Can Litter (Dead Herbage) Management Affect the Production and Composition of a Desert Steppe Community?

Abstract

To examine the effect of litter on ANPP and species composition (using ground cover) in a Desert Steppe community by removing or adding litter, during plant dormancy, in a single event in either fall or spring. Litter was removed or added in three intensity treatments (heavy, moderate, control—undisturbed) as the main plot and season (fall or spring) as the secondary treatment in a split-plot design with five replications. The experiments were repeated in each of 5 years while three of those were resampled twice. The year effect was analyzed by classifying them into high or low precipitation categories and including those in the statistical model. We found few treatment effects one year after treatment and no persistent effect. Therefore, we focus our examination on the first year only. The total ANPP of individual plant types, or their proportions, were not affected by litter treatment or its interaction with season of treatment or precipitation category. Only the ground cover of selected species was influenced by the treatment. The ground cover of Stipa breviflora was greater with heavy litter removal in fall but unaffected by litter removal in spring while Neopallasia pectinata had a greater cover with moderate or heavy removal in years when precipitation was low. Litter addition resulted in a greater ground cover of Neopallasia pectinata and reduced the cover of Convolvulus ammannii in years of low precipitation. The marginal effectiveness of litter treatments on the plan community in the Desert Steppe suggests that it need not be a factor for consideration in grazing management.

1. Introduction

Litter is a component of terrestrial ecosystem and it plays an important role in plant productivity and abundance [1,2]. Most current research on litter focuses on its decomposition processes and effects on soil nutrients [3,4], as well as the relationship between soil microorganisms and litter addition or removal [5,6]. The dead herbage in grasslands has been referred to as a litter. It is herbage that has escaped herbivores and has senesced. As such, its palatability has severely deteriorated for most livestock, other mammals, and insects. Therefore, it is avoided by herbivores and will accumulate unless grazing pressure is increased and its loss is primarily dependent on physical weathering or microbia [7] and photo-degradation [8]. On the fescue grassland, biomass loss over the dormant season (from fall to spring) was 24, 43, and 56% on pastures that had been grazed light, moderate, and heavy for about 40 years [9], which appeared to reflect the characteristics of vegetation that resulted from the long-term grazing treatment. Over a similar period, weight loss in leaves of Altai wildrye (Leymus angustus [Trin.] Pilg.) was 41 and 52% in two different years [10]. Nevertheless, livestock has the greatest effect on the amount of litter present on grazed grasslands. Changes in litter over time can also affect soil carbon cycling processes [11].

While litter will intercept light rainfall [12] which reduces the amount reaching the soil, it can significantly improve soil moisture status by capturing snow in winter [13] and reducing losses from evaporation and surface runoff. Litter is an important ecosystem process, serving as a natural nutrient conduit and a major pathway to supplement available nutrients in plants [14]. This may result in a 60% increase in aboveground net primary production (ANPP) on a Mixed Grass Prairie although the effect is linked to site and year [15,16], which dictates the extent that soil moisture limits production. Litter had a more significant role in regulating soil moisture in dry years and areas than in years and areas with adequate precipitation [17]. Although water deficits are a feature of grasslands, in some years moisture may be sufficient so that it does not constrain ANPP while in years of extreme soil moisture deficits, there may not be sufficient water available for conservation to be effective [16].

The effectiveness of litter for enhancing ANPP appeared to be considerably less on a Typical Steppe in Inner Mongolia [18] that was dominated by Stipa grandis P. A. Smirn and Leymus chinensis (Trin.) Tzvel, than in the Mixedgrass Prairie [16] or Parkland [15]. Here, the primary effects of removing litter were to reduce soil moisture and grass biomass, in particular that of L. chinensis, a rhizomatous species that is sensitive to soil moisture [19]. While this trend is consequential for livestock management, total ANPP was not affected [18].

