Winter Survival, Yield and Yield Components of Alfalfa as Affected by Phosphorus Supply in Two Alkaline Soils

Abstract

Alfalfa (Medicago sativa L.) is an important forage for the development of herbivorous animal husbandry, which is widely planted in the cold climate areas of northern China, where there is low overwintering and forage yield in production, and fertilization is a vital production strategy. A field study was conducted to determine the response of alfalfa’s winter survival rate, yield, and yield components to different gradients of phosphate (P) fertilizer (0, 45, 90, and 135 kg P₂O₅ ha⁻¹) in two types of low-phosphorus alkaline soils. The results show that the winter survival rates and forage yield increased at first and then decreased with the increase of P application; the greater winter survival rates were achieved at 45~90 kg ha⁻¹ of P fertilizer applied, while the greater forage yield were achieved at 90 kg ha⁻¹ in the two sites, and the suitable P fertilizer application rates for greater winter survival were less than those for greater forage yield; plant height and shoot mass made a greater contribution to forage yield than other yield components. To pursue greater forage yield, the suitable P fertilization rates for aeolian sandy soil and silt loam soil are 108.1 and 78.3 kg ha⁻¹, respectively, based on the regression equations. Therefore, in cold winter and soil P deficiency areas, applying P fertilizer can be used as an effective strategy to improve alfalfa’s forage yield and persistence, and the most appropriate amount of fertilizer should be determined according to climate and soil conditions.

1. Introduction

Alfalfa (Medicago sativa L.) is one of the world’s most valuable fodder crops, planted for hay and pasture in more than 80 countries with an area of 30~35 million ha [1,2]. Around the world, alfalfa is grown under a wide range of environmental and climatic conditions with variable precipitation, temperature, and soil fertility [2]. The northern hemisphere includes the mid-west United States, western and eastern Canada, western and northern China, and Northern Europe [3], with cold winters and low temperatures as major factors limiting alfalfa production and distribution, causing severe economic losses [4].

Fertilization is one of the important measures to improve alfalfa’s cold resistance and production performance [5]. Because alfalfa itself has rhizobia to fix nitrogen, no additional nitrogen fertilizer is needed [6], so P and potassium fertilizers are mainly applied in alfalfa production [7]. P deficiency is one of the most limiting factors of crop production [8] and plant survival [9] in many soils worldwide, which also greatly impacts alfalfa production and cold tolerance [10]. Sometimes the problem with P deficiency is not the total P content in the soil, but P is relatively immobile and often poorly available to plants [11]. Because of the low concentration and mobility of available P in the soil, P fertilizer has been used widely in crop production and increased crop yield by a large margin [12,13]. However, P fertilizers are often applied to soils above plant requirements, and many soils have accumulated large stores of residual P and non-labile inorganic P over time [14]. Excessive application of P fertilizer in the soil not only increases the production cost but also causes environmental pollution [15]. Furthermore, P is a non-renewable resource [16]. It is very important to apply an appropriate amount of P fertilizer to alfalfa.

Alfalfa forage yield can be described as the product of three plant characteristics: plants m⁻², shoots per plant, and shoot mass [17]. Understanding which yield component has a greater impact on forage yield is important, but people have different views on this. Several investigations concluded that shoot mass is the most important component of forage yield [7,18]. By contrast, Volenec et al. [17] and Undersander et al. [19] reported that plant population plays an important role in forage yield, while the plant population and shoots per plant can determine shoots m⁻². The shoots plant⁻¹ would decrease as populations increased from 11 to 86 plants m⁻² [20]. The decline in plant population may be partially or fully compensated for by increases in other yield components such as shoots plants and shoot mass, so as to maintain the forage yield at a higher level [18]. Shoot mass could be decided by plant height and mass per centimeter of the shoot (g cm⁻¹) [12], and thus plant height also can be used as a trait explaining the variation of forage yield [21]. Fertilization with P often increases forage yield and affects the yield components [7,22,23]. Jung and Smith [9] reported that P fertilization was essential for plant survival. Sanderson and Jones [22] reported that P fertilization increased shoots per plant of field-grown alfalfa when compared with alfalfa not receiving P. Yang and Shi [24] reported that P fertilizer could significantly increase the plant height and shoots per plant of alfalfa. Shoot mass is the yield component most closely associated with yield increases in response to P fertilization, caused by new shoot growth that is quicker because increased nutrient mobilization occurs between roots and actively growing shoots [25].

Alfalfa is an important perennial herbaceous forage legume for the animal industry in Inner Mongolia [26]. Due to the rapid development of animal husbandry, the alfalfa cultivation area continued to expand. Since the government has set a red line of 120 million ha of arable land area, farmers can only plant alfalfa on marginal land, many soils in Inner Mongolia contain inadequate amounts of plant-available P, and their fertilization with P is required to achieve an optimal yield of alfalfa. Many studies conducted with P fertilizer on crops indicate that soil-applied fertilizer P may be available for many years [27]. The first hypothesis of this study was that P fertilization will improve one or more yield components of alfalfa, and those components affect the forage yield. The second hypothesis was that the effect of P fertilization will be different under different climatic conditions in the alfalfa planting area. The objectives of this study were to: (i) identify the winter survival, forage yield, and yield components of alfalfa response to P fertilizer on two sites; (ii) optimize P application rates that maximize forage yield and winter survival on the two sites.

2. Materials and Methods

2.1. Growth Conditions

Two field experiments were conducted in central Inner Mongolia. The two experimental sites were situated on two alkaline soils, i.e., an aeolian sandy soil and silt loam soil. The first one was aeolian sandy soil in Siziwang County (41°40′ N, 111°78′ E), and the second one was silt loam soil in Tuzuo County (40°34′ N, 111°46′ E); analysis was performed on soil samples taken to a 15 cm depth before planting (

Table 1. Initial soil characteristics (15 cm depth) for the two sites.

