1. Cultivar and WAR Classes Adaptation to the Microenvironments

Exploiting GEI, by selecting cultivars tailored to specific microenvironments holds significant promise for enhancing productivity and sustainability in lucerne cropping systems [37]. However, considering frequent climatic extremes, cultivar productivity must be balanced with stability and resilience [7]. The GGE biplot analysis for yield uncovered crossover interactions, indicating different top-performing cultivars in the two locations. Thus, the HWA cvs. ‘Kalliopi’ and ‘59N59’ were well-adapted to Fanari, the more productive site, while the SWA cv. ‘Yliki’ matched with Dolichi, furthermore, the SWA cv. ‘Icon’ demonstrated the highest stability and adaptability across both microenvironments, though less productive as compared to the top-yielding cultivars. In a broader context, across the USA and Canada, cultivar stability was unrelated to productivity, with resilience negatively correlated with productivity [7].

Given the well-irrigated conditions of the trials, it is evident that the temperature difference during the growing season played a pivotal role in the observed crossover interactions. Considering the base growth temperature of 5 °C for field-grown alfalfa, and depending on genotype, plants could become less winter-active in Dolichi, whereas they remained quite active in Fanari [38].

The comparison between the WAR classes revealed a distinct response in the two microenvironments. Specifically, WAR 6 cultivars (e.g., ‘Yliki’) exhibited a 10.4% higher yield in Dolichi, whereas WAR 9 and WAR 8 cultivars (e.g., ‘Kalliopi’, ‘58N59’) produced 10.5% more in Fanari. Photoperiodism and air temperatures during autumn and winter play crucial roles in determining WAR, a trait that defines lucerne adaptation to varying climatic conditions [20]. Non-dormant cultivars are more productive in regions with mild winters because of the higher shoot elongation in autumn and faster regrowth after cutting. On the other hand, dormant cultivars have reduced shoot elongation and decumbent shoot orientation in autumn, but generally possess greater winter hardiness [21,39].

The assessment of seasonal yield distribution (R_H) proved effective in exploiting cultivar adaptive responses concerning the WAR effect [25]. The spring and autumn harvests demonstrated greater productivity in Fanari, whereas summer yields were elevated in Dolichi. Accordingly, the two WAR classes exhibited crossover interactions; the HWA class displayed higher yields in spring and autumn, while the SWA class performed better in summer. In Fanari, the HWA cvs. ‘Kalliopi’ and ‘59N59’ showed greater responsiveness in autumn. Contrastingly, in Dolichi, the most responsive cultivar during summer was the SWA cv. ‘Yliki’. Previously, under irrigated, warm conditions in central Greece, HWA cultivars yielded higher in autumn, but certain native SWA cultivars were equally productive due to their higher yields in summer [26].

It has been reported that the peak DM_A in lucerne can be achieved either in the second and third growing years or in the third and fourth ones [40,41]. In the present study, the highest DM_A was obtained in the second growing season in Fanari, possibly because the warmer conditions enhanced faster growth. Conversely, in the cooler site Dolichi, the peak DM_A was reached in the third season. The DM_A decreased as the stand aged, but certain cultivars showed more persistence in aging. Environmental and management stresses can weaken plants, making them more susceptible to a combination of environmental stresses, resident insects, and pathogens, ultimately leading to their demise [24]. In our study, the HWA cultivars had nearly half the percentage of overwintered plants compared to the SWA genotypes under the harsh winters in Dolichi; however, no differentiation was evident under the mild winters in Fanari. Consequently, plant survival was a more indicative measure of productivity in Dolichi (r = 0.83) than in Fanari (r = 0.57). Winter hardiness in lucerne encompasses two processes: low-temperature acclimation in autumn and freezing-tolerance adaptation in winter [19,42,43]. Acclimation induced by falling temperatures and shortening photoperiods in early autumn is characterized by a reduced rate of aboveground biomass accumulation, with photosynthetic products being allocated to the rooting system to ensure cold tolerance [20,44]. Strong associations between autumn dormancy and winter hardiness have been documented [45], but breeders have successfully decoupled fall dormancy from winter hardiness, enabling optimization of cultivars for both traits [46,47].