The community response to litter is a function of its species’ composition and adaptation to their environment. Community composition and its productivity will be affected by litter treatment [20]. The vegetation in the Desert Steppe of Inner Mongolia consists of drought-tolerant species that include bunchgrasses (Stipa breviflora Griseb., Cleistogenes songorica (Roshev.) Ohwi), annual forbs (Convolvulus ammannii Desr., Heteropappus altaicus (Willd.) Novopokr., Neopallasia petinata (Pall.) Poljak., Kochia prostrata (L.) Schrad.), and shrubs (Caragana stenophylla Pojark.). When taken together with their tolerance to grazing, it suggests that the importance of litter may not be significant to their productivity. At present, the relevant studies on litters mainly focus on their accumulation, decomposition dynamics and microbiome [3,11,21], and there are relatively few studies on the effects of litters on grassland, especially in desert steppe which is most sensitive to external interference.

Desert steppe is one of the main grassland types in the central and western Inner Mongolia Autonomous Region, accounting for 39% of the grassland area. Stipa breviflora desert steppe occupies 11.2% of the total area of warm desert steppe [22], which is an important type of desert steppe in Inner Mongolia and plays an important role in the economic development of animal husbandry and ecological services [23]. Therefore, we conducted a study to examine the effect of removing or adding litter, during plant dormancy, in a single event on a plant community in the Desert Steppe, so as to more accurately understand the role of litters in the desert steppe ecosystem and provide data support for rational utilization of Stipa breviflora grassland. Therefore, we conducted research to examine the effect of removing or adding litter, during plant dormancy, in a single event on a plant community in the Desert Steppe. We tested the hypothesis: (1) litter removal reduced ANPP while litter addition increased ANPP and (2) these effects were influenced by the season of treatment, whether fall or spring.

2. Materials and Methods

2.1. Site Description

The study was conducted in a Desert Steppe at the field station of the Inner Mongolia Academy of Agriculture and Animal Husbandry Science (lat 41°47′17″ N, long 111°53′46″ E), which is located in the in the Siziwang Banner, Inner Mongolia, China. The average altitude is 1456 m above sea level. The climate of the area is classified as Temperate Continental Monsoon Climate with average annual precipitation of 250 mm with 80% occurring between June and September. Precipitation in winter is low and any snow tends to sublimate so that it does not contribute to soil moisture [20]. Spring is often windy, winters are cold and dry, and summers are hot and wet. The soils are classified as Light Chestnut and the dominant plant species include Stipa breviflora, Artemisia frigida Willd. and Cleistogenes songorica. The soil in the study site is light chestnut soil with a thickness of about 1 m, but calcium accumulation occurs in 50–60 cm in most places. The soil is hard and the permeability is poor.

The study area had historically been used for grazing sheep and goats from 1950 to 1990 with a significant number of cattle [24]. Livestock numbers increased from about 0.3 Sheep-Equivalent Units (SE)/ha in 1949 to about 1 SE/ha in 1963, where it stabilized [24].

Litter removal and addition treatments were carried out on two neighboring sites: The removal treatment was conducted in a livestock exclusion that had been erected in 1996 while the addition treatment was conducted on a grazed site. Further grazing on the latter site was controlled with fencing (Figure 1).

2.2. Experimental Method

Litter was removed or added in two separate experiments. Each experiment consisted of a split-plot design with litter removal or addition treatments (heavy, moderate, control—undisturbed) as the main plot (2 m × 6 m) and season (fall, spring) as the secondary plot (2 m × 3 m). The litter treatments were conducted once in five consecutive treatment-years from fall, 2016 to 2020 and spring, 2017 to 2021. There were 5 replications with one treatment in each treatment-year. The vegetation response was measured at peak standing crop in the first growing year after treatment (2017 to 2021) and for up to 2 additional growing years to observe the post-treatment response. The first 3 treatment-years were sampled a total of 3 times, the 4th a total of 2 times and the 5th only once.

Growing season precipitation (May to September) over the five-year period could be classified into two categories as low, in 2017 (130 mm) and 2021 (130 mm), and high, in 2018 (210 mm), 2019 (207 mm) and 2020 (200 mm). The proportion of annual precipitation occurring during the growing season was 0.80, 0.91, 0.83, 0.76 and 0.85 from 2017 to 2021, respectively (Figure 2).