Site	Soil Type	Organic C (g kg−1)	Total N (g kg−1)	Available P (mg kg−1)	Available K (mg kg−1)	Salt (g kg−1)	pH
Siziwang	Aeolian sandy soil	30.5	1.71	21.4	306.5	1.43	8.6
Tuzuo	Silt loam soil	7.30	1.09	13.3	384.0	1.39	8.5

). The soil in Inner Mongolia is divided into 11 grades according to the level of P deficiency [28], with the soils in the two sites belonging to levels 2 and 3, respectively, both of which are mildly deficient in P. The average precipitation and temperature of each month from 2019 to 2021 are shown in

Table 2. The air temperatures (°C) and precipitation (mm) by month for each year.

Month	Siziwang County						Tuzuo County
	Temperature (°C)			Precipitation (mm)			Temperature (°C)			Precipitation (mm)
	2019	2020	2021	2019	2020	2021	2019	2020	2021	2019	2020	2021
Jan.	−11.9	−11.4	−11.9	0	4.8	0	−9.8	−8.5	−9.5	0	3.3	5.3
Feb.	−10.8	−5.8	−4.1	4.4	0	2.0	−8.0	−4	0	6.3	14.5	0.1
Mar.	−0.2	0.1	1.9	8.9	15.1	9.2	2.0	3	4.5	3.7	25.1	6.7
Apr.	8.5	6.5	7.1	29.5	0.2	0	10.9	8.5	8.5	17.2	3.7	0.1
May	13.6	14.3	12.7	31.4	19.6	36.4	15.5	16	15	9.8	29.2	23.5
June	19.2	19.4	18.5	48.9	51	62.7	21.4	20	20.5	47.9	73.2	19.4
July	19.9	19.6	20.8	127.6	176.8	81.9	21.4	21	23	102.3	89.1	104.7
Aug.	18.2	18.6	17.6	82.5	103.1	67.3	19.8	20	20	91.3	68.9	171.6
Sept.	14.9	12.7	15.1	62.5	9.9	30.6	16.8	14.5	17.5	101.4	25	54.2
Oct.	6.3	4.4	4.7	11.3	14.6	13.7	8.0	5.5	7.5	18.3	23	10.2
Nov.	−2.0	−3.6	−3.4	4.7	18.8	18.7	−0.5	−0.5	−1	7.1	0.1	5.7
Dec.	−9.9	−13.7	−8.6	1.6	0	0	−9.7	−11	−6.5	6.8	0.1	0.3

.

2.2. Plant Materials

The cultivars used in the study were Zhongcao No.3 and Gannong No.3. The cultivar Zhongcao No.3 was bred by the grassland research institute of the Chinese Academy of Agricultural Sciences; it is a dormant cultivar, and suitable for cultivation in dry and cold areas. The cultivar Gannong No.3 was bred by Gansu Agricultural University; it is a semi-dormant cultivar, and suitable for cultivation in irrigated and template areas. The germination rates of the two cultivars were both more than 93%, based on a germination experiment.

2.3. Field Design

The experiment was planted on the 23rd of July in Siziwang County and on the 25th of July in Tuozuo County in 2019. Seeds were planted manually in each plot (3 m × 5 m), at 15 kg ha⁻¹ at a soil depth of 1~2 cm and 30 cm row spacing. Treatments were arranged in a randomized complete block design with a split-plot arrangement with three replications at two sites. The whole treatments were P fertilization treatments, and the subplot treatments were cultivars. Whole treatments were surrounded by a one-meter-wide strip of unfertilized Zhongcao No.3 alfalfa to prevent border effects. Application of P fertilization was carried out on the 23rd and 24th of August, four P fertilizer (P₂O₅) treatments (0, 45, 90, and 135 kg ha⁻¹) were applied to a 5 cm depth in the middle of two rows and then covered and irrigated, which was produced by Henan Jixin Chemical Products Co., Ltd., Zhengzhou, China. No other fertilizer was supplied throughout the experiment. Plots were hand-weeded during the growing period whenever necessary for proper weed control. Plants were kept well-watered, and no diseases or pests occurred during the experiment period.

2.4. Investigation of Winter Survival Rate and Sampling

The winter survival rate was calculated depending on the number of initial plants (W) on 5–10 October 2019, and the number of living plants (L) was counted on 23–26 April 2020 and again on 25–30 April 2021 within 0.3 m² in each plot, according to the following formula:

$$\mathrm{Winter} \mathrm{survival} \mathrm{rate} \left(\%\right)=\frac{\mathrm{L} }{\mathrm{W}}\times 100$$

Two forage harvests at the early flowering stage from early July to late August were carried out in Siziwang County and three forage harvests at the early flowering stage from late June to middle September were carried out in Tuzuo County in 2020 and 2021. Sampling should be avoided on borderlines to prevent marginal utility. A an area of 1.0 m² from the center of each plot was harvested, approximately 5 cm above the ground. Fresh weight was determined, dried at 105 °C for 30 min and then at 65 °C for 48 h, re-dried and re-weighed, until it reached a constant weight; the percentage of dry matter of the forage was determined, and the dry matter yield per hectare was calculated based on the forage fresh weight and the percentage of dry matter of the forage. Shoots per plant were calculated by counting ten plants in each plot before harvest. Plant height was measured for ten plants in each plot before harvest. A subsample of 10 shoots was hand-collected in each plot before harvest, weighted, and the percentage of dry matter of forage was used to calculate the dry mass shoot⁻¹. Taproots in a 1.0 m² area of each plot were excavated to a depth of 10 cm and counted before each harvest in each year to determine plant density. The annual forage yield was counted by summing the yield of each harvest. After each harvest, all border areas around plots were removed.