Specific adaptation is crucial for optimal yield expression [48]. One method of improving locally adapted forage legumes has been to hybridize ecotypes with elite populations to leverage specific adaptation effects commonly found in landraces and natural populations [4,49]. Moreover, crossing between lucerne populations exhibiting high per se performance and varying WAR may provide a pathway to developing populations suited to environments with intermittent freezing conditions [50]. Greek lucerne genotypes have demonstrated winter hardiness even under local and even colder conditions, suggesting their utility as sources of cold tolerance [51,52]. An additional strategy involves introducing elite lucerne cultivars with specific autumn dormancy levels that match the winter hardiness needs of their target region [53].

Forage production in many parts of the Mediterranean region is expected to face challenges due to climate change, resulting in shifting growing seasons towards autumn–winter–spring due to drier, longer summers accompanied by severe heat waves [15]. Additionally, declining water tables and increased demand for urban water use have sparked interest in cultivating lucerne under alternative management systems, such as rainfed conditions or without irrigation during summer. Consequently, genotype × management system interactions should be considered [54,55]. These circumstances underscore the need for selecting cultivars with increased yield stability and resilience. Although cultivar productivity has increased with the year of cultivar release, stability has remained unchanged, while resilience to withstand climatic extremes (e.g., drought) has decreased [7]. The points emphasized above underscore the critical need for implementing regional breeding programs with tailored breeding strategies. Specifically, for high elevations, the selection of winter-hardy germplasm with enhanced spring–autumn productivity and accelerated growth immediately after establishment is imperative. Conversely, in lowlands, improving heat tolerance and summer productivity is essential. Early screening could be conducted in each target environment to optimize productivity through specific adaptation. Yet, advanced lines’ testing should take place across various environments, the marginal included, for exploiting GEI and genotype × management system interactions and selecting for increased stability and resilience.

4.2. Variation in Agronomic Traits

Lucerne genotypes exhibit significant variation in stem length and node number [56]. Despite being strongly influenced by environmental factors, these traits are heritable and easy to measure, making them cost-effective indirect selection indices for evaluating DM potential and quality [57]. In the present study, the PH and NN were more indicative assessments of DM_T in the cool environment of Dolichi (r ≥ 0.73) than in Fanari (r ≤ 0.50).

The NPH serves as a proxy for lucerne growth in autumn and helps assess cultivar’s WAR by ranking cultivars based on their autumn growth. While WAR is primarily genetically determined, GEI plays a significant role, influenced by location [16]. This study reaffirmed the significant genetic influence on NPH and, by extension, WAR, as the cultivar ranking for NPH remained consistent across both locations.

4.3. Forage Quality

The nutritive value of lucerne is linked to high CP, RFV, and low NDF and ADF (Annicchiarico et al. 2015) [4]. In this study, lucerne hay demonstrated high quality based on CP (162–246 g/kg), NDF (352–426 g/kg), ADF (243–355 g/kg), and RFV (>140.4) [58]. The SWA cultivars, notably ‘Yliki’ and ‘Icon’, exhibited higher CP levels. Differentiation between cultivars for NDF, ADF and RFV was minimal across locations. Despite the absence of a relationship between cultivar release date and quality, improvements in forage quality can be achieved through agronomic practices [6,59]. Notably, CP levels correlated with high NN, reflecting a relationship between NN and leaf/stem ratio, and subsequently, CP levels [57].

5. Conclusions

Our results underscored the importance of aligning cultivar winter activity and hardiness with specific microenvironments. The SWA cultivars demonstrated superior yields in high-elevation cool sites, attributed to their heightened winter hardiness and increased yields during summer. In contrast, the HWA cultivars excelled in warm, low-elevation sites, benefiting from the absence of winter-related reductions in plant population and the enhanced spring and autumn yield. Additionally, stability analyses revealed trade-offs between productivity and stability, with the most stable and resilient cultivar, showing lower productivity. It is recommended that early screening occur regionally to exploit specific adaptations. Advanced line testing could be conducted interregional to exploit genotype stability and resilience. Specifically, in cooler, high-elevation sites, breeders could aim to decouple fall dormancy from winter hardiness and select germplasm that is more winter-active and hardy, enabling the optimization of cultivars for both traits. Conversely, in the lowlands, the focus could be on improving heat tolerance and summer productivity to achieve optimal yield expression and adaptation to anticipated climatic changes.