The heavy litter removal treatments were applied by harvesting all litter, which includes the fallen (fragmented) and standing (attached to the crown) dead herbage by clipping the plants at 2 cm height and raking the material while the moderate removal treatment involved only raking. The removed litter was taken by the plot and then applied evenly by hand to the assigned plot in the litter addition experiment. Therefore, the identical amount harvested from the heavy or moderate removal treatments contributed to the comparable heavy or moderate addition plots, respectively (Figure 3). The mass of litter removed from the moderate and heavy treatments was about 40 and 69 g·m⁻², respectively. Therefore, the moderate treatment removed about 58% as much as the heavy treatment. As the heavy removal treatment removed all standing and fallen litter from the plots, this approximated the litter mass in the control. The litter addition experiment was conducted on a grazed site where litter represented only trace amounts.

2.3. Vegetation Sampling

ANPP was sampled at peak standing crop in early August by clipping plants of each species at 2 cm height in two 20 cm × 50 cm quadrats within each plot. Standing litter was removed by hand at the time of sampling and combined with fallen litter that was sampled by raking. Raking removed most fallen litter that was greater than 2 cm in length. All the samples were bagged, brought back to the laboratory, dried at 65 °C for 48 h and weighed by species. Total ANPP was determined by summing the contribution of all species. Plant height, ground cover, and density were also sampled at the same time. Plant height and cover were determined by species in each quadrat before harvesting. Plant height was estimated by measuring the average leaf length using a meter stick. Ground cover was determined by visually estimating leaf area (%) using a marked grid at classes of 1% to 5%, 5% to 25% and 25% classes to 100%. Plant densities were determined by counting the number of plants of each species at the time that plants were being harvested.

2.4. Statistical Analysis

The effect of litter removal or addition on plant variables (ANPP, height, ground cover, and density) were tested by dominant species, totals within functional groups, and totals within the community. The functional groups were rhizomatous grass, bunch grass, forbs and annual species. Tests were also conducted on the biomass proportions of perennial to annual species as well as of grass to forbs.

Data were analyzed in two split-plot mixed models: a whole model that included repeated sampling for three growing years after treatment to test post-treatment response and a partial model of only the first growing year after treatment. This was done to balance the whole model with only the first three treatment-years that were sampled successively in three growing years while the partial model tested only the first growing year after treatment but included five treatment-years. The litter removal and litter addition treatments were analyzed separately.

The factors included in the whole model were precipitation (high, low), growing years (1, 2 or 3), season (fall, spring) and three litter treatments (either removal or addition at high, moderate, or control—undisturbed) and their interactions. These factors, with the exception of growing years, were included in the partial model. Analysis of variance was performed using the Mixed Procedure (SAS, Version 9.1.3, SAS Institute Inc., Cary, NC, USA) after normalizing (Shapiro–Wilk test) the data by examining the distribution of residuals and either removing outliers or applying a transformation. The effects of litter treatment, season, precipitation and post-treatment response (in whole model) were treated as fixed effects, with block nested in treatment-year. The interaction of the block (random effect) with the main effects produced the error terms. Post-treatment years were analyzed as a repeated measure and its residuals were tested with four co-variance structure matrices (autoregressive, heterogeneous autoregressive, compound symmetry and heterogeneous compound symmetry) using the Akaike’s information criterion to select the best structure. Treatment means were separated using Fisher’s protected LSD_0.05.

Species differences (by weight) among a priori groups were examined using indicator species analysis with PCORD (Version 5, MjM Software, Corvallis, OR, USA). This analysis calculates the proportional abundance of a species in a group relative to its abundance in all groups, and the proportional frequency of the species in each group, which are then combined by multiplication to yield an indicator value. This value is then evaluated with a Monte Carlo method [25]. For this analysis, 4999 randomizations in the Monte Carlo test were used. Several comparisons were made, for the first year effects only, using indicator species analysis: (1) among treatments by years at each site combined to determine whether litter removal or addition affected species composition (based on mass), and (2) between precipitation classes of the years combined that were included in each for both litter addition or removal sites. Summary data of species richness and diversity (Shannon’s diversity index [25]) were calculated using PC-ORD by the same groups.