2.5. Statistical Analysis

Analysis of variance was performed on winter survival rate, yield, and yield components using SPSS 26. Where F-tests were significant (p < 0.05), LSD was calculated to compare means. Data were analyzed using the general linear models (GLM) procedure. Because forage yield is a linear function of plants m⁻², shoots plant⁻¹, plant height, and mass shoot⁻¹, we used correlation analysis to determine the effects of each yield component on forage yield. These values were assigned to dummy variables representing linear, quadratic, cubic, and quartic polynomial contrasts and their interaction. Occasionally, a regression model exhibiting an anomalous relationship was rejected. Differences in regression slopes were determined using the standard error of the difference between the slopes adjusted for the experimental error [7]. We used the general linear model procedure to identify the P fertilization rates that maximize forage yield and winter survival. Because forage yield is a linear function of plants m⁻², shoots plant⁻¹, plant height, and shoot mass, and given those yield components themselves are often correlated with one another, we used path analysis to determine the direct effects of individual yield components on forage yield [7].

3. Results

3.1. Main Effect of Soil Types and P fertilizer

Although differences among cultivars with fall dormancy category existed for winter survival, forage yield of each harvest, plants m⁻², and shoots plant⁻¹, as expected (

Table 3. Analysis of variance for yield, plants m⁻², shoots plant⁻¹, height, shoot mass, annual yield and winter survival for site, harvest, year, cultivar and P treatment.

Source	df+	Yield	Plants m−2	Shoots Plant−1	Height	Shoot Mass	Annual Yield	Winter Survival
Site	1	***	***	***	***	***	**	*
Harvest	2	***	***	***	***	***
Year	1	***	***	***	ns	***	***	***
Cultivar	1	***	***	***	ns	ns	***	***
P	3	***	***	***	***	***	***	***
Site × harvest	1	***	ns	ns	***	ns
Site × year	1	***	***	ns	***	***	***	**
Site × cultivar	1	ns	***	ns	***	*	ns	ns
Site × P	3	***	***	ns	***	***	**	ns
Harvest × year	2	***	ns	***	*	*
Harvest × cultivar	2	ns	*	***	ns	**
Harvest × P	6	**	ns	ns	***	***
Year × cultivar	1	ns	ns	*	ns	*	ns	ns
Year × P	3	*	**	ns	**	***	ns	ns
Cultivar × P	3	ns	***	*	ns	ns	ns	**
Site × harvest × year	1	**	ns	***	ns	***
Site × harvest × cultivar	1	ns	ns	ns	ns	***
Site × harvest × P	3	ns	ns	ns	ns	ns
Site × year × cultivar	1	ns	ns	ns	ns	ns	ns	ns
Site × year × P	3	***	ns	ns	***	***	***	ns
Site × cultivar × P	3	**	**	ns	ns	ns	*	ns
Harvest × year × cultivar	2	ns	ns	*	ns	**
Harvest × year × P	6	*	ns	ns	***	***
Harvest × cultivar × P	6	ns	ns	**	ns	ns
Year × cultivar × P	3	ns	ns	ns	ns	**	ns	ns
Site × harvest × year × cultivar	1	ns	ns	ns	ns	*
Site × harvest × year × P	3	ns	ns	ns	ns	*
Site × harvest × cultivar × P	3	ns	ns	ns	ns	*
Site × year × cultivar × P	3	ns	ns	ns	ns	ns	ns	ns
Harvest × year × cultivar × P	6	ns	ns	ns	ns	*
Site × harvest × year × cultivar × P	3	ns	ns	ns	*	***

Note: *, **, and *** Significant at 0.05, 0.01, and 0.001 probability levels, respectively. df+: Degrees of freedom. ns: Not significant at the 0.05 probability level.

), the focus of the study was on soil types and P fertilizer effects. Consequently, cultivars with fall dormancy category effects will not be discussed.

3.2. Winter Survival Rate

The effects of P fertilizer on winter survival were investigated in the two sites (Figure 1). In the two sites, the winter survival rates both increased first and then decreased with the increase of P application over two years, and the winter survival rates of 2020 were higher than those of 2021 in the two sites. In Siziwang County, the winter survival rate reached 85.6%% at 90 kg ha⁻¹, but did not have a significant difference from the control in 2020, while the winter survival rate reduced to 63.8% at 45 kg ha⁻¹ in 2021, and had a significant difference from the control (Figure 1a). In Tuzuo County, the winter survival rate reached 93.4% at 45 kg ha⁻¹ in 2020, which was significantly higher than that of the control, while the winter survival rates in 2021 did not have a significant difference among P treatments, which reached 65.3% at 90 kg ha⁻¹, a little higher than other treatments (Figure 1b).

The response curve and degree of response were different for the two sites. However, P fertilizer had significant linear effects on the mean winter survival rate for the two years (Figure 2). The quadric curve fit the data best for the winter survival rate in the two sites. In Siziwang County, the winter survival rate and P fertilizer had a significant quadratic regression relationship (p < 0.01), where the winter survival rate reached a maximum value at the P fertilizer application of 60.5 kg P₂O₅ ha⁻¹ (Figure 2a). In Tuzuo County, the winter survival rate and P fertilizer also had a significant quadratic regression relationship (p < 0.05), where the winter survival rate reached a maximum value at the P fertilizer application rate of 58.0 kg P₂O₅ ha⁻¹ (Figure 2b).

3.3. Forage Yield

The effects of P fertilizer on forage yield were investigated in the two sites (

Table 4. Effects of P fertilization on the yield of alfalfa in the two sites.