3. Results

The persistent effect of litter removal or addition over 3 years after treatment (growing years), or its interaction with precipitation, season of application or litter application treatment was not significant (p > 0.05). Therefore, we focus on the first year effect when the most dramatic changes might be expected.

Vegetation ground cover over the 5-yr study period on both the litter removal and litter addition sites was unaffected (p > 0.05) by the litter treatment. Most covers were represented by forbs (23%) and a minor proportion by shrubs (<1%;

Table 1. Ground cover of primary plant species of the undisturbed (control) plots on the Desert Steppe of Inner Mongolia over the period from 2007 to 2011 (n = 5).

Species	Mean (%) 1	SD 1
Graminoides
Cleistogenes songorica (Roshevitz) Ohwi C4	3.43	3.11
Stipa breviflora Griseb.	2.42	2.37
Agropyron cristatum (L.) Gaertn.	0.16	0.54
Leymus chinensis (Trin.) Tzvelev	0.28	0.93
Carex duriuscula C.A. Mey.	0.01	0.08
Total Graminoides	6.30	5.05
Forbs
Perennial
Artemisia frigida Willd.	8.79	6.77
Convolvulus ammannii Desr.	4.97	1.92
Allium tenuissimum L.	0.51	1.08
Bassia prostrata (L.) A.J. Scott	0.57	1.27
Lagochilus ilicifolius Bunge ex Benth.	1.10	1.82
Allium mongolicum Regel.	0.02	0.16
Krascheninnikovia ceratoides (L.) Gueldenst.	0.40	1.36
Potentilla bifurca L.	0.02	0.16
Iris tenuifolia Pall.	0.01	0.04
Cymbaria dahurica L.	0.01	0.08
Heteropappus altaicus (Willd.) Novopokr.	0.07	0.40
Annual
Salsola collina Pall. C4	4.90	11.35
Neopallasia pectinata (Pall.) Poljak.	1.43	2.69
Total Forbs	23.09	12.00
Shrubs
Caragana stenophylla Pojark.	0.13	0.51
Caragana microphylla Lam.	0.16	1.01
Total Cover	29.67	12.22

¹ Means and standard deviations are calculated across all plots and years.

). Of the forbs, Artemisia frigida and Convolvulus ammannii accounted for a combined total of 14% ground cover while Cleistogenes songorica and Stipa breviflora accounted for a combined total of about 6% (Table 1). Ground cover on the litter addition site was represented by similar species but total ground cover was less (23.9 vs. 29.7%), which reflected a diminished (about 50%) cover of both Artemisia frigida and Salsola collina.

Litter mass sampled in the first growing season after treatment represents the residual of the mass that was present after treatment and is the net mass after weathering losses. Litter mass in the removal experiment had accumulated as a result of its protection from grazing while litter in the addition experiment was nearly absent because of its history of heavy grazing immediately before the trial. In the removal experiment, the amount of litter present at the time of sampling after the first growing season did not reflect the treatments (p > 0.05; 1.46, 1.39 and 1.06 g·0.1 m² in the control, moderate, and heavy removal treatments, respectively) where the mass of litter remaining after the first growing season was affected only by season (

Table 2. Response of Annual Net Primary Production (ANPP) of selected plant types and proportions to litter removal after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) ¹, season (Spring, Fall) and intensity (Control, Moderate, High).

	ANPP					Proportion
Effect	Total	Grass	Forbs	Saco	Litter	Grass:Forbs	Peren.:Ann.
	Probability
Precipitation 1 (P)	<0.01	>0.05	<0.01	<0.01	>0.05	>0.05	0.05
Season (S)	>0.05	>0.05	>0.05	>0.05	0.01	>0.05	>0.05
P × S	>0.05	>0.05	>0.05	>0.05	>0.05	0.02	0.04
Intensity (I)	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05

¹ May to September precipitation: High in 2008, 2009 and 2010 of 210, 207, and 200 mm, respectively; Low in 2007 and 2011 of 130 mm in each year.

and

Table 3. ANPP response to litter addition after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) ¹ and season (Spring, Fall).

	Biomass					Proportion
Effect	Total	Grass	Forbs	Saco 1	Litter	Grass:Forbs	Peren.:Ann.
	Probability
Precipitation 2 (P)	0.01	>0.05	0.01	<0.01	0.01	>0.05	<0.01
Season (S)	>0.05	>0.05	>0.05	0.01	<0.01	0.05	<0.01
P × S	>0.05	>0.05	>0.05	<0.01	0.02	>0.05	0.05
Intensity (I)	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × I	>0.05	>0.05	>0.05	>0.05	0.02	>0.05	>0.05
S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × S × I	>0.05	>0.05	>0.05	>0.05	0.04	>0.05	>0.05

¹ Saco, Salsola collina, the same below. ² May to September precipitation: High in 2008, 2009 and 2010 of 210, 207, and 200 mm, respectively; Low in 2007 and 2011 of 130 mm in each year.

), with greater amounts present in the fall than in the spring treatment plots. In the litter addition experiment, the mass was greater (p < 0.05) in spring than in fall and in years of high precipitation (

Table 4. The response of plant ground cover (%) of selected species to litter removal after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) ¹, season (Spring, Fall) and intensity (Control, Moderate, High).

	Ground Cover (%)
	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Probablity
Precipitation 2 (P)	0.01	0.03	>0.05	<0.01	>0.05	0.01	<0.01
Season (S)	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × S	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
Intensity (I)	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × I	>0.05	>0.05	>0.05	>0.05	>0.05	0.05	>0.05
S × I	>0.05	<0.01	>0.05	>0.05	>0.05	>0.05	>0.05
P × S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. ² May to September precipitation: High in 2018, 2019 and 2020 of 210, 207, and 200 mm, respectively; Low in 2017 and 2021 of 130 mm in each year.

and

Table 5. The mean response of plant ground cover by selected species to litter removal after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) and season (Spring, Fall).

	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Ground Cover (%)
Precipitation (P)
High	2.4 a	1.9 a	8.3	5.0 b	0.9	0.4 a	13.3 b
Low	4.5 b	2.4 b	8.3	3.0 a	0.7	2.9 b	1.5 a
	0.4	0.2	0.9	0.5	0.2	0.2	0.8
Season
Fall	3.3	2.0	8.4	4.0	0.6	1.4	7.0
Spring	3.6	2.3	8.2	4.0	0.9	1.8	7.8
	0.4	0.2	0.9	0.5	0.2	0.2	0.9
P × S
Low—Fall	4.7	2.4	7.7	3.5	0.6	2.7	1.3
Low—Spring	4.3	2.5	8.8	2.4	0.8	3.1	1.7
High—Fall	1.9	1.6	9.0	4.5	0.8	0.2	12.6
High—Spring	2.8	2.1	7.6	5.6	1.0	0.6	13.9
	0.5	0.3	1.2	0.7	0.2	0.3	1.1

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. The same superscript lowercase letter means no significant difference (p > 0.05).

). The level of litter addition was only affected by its interaction with precipitation (Table 4). In years of low precipitation it was 0a, 0.2ab and 0.14bc g·0.1 m² in the control, moderate and heavy application treatments, respectively, while in years of high precipitation, the litter mass was 0.09ab, 0.27c, and 0.10ab g·0.1 m², respectively.

There was no significant effect on total ANPP and that of grass, forbs or Salsola collina (Table 2 and Figure 4) in both removal and addition treatments, in the first year after treatment. The effects of precipitation in the growing season (defined by classification values as a high ($\overline{\mathrm{x}}$ = 203 mm) or low ($\overline{\mathrm{x}}$ = 130 mm) annual total) in the removal and addition treatments were most significant and the consistent effects on the variables measured were similar. In the litter removal experiment, total ANPP in high precipitation years was about 40% greater than in low precipitation years, with most contributed by a greater yield of Salsola collina (Figure 4). In the litter addition treatments, the differences were smaller but also contributed mostly by the response of Salsola collina (Figure 5).