Site	Year	Harvest	Levels of Applied P (kg ha−1)
			0	45	90	135
Siziwang	2020	H1	3859.8 ± 367.6 b	5508.8 ± 851.9 ab	5809.2 ± 448.6 a	5943.8 ± 598.3 a
		H2	2038.7 ± 304.1 b	2648.1 ± 264.9 b	3242.3 ± 623.9 ab	4348.4 ± 487.9 a
		Annual	5898.5 ± 564.2 b	8206.9 ± 927.2 ab	9101.5 ± 1006.1 a	10,292.2 ± 934.1 a
	2021	H1	3882.5 ± 387.9 a	4638.6 ± 401.5 a	4537.5 ± 190.2 a	4019.6 ± 210.3 a
		H2	2142.6 ± 200.9 b	2687.1 ± 245.8 b	3849.4 ± 397.3 a	2873.2 ± 405.3 b
		Annual	6025.1 ± 516.3 b	7325.8 ± 487.3 ab	8386.8 ± 487.9 a	6892.8 ± 567.5 ab
Tuzuo	2020	H1	3078.0 ± 164.8 c	4999.1 ± 354.1 b	6562.2 ± 474.2 a	2437.0 ± 161.5 c
		H2	2321.2 ± 119.4 c	3292.2 ± 327.5 b	4064.4 ± 258.7 a	2072.1 ± 154.4 c
		H3	1625.3 ± 93.9 c	2143.1 ± 152.5 b	2709.1 ± 131.8 a	1809.2 ± 227.1 bc
		Annual	7024.5 ± 313.6 c	10,434.4 ± 645.5 b	13,335.6 ± 320.8 a	6318.3 ± 392.8 c
	2021	H1	3569.7 ± 563.0 c	4498.7 ± 627.3 bc	6331.7 ± 709.9 a	5543.9 ± 320.8 ab
		H2	3375.8 ± 296.1 b	5327.0 ± 487.1 a	4952.0 ± 293.1 a	5535.7 ± 225.8 a
		H3	2339.6 ± 214.3 ab	2111.9 ± 130.9 b	3003.4 ± 133.9 a	2624.4 ± 379.9 ab
		Annual	9285.1 ± 866.6 b	11,937.6 ± 977.5 a	14,287.2 ± 846.2 a	13,704.0 ± 697.1 a

Note: Different letters in the same row indicate a significant difference among P fertilization treatments (p < 0.05).

). The site, year, harvest, and P fertilizer all impacted the forage yield.

In Siziwang County, the forage yield of each harvest increased with the increase of P fertilization in 2020; when P was applied at more than 90 kg ha⁻¹, the forage yield of the first harvest was significantly higher than that of the control, and when P was applied at 135 kg ha⁻¹, the forage yield of the second harvest was significantly higher than that of the control. The annual forage yield of 2020 also increased with the increase of P fertilization. When P was applied at more than 90 kg ha⁻¹, it reached more than 9000 kg ha⁻¹, which was significantly higher than that of the control. The forage yield of each harvest both increased first and then decreased with the increase of P fertilization in 2021; when P was applied at 45 kg ha⁻¹, the forage yield of the first harvest reached 4638.6 kg ha⁻¹, which was not significantly higher than the control, while when P applied at 90 kg ha⁻¹, the forage yield of the second harvest reached 3849.4 kg ha⁻¹, which was higher than the control (p < 0.05). The annual forage yield also increased first and then decreased with the increase of P fertilization; it reached 8386.6 kg ha⁻¹ under 90 kg P₂O₅ ha⁻¹, which was increased by 39.2% relative to the control.

In Tuzuo County, the forage yield of each harvest and the annual yield both increased first and then decreased with the increase of P fertilization in 2020. When P was applied at 90 kg ha⁻¹, each harvest and annual forage yield reached the greatest value, the forage yields of each harvest were 6562.2 kg ha⁻¹, 4064.4 kg ha⁻¹ and 2709.1 kg ha⁻¹, and the forage yield was 13,335.6 kg ha⁻¹, all of them significantly higher than those of the controls. The forage yield of the first harvest increased first and then decreased with the increase of P fertilization in 2021; the application of P fertilizer at 90 kg ha⁻¹ reached 6331.7 kg ha⁻¹, which was significantly higher than that of control, while the forage yields of the second harvest under P fertilization were all greater than those of the control (p < 0.05). The forage yield of the third harvest did not have an obvious difference between the P treatments and the control. However, when P was applied at 90 kg ha⁻¹, the forage yield was much greater than the others. The annual forage yield of 2021 increased first and then decreased with the increase of P fertilization; when P₂O₅ was applied at 90 kg ha⁻¹, it reached 14,287.6 kg ha⁻¹, which was increased by 53.9% relative to the control.

The application of P fertilizer had significant linear effects on forage yield, but the response curve and degree were different in the two sites (Figure 3). The forage yield and P fertilizer had significant quadratic regression relationships in the two sites (p < 0.01). In Siziwang County, through derivative calculation, the maximum mean forage yield of the two years reached 8774.4 kg ha⁻¹ under 108.1 kg P₂O₅ ha⁻¹. In Tuzuo County, through derivative calculation, the maximum mean forage yield of the two years reached 13,022.3 kg ha⁻¹ under 78.3 kg P₂O₅ ha⁻¹.

3.4. Yield Components

3.4.1. Plant Population Density

Alfalfa forage yield can be described as the product of three components: plants m⁻², shoots plant⁻¹, and mass shoot⁻¹ [17]. The application of P fertilizer impacted the plant population density of alfalfa in the two sites (Figure 4). The plant population density increased first and then decreased with the increase of P fertilizer in the two sites. In Siziwang County, when P was applied at 90 kg ha⁻¹, the plant population density of each harvest was greater than the control (p < 0.05) (Figure 4a). Ignoring the influence of P and cultivar, plots averaged 56.25 plants m⁻² in July 2020, and this decreased to 51.92 plants m⁻² in late August 2020, decreased to 41.08 plants m⁻² in July 2021, and decreased to 29.38 plants m⁻² in late August 2021 (not shown in the figure). By the regression analysis, the plant population density had significant linear effects on forage yield (

Table 5. Regression equations, R² values, and p values for the influence of plants m⁻² on forage yield at each harvest in 2020 and 2021.

Site	Harvest	2020			2021
		Plants m−2	R2	p Value	Plants m−2	R2	p Value
Siziwang	H1	y = 2875 + 42.76x	0.087	0.163	y = 3948 + 7.83x	0.006	0.720
	H2	y = 2879 + 4.14x	0.001	0.880	y = 7816 − 401.7x + 7.66x2	0.232	0.063
Tuzzuo	H1	y = 1038 + 30.125x	0.112	0.110	y = 14,351 − 250.92x + 1.606x2	0.070	0.467
	H2	y = 601 + 23.041x	0.222	0.020	y = 3266 + 21.53x	0.060	0.248
	H3	y = 360 + 17.849x	0.380	0.001	y = 4846 − 61.041x + 0.374x2	0.060	0.552

). The regression models subsequently developed exhibited a linear relationship between forage yield and plants m⁻² of each harvest. The forage yield increased with the increase in the plant population density.

In Tuzuo County, when P was applied at 45 kg ha⁻¹, the plant population density of each harvest was greater than that of the control. Ignoring the influence of P fertilizer and cultivar, plots averaged 107.25 plants m⁻² in June 2020, decreased to 95.9 plants m⁻² in September 2020, and then decreased to 79.0 plants m⁻² in June 2021 and decreased to 66.0 plants m⁻² in September 2021 (Figure 4b). The regression models developed exhibited a linear relationship between forage yield and plants m⁻² of each harvest. The forage yield increased with the increase in the plant population density (Table 5).

3.4.2. Shoots per Plant

The effects of P fertilizer on shoots number were investigated in the two sites (Figure 5). In Siziwang County, the shoots per plant increased with the increase of P fertilization in the two years. In 2020, when P was applied at 135 kg ha⁻¹, the shoots plant⁻¹ of the first harvest was 10.1, increased by 28.8% relative to the control (p < 0.05); the shoots plant⁻¹ of the second harvest did not have an obvious difference among fertilization treatments, from 12.5 to 14.2. In 2021, when P was applied at more than 90 kg ha⁻¹, the shoots plant⁻¹ of the first harvest was greater than that of the control (p < 0.05), increased by more than 19.2%; and the shoots plant⁻¹ of the second harvest did not have a significant difference among fertilization treatments, arranged from 12.0 to 13.1. In the two years, the shoots plant⁻¹ of the first harvest was less than that of the second harvest (Figure 5a). The regression models developed exhibited a consistent positive linear relationship between forage yield and shoots plant⁻¹ of each harvest, and the forage yield increased with the increase of the shoots per plant (

Table 6. Regression equations, R² values, and p values for the influence of shoots per plant on forage yield at each harvest in 2020 and 2021.

Site	Harvest	2020			2021
		Shoots Plant−1	R2	p Value	Shoots Plant−1	R2	p Value
Siziwang	H1	y = 2955 + 266.348x	0.070	0.213	y = 2030 + 224.385x	0.170	0.045
	H2	y = −503 + 275.734x	0.231	0.017	y = 1462 + 112.16x	0.041	0.345
Tuzzuo	H1	y = 3388 + 94.746x	0.031	0.411	y = 3811 + 142.593x	0.019	0.519
	H2	y = −1556 + 1085x − 61.508x2	0.219	0.075	y = 3148 + 125.969x	0.064	0.233
	H3	y = 10.632 + 0.0002x	0.001	0.883	y = −524 + 298.51x − 6.312x2	0.274	0.035

).

In Tuzuo County, the shoots plant⁻¹ of the first harvest increased first and then decreased with the increase of P fertilization treatments in 2020; when P was applied at more than 90 kg ha⁻¹, the shoots plant⁻¹ was significantly greater than that of the control; the shoots plant⁻¹ of the second harvest of 2020 increased with the increase of P fertilization treatments and increased from 25.4% to 78.0%, while the shoots plant⁻¹ of the third harvest did not change obviously under any P fertilization. In 2021, the shoots plant⁻¹ of the first harvest under any P fertilization did not have a significant difference from the control, and the shoots plant⁻¹ of the last two harvests did not have a significant difference between applied P fertilizer treatments and the control. In the same year, the shoots plant⁻¹ of the following harvest was more than that of the previous harvest (Figure 5b). The regression analysis, there was a linear correlation between forage yield and shoots plant⁻¹ of the first harvest in 2020 (y = 3388 + 94.746x), the forage yield increased with the increase of shoots plant⁻¹, and there was a quadratic regression correlation between forage yield and shoots plant⁻¹ of the second harvest in 2020 (y = 1556 + 1085x − 61.508x²). The optimum shoots plant⁻¹ was 8.82; the linear relationship between forage yield and shoots plant⁻¹ of the third harvest was not obvious, and the slope was 0.0002 (Table 6). In 2021, there were linear correlations between forage yield and shoots plant⁻¹ of the first two harvests, and the forage yield increased with the increase of shoots plant⁻¹. The forage yield of the third harvest had a quadratic regression correlation with the shoots plant⁻¹, and the optimum shoots plant⁻¹ was 23.6 (Table 6).