The effect of the season that litter was removed was mostly to modify (p < 0.05) the effects of precipitation on the ratio of grass:forbs and perennial:annual species (Table 2) while the season of litter addition modified (p < 0.05) the effects of precipitation on the ANPP of Salsola collina and the ratio of perennial:annual plants (Table 3). The ratio of grass:forbs was greater (p < 0.05) with litter removal in spring than in fall when precipitation was high, but similar (p > 0.05) between seasons when precipitation was low (Table 2). The ratio of perennial:annual species with litter removal was similar (p > 0.05) between seasons within a precipitation category but it was greater (p < 0.05) in spring when precipitation was low than when it was high (Table 2 and Figure 4).

The season of litter addition had no effect (p > 0.05) on the ANPP of Salsola collina when precipitation was low but in years of high precipitation the effect was greater (p > 0.05) with addition in fall rather than spring (Table 3 and Figure 5). Within a precipitation category, litter addition in spring resulted in a higher ratio of perennial:annual plants than in fall (Figure 4).

In both the litter removal and addition experiments, the ground cover of most species selected for examination was not affected (p > 0.05) by litter treatment but primarily by precipitation (Table 4 and

Table 6. The response of plant ground cover of selected species to litter addition after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) ¹, season (Spring, Fall) and intensity (Control, Moderate, High).

	Ground Cover (%)
	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Probability
Precipitation 2 (P)	>0.05	0.04	>0.05	<0.01	>0.05	<0.01	<0.01
Season (S)	>0.05	>0.05	>0.05	>0.05	>0.05	0.01	0.03
P × S	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	<0.01
Intensity (I)	>0.05	>0.05	>0.05	<0.01	>0.05	<0.01	>0.05
P × I	>0.05	>0.05	>0.05	<0.01	>0.05	>0.05	>0.05
S × I	>0.05	>0.05	>0.05	0.04	>0.05	>0.05	>0.05
P × S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. ² May to September precipitation: High in 2018, 2019 and 2020 of 210, 207, and 200 mm, respectively; Low in 2017 and 2021 of 130 mm in each year.

). Nevertheless, in the removal experiment, the ground cover of Stipa breviflora was affected (p < 0.05) by the interaction of litter removal intensity and season of treatment (Table 4 and Table 5). In fall, the percent ground cover of Stipa breviflora was 1.6a, 1.6a and 2.8b% (means with different superscript letters are different, p < 0.05) in the control, moderate and heavy removal treatments, respectively, and in spring the ground cover was 2.6b, 2.4b, and 1.8b%, respectively, for the same removal treatments. The ground cover of Neopallasia pectinata was affected (p < 0.05) by the interaction of litter removal and precipitation. In years of low precipitation its percent ground cover was 2.0b, 3.3c, and 3.4c% in the control, moderate and heavy removal treatments, respectively, while in years of high precipitation the cover was, 0.4a, 0.2a, and 0.4a%, respectively, for the same removal treatments.

In the litter addition experiment, the ground cover of Neopallasia pectinata was affected (p < 0.05) with litter addition (1.1a, 2.1b and 1.5ab%; means with different superscript letters are different, in the control, moderate and heavy addition treatments, respectively). The cover of Neopallasia pectinata and Salsola collina was greater (p < 0.05) when litter was added in the fall than in the spring (Table 6 and

Table 7. The mean response of plant ground cover by selected species to litter addition after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) ¹ and season (Spring, Fall).

	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Ground Cover (%)
Precipitation (P)
High	5.8	3.1 a	4.3	2.1 a	0.3	0.9 a	6.0 b
Low	5.7	4.1 b	4.9	3.5 b	0.4	2.3 b	0.8 a
	0.4	0.4	0.3	0.2	0.1	0.2	0.6
Season
Fall	5.4	3.8	4.3	2.8	0.4	1.9 b	3.9 b
Spring	6.2	3.3	4.8	2.7	0.3	1.2 a	2.8 a
	0.4	0.4	0.3	0.2	0.1	0.2	0.6
P × S
Low—Fall	5.7	3.7	4.7	3.4	0.4	2.5	0.6 a
Low—Spring	5.7	2.5	5.1	3.5	0.3	2.0	1.0 a
High—Fall	5.1	4.0	4.0	2.2	0.3	1.3	7.2 c
High—Spring	6.6	4.2	4.6	1.9	0.2	0.5	4.7 b
	0.6	0.5	0.4	0.3	0.1	0.2	0.6

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. The same superscript lowercase letter means no significant difference (p > 0.05).