3.4.3. Plant Height

The effects of the P fertilizer rate on plant height were investigated in the two sites (Figure 6). In Siziwang County, the plant height did not have an obvious change in 2020; the plant height of the first harvest ranged from 60.7 to 67.8 cm, and the second harvest ranged from 49.2 to 55.0 cm. The plant height of the first harvest increased first and then decreased in 2021, when P was applied at 45 kg ha⁻¹, the plant height was 52.05 cm, which was an increase of 12.7% relative to the control; and the plant height of the second harvest did not show a significant change with the increase of P fertilization, ranging from 40.7 to 44.9 cm. The plant height of the first harvest was greater than that of the second harvest in the two years (Figure 6a). The regression models developed exhibited a consistent positive linear relationship between forage yield and plant height of each harvest in 2020; the forage yield increased with the increase of plant height (

Table 7. Regression equations, R² values, and p values for the influence of plant height on forage yield at each harvest in 2020 and 2021.

Site	Harvest	2020			2021
		Plant Height	R2	p Value	Plant Height	R2	p Value
Siziwang	H1	y = −804 + 94.459x	0.277	0.008	y = −21,333 + 966.187x − 9.014x2	0.233	0.061
	H2	y = −1830 + 94.493x	0.354	0.002	y = −30,635 + 1514.672x − 16.894x2	0.221	0.072
Tuzzuo	H1	y = −1656 + 92.598x	0.497	0.000	y = −37,232 + 1289.212x − 9.692x2	0.277	0.033
	H2	y = −1027 + 32.13x	0.256	0.012	y = 3264 + 21.577x	0.023	0.483
	H3	y = 575 + 40.027x	0.184	0.036	y = −9035 − 311.448x + 3.618x2	0.348	0.011

). As plant height increased, the forage yield increased until plant height reached 53.6 and 44.8 cm for every harvest in 2021, respectively (Table 7).

In Tuzuo County, the plant heights of the first two harvests increased first and then decreased with the increase of P fertilization in 2020; the plant height reached 76.3 and 73.4 cm, respectively. When P was applied at more than 90 kg ha⁻¹, the plant heights of the third harvest were greater than those of the control. The plant heights of 2021 did not change obviously with the increase of P fertilization. The plant heights of the first two harvests were greater than the first harvest (Figure 6b). Forage yield also had a linear correlation with plant height in 2020; the forage yield increased with greater plant height (Table 7). The regression models developed exhibited a consistent quadratic correlation between forage yield and plant height of the first harvest; the forage yield reached a greater value when the plant height was 66.5 cm; the forage yield of the second harvest showed a linear correlation with plant height, the forage yield increased with greater plant height; the forage yield of the third harvest also exhibited a consistent quadratic correlation with plant height; when the plant height was more than 43.0 cm, the forage yield increased with the plant height (Table 7).

3.4.4. Shoot Mass

The effects of P fertilizer on shoot mass were investigated in the two sites (Figure 7). In Siziwang County, the shoot mass of each harvest increased first and then decreased with the increase of P fertilization in 2020 and 2021. When P was applied at more than 90 kg ha⁻¹, the shoot mass of each harvest was greater than that of the control, increased by 39.1~70.1% in 2020; when P was applied at 135 kg ha⁻¹, the shoot mass of the first harvest in 2021 was significantly greater than that of the control, increased by 38.4%; and when P was applied at more than 90 and 135 kg ha⁻¹, the shoot mass of the second harvest in 2021 was greater than that of the control, increased by 85.1% and 93.9%. The shoot mass of the first harvest was greater than that of the second harvest in the two years (Figure 7a). Through the regression equation analysis, we found that the forage yield of 2020 increased with greater shoot mass. The forage yield of 2021 exhibited a consistent quadratic correlation with shoot mass; when the shoot mass was 1.76 and 2.02 g, the alfalfa reached the ideal forage yield (

Table 8. Regression equations, R² values, and p values for the influence of shoot mass on forage yield at each harvest in 2020 and 2021.

Site	Harvest	2020			2021
		Shoot Mass	R2	p Value	Shoot Mass	R2	p Value
Siziwang	H1	y = −3378 + 1378.802x	0.092	0.151	y = −4078 + 293.026x − 83.432x2	0.011	0.892
	H2	y = −92 + 3305.794x	0.442	0.000	y = −1799 + 1283.273x − 317.444x2	0.068	0.478
Tuzzuo	H1	y = −2450 + 1533.601x	0.241	0.015	y = −2850 + 697.666x	0.244	0.014
	H2	y = −1381 + 1379.479x	0.397	0.001	y = 3899 + 649.409x	0.100	0.133
	H3	y = 1664 + 566.934x	0.032	0.403	y = −1521 + 1192.632x	0.163	0.051

).

In Tuzuo County, the shoot mass of the first two harvests in 2020 increased first and then decreased with the P fertilization treatment; when P was applied at 90 kg ha⁻¹, the shoot masses of the first two harvests were 1.81 and 1.45 g, increased by 45.3% and 42.9% relative to the control, respectively, while the shoot mass of the third harvest did not have obvious differences among treatments. The shoot mass of each harvest in 2021 did not have a significant difference between P fertilization and the control (Figure 7b). The regression models developed exhibited a consistent positive linear relationship between the forage yield and shoot mass of each harvest in 2020 and 2021; the forage yield increased with greater shoot mass (Table 8).

3.5. Relative Importance of Yield Components

To better understand the direct effect of each yield component on forage yield, a four-component path analysis (plants m⁻², shoots plant⁻¹, plant height, shoot mass) was performed at each harvest and year. The ratio of path coefficients of the yield components provides an estimate of the relative importance of each component on forage yield [12].

In Siziwang County, path analysis revealed that plants m⁻² did not make a significant contribution to forage yield (p > 0.05) (

Table 9. Path coefficient analysis at each harvest and year in Siziwang County.

Siziwang	H1 a	H2	Year 2020 b	Year 2021
Plants m−2 → yield c	---	---	---	---
Shoots plant−1 → yield c	0.38	---	---	---
Plant height → yield c	0.65	0.46	0.58	0.55
Shoot mass → yield c	---	---	0.44	0.45

^a By each harvest n = 48. ^b By year n = 48. ^c Direct effect of yield components on forage yield. --- Not significant at the 0.05 level.

). In the first harvest, plant height made a greater contribution than shoot mass and shoots plant⁻¹ (ratio 1.71:1 for plant height:shoots plant⁻¹); in the second harvest, only plant height made a greater contribution to forage yield, and others did not make a significant contribution (p > 0.05). With regards to the year, in 2020, plant height made a greater contribution to forage yield than shoot mass (ratio 1.32:1 for plant height:shoot mass); in 2021, plant height made a greater contribution than plants m⁻² (ratio 1.43:1 for plant height:plants m⁻²).

In Tuzuo County, the plants m⁻² did not make a significant contribution to forage yield at each harvest and year either (p > 0.05) (

Table 10. Path coefficient analysis at each harvest and year in Tuzuo County.

Tuzuo	H1 a	H2	H3	Year 2020 b	Year 2021
Plants m−2 → yieldc	---	---	---	---	---
Shoots plant−1 → yield c	---	0.47	---	---	---
Plant height → yield c	0.33	0.48	0.52	0.56	0.36
Shoot mass → yield c	0.29	---	---	0.24	0.43

^a By each harvest n = 48. ^b By year n = 48. ^c Direct effect of yield components on forage yield. --- Not significant at the 0.05 level.

). In the first harvest, only plant height and shoot mass made a contribution to forage yield, and plant height made a greater contribution than shoot mass (ratio 1.14:1 for plant height:shoot mass); in the second harvest, only plant height and shoots plant⁻¹ made contributions to forage yield, and they made an approximately equal impact on forage yield (ratio 1.02:1 for plant height:shoots plant⁻¹); in the third harvest, only plant height contributed to forage yield. With regards to the year, plant height and shoot mass made a contribution to forage yield, and plant height was more than twice as important in defining forage yield compared to shoot mass in 2020 (ratio 2.33:1 for plant height:shoot mass); in 2021, plant height and shoot mass had an approximately equal impact on forage yield (ratio 0.83:1 for plant height:shoot mass).

4. Discussion

4.1. Application of P Fertilizer Affects Winter Survival of Alfalfa

The cold resistance of alfalfa refers to its ability to survive in winter and regenerate in the following Spring [29], which directly affects its persistence and production capacity. Plant nutrition is one management factor to improve alfalfa winter survival. Fertilization with P and K has long been known to improve the persistence and yield of alfalfa [30,31]. Jung and Smith [9] reported that P fertilization was essential for plant survival. Some reports indicated no impact, and occasionally a negative effect of enhanced P and K nutrition on winter survival. Gross et al. [32] found that plant density decreased with P fertilization. Berg et al. [7] also reported that fertilization with P decreased plant population. Sanderson and Jones [23] found that delayed application of P on fall-planted alfalfa did not affect plant density. In our study, despite some reductions in the alfalfa population following emergence, the winter of 2019–2020 was warmer than usual, and the winter survival rates of alfalfa in the two sites were both more than 80%; the effects of P fertilizer on winter survival were obvious in the Tuzuo experiment site, where the winter survival rate of which reached 93.4% at 45 kg ha⁻¹ in 2020, whereas it reached 65.3% at 90 kg ha⁻¹ in 2021. The winter of 2020–2021 was colder than usual, and although some alfalfa plants survived in the first winter, but died in the winter of 2020–2021, the alfalfa under a proper amount of P fertilization had a greater overwintering rate, especially for the Siziwang experiment site, where the maximum winter survival rate in 2020 did not have a significant difference from the control, whereas the maximum winter survival rate in 2021 had a significant difference from the control. The suitable amount of P fertilizer for a greater winter survival rate was 45~90 kg ha⁻¹ in the two sites. The effect of P fertilizer on winter survival was related to other factors, including temperature, soil test P levels, and the content of other mineral elements, which may be responsible for this overwintering difference. The initial P content of the soil was relatively high, and in a warm winter, the addition of P would not significantly affect the overwintering of alfalfa. If more severe stress had been placed on the soils to supply P to the plant, such as done today with exceptionally high yields and the cold winter of 2020–2021, the initial P soil supply may not be sufficient to carry the crop through the winter [12,33]. The other interpretation of the confusion may be that it resulted from inaccuracies associated with estimating plant numbers by counting crowns from aboveground instead of directly digging and counting plants [3].

4.2. Application of P Fertilizer Affects Alfalfa Yield and Yield Components

P is a crucial element for alfalfa growth and agronomic performance [34]. When soil P is limited, plants tend to show a positive growth response to P addition [35]. Berg et al. [7] found that the addition of P fertilizer can increase alfalfa yield and stand persistence. Malhi et al. [27] reported that the alfalfa forage yield increased with P application, but its magnitude of response to added P was lower at Botha than at Lacombe, and the residual effect of large single P application on forage yield lasted at least for five years. However, the addition of P to P-deficient plants stimulated the growth of shoots but not roots [34]. In our study, proper application of P fertilizer increased alfalfa forage yield, and the effects of applying P fertilizer on alfalfa forage yield were significantly different at the two sites. Applying 90 kg P₂O₅ ha⁻¹, the forage yield increased in the two sites; in Siziwang County, the forage yield of 2020 reached 9101.5 kg ha⁻¹, which was an increase of 54.0% relative to the control. Because a large amount plants died in the cold winter of 2020–2021 and the consumption of fertility, the forage yield of 2021 decreased relative to that of 2020; the forage yield under 90 kg ha⁻¹ reached 8386.8 kg ha⁻¹, which was increased by 39.2% relative to the control. To obtain the ideal forage yield, the optimum amount of P₂O₅ application was 108.1 kg ha⁻¹. In Tuzuo County, the forage yield of 2020 under 90 kg P₂O₅ ha⁻¹ reached 13,335.6 kg ha⁻¹, which was an increase of 89.9% relative to the control, and reached 14,287.2 kg ha⁻¹ in 2021, which was an increase of 53.9% relative to the control. The optimum P₂O₅ application rate for alfalfa production was 78.3 kg ha⁻¹ for Tuzuo County. The test P levels of the two soil types were different, and the different contents of other nutrient elements and water conditions in the two sites may also have led to the different demands of P fertilizer for higher forage yield. Insufficient or excessive supply of P fertilizer would reduce the forage yield of alfalfa to some extent, and only appropriate application of P fertilizer can maximize higher yield and the ideal economic benefits [36].