). Also, litter addition reduced (p < 0.05) the cover of Convolvulus ammannii although this effect was only evident in years of low precipitation when the cover was 4.9b, 2.8a and 2.7a% in the control, moderate and heavy addition treatments, respectively, compared with 2.2a, 2.1a and 1.9a% in the same treatments, respectively.

Precipitation had an inconsistent effect on plant height in both litter removal and addition experiments (

Table 8. The response of plant height by selected species to litter removal after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) ¹, season (Spring, Fall) and intensity (Control, Moderate, High).

	Plant Height
	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Probability
Precipitation 2 (P)	0.02	<0.01	>0.05	<0.01	>0.05	<0.01	<0.01
Season (S)	0.02	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × S	0.05	>0.05	>0.05	>0.05	>0.05	>0.05	<0.01
Intensity (I)	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × I	>0.05	0.02	>0.05	>0.05	>0.05	>0.05	>0.05
S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. ² May to September precipitation: High in 2018, 2019 and 2020 of 210, 207, and 200 mm, respectively; Low in 2017 and 2021 of 130 mm in each year.

and

Table 9. The response of plant height by selected species to litter addition after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) ¹, season (Spring, Fall) and intensity (Control, Moderate, High).

	Plant Height
	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Probability
Precipitation 2 (P)	<0.01	<0.01	<0.01	>0.05	>0.05	<0.01	<0.01
Season (S)	0.02	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × S	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
Intensity (I)	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
S × I	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05	>0.05
P × S × I	>0.05	0.02	0.03	>0.05	>0.05	>0.05	>0.05

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. ² May to September precipitation: High in 2018, 2019 and 2020 of 210, 207, and 200 mm, respectively; Low in 2017 and 2021 of 130 mm in each year.

). In the litter removal experiment years of high precipitation resulted in taller plants of Stipa breviflora and Salsola collina but Cleistogenes songorica and Neopallasia pectinata were shorter (p < 0.05;

Table 10. The mean response of plant height by selected species to litter removal after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) and season (Spring, Fall).

	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Plant height (cm)
Precipitation (P)
High	3.7 a	7.5 b	6.7	4.5	6.4	5.3 a	11.6 b
Low	4.3 b	5.2 a	7.1	3.3	5.9	14.8 b	7.6 a
	0.3	0.3	0.7	0.1	0.5	0.8	0.5
Season
Fall	4.4 b	6.3	7.0	4.1	6.5	10.4	10.5
Spring	3.7 a	6.4	6.8	3.8	5.7	9.7	8.8
	0.3	0.3	0.7	0.1	0.7	0.9	0.5
P × S
Low—Fall	4.8 b	5.2	6.7	3.4	6.3	15.6	7.2 a
Low—Spring	3.7 a	5.2	7.5	3.2	5.4	14.0	8.1 b
High—Fall	3.9 a	7.3	7.2	4.6	6.7	5.2	13.9 c
High—Spring	3.6 a	7.6	6.2	4.3	6.1	5.4	9.4 b
	0.3	0.4	0.8	0.2	0.7	1.2	0.7

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. The same superscript lowercase letter means no significant difference (p > 0.05).

) while in the litter addition experiment both grasses were taller but the affected forbs were shorter (p < 0.05;

Table 11. The mean response of plant height by selected species to litter addition after the first growing season following treatment on the Desert Steppe in Inner Mongolia, in relation to precipitation category (High, Low) and season (Spring, Fall).