Alfalfa forage yield can be described as the product of three components: plants area⁻¹, shoots plant⁻¹, and mass shoot⁻¹ [17]. Adequate soil P at planting is essential to establishing productive stands of legumes [23]. Supplemental P should be placed according to the native soil fertility and soil P-fixing capacity [37]. Previous reports on the effect of P fertilizer on alfalfa density have had different results. Markus and Battle [30] and Berg et al. [38] found that decreased alfalfa populations were observed in response to enhanced P fertilization. Sanderson and Jones [23] found that the delayed application of P to fall-planted alfalfa did not affect plant density. In a decade-long alfalfa fertility study, stand densities in plots fertilized with both P and K were greater than stand densities of plots fertilized with P alone [39]. In our study, applied P fertilizer improved plant population density in the two sites; however, the amount of P fertilizer required for high plant population in the two sites was different. The differences in the results are related to the initial P levels of soils in various regions. In the two years, plant losses totaled 27 plants m⁻² in Siziwang County and 41 plants m⁻² in Tuzuo County. As the climate and environmental conditions of Tuzuo County during the alfalfa planting period were better than those of Siziwang County, Tuzuo County had more initial plants m⁻². Plants die not only in winter but also during the growing season [38]. As the plant grows, intraspecific competition gradually intensifies later, including competition for water, light, and nutrients. Moreover, the winter-injured plants may survive through the first harvest but subsequently die later in summer [38]. These reasons may lead to Tuzuo County losing more plants m⁻².

Improved plant nutrition is helpful to increase plant persistence. P fertilizer application can increase the number of leaves and stems, promote root development, increase the forage yield, and improve soil fertility [40]. Application of P fertilizer increased the initial plant population, but plants were also lost during the growing period; the magnitude reduction of the plants m⁻² was equal to that of the no fertilization treatment. The decline in the alfalfa population may have resulted from enhanced interplant competition for light, water, and nutrients that eliminated smaller, less vigorous plants from the genetically heterogeneous population of plants that comprise an alfalfa stand. This is because extensive shading and low nutrition content cause preferential death of the weakest plants in the population [41]. Robust, P-responsive plants have greater mass and have more rapid regrowth after harvest, eventually crowding out smaller, less vigorous, slower-growing alfalfa plants [38].

The shoots number of alfalfa is the number of branches derived from the taproot, which are needed for forage yield [42]. Shoots plant⁻¹ is the yield component that is thought to increase as stand density declines to maintain a high yield [43]. The application of P fertilizer can increase the shoots number and yield of alfalfa [44]. In our study, the shoots plant⁻¹ of alfalfa in Siziwang County increased with the P fertilization and shoots plant⁻¹ of the second harvest was larger than that of the first harvest. However, the shoots plant⁻¹ in Tuzuo County also increased with the P fertilization in 2020, while the shoots plant⁻¹ in 2021 did not have an obvious difference between P fertilization treatments. There was a slight increase in shoots plant⁻¹ at the same harvest in the two years. The greater forage yield needed more shoots plant⁻¹, except for the second harvest of 2020 and the third harvest of 2021 in Tuzuo County. Berg et al. [38] found that slight increases in shoots plant⁻¹ occurred as plant populations decreased, but these changes were independent of P fertilization. The shoots plant⁻¹ of alfalfa is determined by genetics, which would be affected by nutrition and other environmental conditions to a certain extent, but it will not continue to increase with the improvement of nutritional conditions [17].

Mass shoot⁻¹ is the most important component of forage yield [18]. Increased mass shoot^–1 has consistently been associated with the improved agronomic performance of alfalfa, and improved soil fertility is a strategy to increase mass shoot⁻¹ [25]. When applying an appropriate amount of P fertilizer, the shoot mass increased in a consistently positive relationship with forage yield. Two components determine mass shoot⁻¹: mass shoot height⁻¹ and height shoot⁻¹. The latter has a more direct impact on forage yield [12]. In our study, shoot mass and plant height both increased when the P application rate increased, and were also linearly correlated with forage yield. Rapid shoot initiation rates after harvest permit shoot regrowth to resume quickly after harvest, resulting in high plant height and shoot mass [45]. Plants receiving P initiated shoot growth quicker, leading to higher plant height and greater shoot mass [25].

5. Conclusions

When alfalfa was planted in soil with mild P deficiency in Inner Mongolia, the winter survival rates and forage yield increased at first and then decreased with the increase of P application; the greater winter survival rates were achieved at 45~90 kg ha⁻¹ of P fertilizer applied, while the greater forage yield was achieved at 90 kg ha⁻¹ in the two sites. The suitable P fertilizer application rates for greater winter survival were less than those for higher forage yield; plant height and shoot mass made a greater contribution to forage yield than other yield components. To pursue a higher forage yield, the suitable P fertilization rates for aeolian sandy soil and silt loam soil were 108.1 and 78.3 kg P₂O₅ ha⁻¹, respectively, based on the regression equations. Therefore, when fertilizing alfalfa production in alpine and soil P deficiency areas, the most appropriate amount of fertilizer should be determined according to climate and soil conditions.