	Grass 1		Forbs 1
Effect	Clso	Stbr	Arfr	Coam	Lail	Nepe	Saco
	Plant height (cm)
Precipitation (P)
High	3.5 b	6.3 b	3.2 a	2.7	4.4	4.0 a	6.7 a
Low	2.6 a	3.5 a	4.9 b	2.6	3.7	14.1 b	9.2 b
	0.1	0.3	0.3	0.1	0.4	0.5	0.4
Season
Fall	3.2 b	5.1	4.1	2.7	3.9	9.4	8.1
Spring	2.8 a	4.7	4.2	2.7	4.2	8.7	7.8
	0.1	0.3	0.3	0.1	0.4	0.5	0.4
P × S
Low—Fall	2.7	3.6	5.2	2.6	3.8	14.9	9.2
Low—Spring	2.4	3.4	4.6	2.7	3.7	13.2	9.2
High—Fall	3.6	6.6	3.0	2.7	4.0	3.8	7.0
High—Spring	3.3	6.0	3.5	2.6	4.8	4.2	6.4
	0.1	0.3	0.4	0.1	0.5	0.6	0.5

¹ Species codes: Clso, Cleistogenes songorica; Stbr, Stipa breviflora; Arfr, Artemisia frigida; Coam, Convolvulus ammannii; Lail, Lagochilus ilicifolius; Nepe, Neopallasia pectinata; Saco, Salsola collina. The same superscript lowercase letter means no significant difference (p > 0.05).

). Fall litter removal resulted in taller plants of Cleistogenes songorica (p < 0.05) but its effect was modified by precipitation (Table 8 and Table 10). The height of Salsola collina was also affected by season and precipitation (Table 8) while the height of Stipa breviflora was influenced by the interaction of litter removal and precipitation. Here, the height of Stipa breviflora was greatest with heavy removal in years of low precipitation (4.5a, 4.6a and 6.5b cm, respectively, in the Control, Moderate and Heavy removal treatments; means having different letters are different, p < 0.05) but similar (p > 0.05) among treatments in years of high precipitation (8.0, 7.3 and 7.2 cm, respectively).

The significant (p < 0.05) 3-way interaction of precipitation, season and litter addition for Stipa breviflora appeared to be caused primarily by taller plants (p < 0.05) produced with high precipitation and heavy litter addition in fall (7.2 cm) than in spring (5.6 cm) for the same conditions. For Artemisia frigida the effect appeared to be produced by taller plants when precipitation was high and with heavy litter addition in spring (4.5 cm) compared with low precipitation and heavy litter addition in spring (5.9 cm).

Indicator species analyses of litter removal (based on species mass) showed that only Krascheninnikovia ceratoides provided an indication of treatment effect [Indicator value (IV) = 15.5; p = 0.04]. Its relative abundance was 13%, 22% and 65% for the control, moderate and heavy litter removal treatments, respectively. A further examination of this species using analysis of variance also indicated a significant (p = 0.01) response to litter removal (0.3%, 0.2%, and 1.2%, for the same treatments, respectively) with a suggestion that its cover was influenced by precipitation (p = 0.07) where it exhibited similar ranking among treatments in both years but was greater (p = 0.02) in dry years ($\overline{\mathrm{x}}$ = 1.0% and 0.1%, in dry and wet years, respectively). Species that distinguished between low and high precipitation years were: Cleistogenes songorica (IV = 55.4, p = 0.01), Convolvulus ammannii (IV = 60.7; p ≤ 0.01), Salsola collina (IV = 62.7; p < 0.01), Neopallasia pectinata (IV = 59.5; p ≤ 0.01), Bassia prostrata (IV = 33.1; p < 0.01) and Agropyron cristatum (IV = 13.6; p = 0.01). Their relative abundance, in the low and high precipitation years were, respectively for each species: 61 and 39, 36 and 64, 11 and 89, 90 and 10, 64 and 36, and 0 and 100.

In the litter addition experiment, there were no species that provided significant (p < 0.05) indication of litter addition. Species that were an indicator for low and high precipitation years were: Salsola collina (IV = 74.8; p < 0.01), Neopallasia pectinata (IV = 44.3; p ≤ 0.01), Allium tenuissimum (IV = 15.9; p = 0.05), Agropyron cristatum (IV = 12.2; p = 0.05), Leymus chinensis (20.9; p = 0.02); and Artemisia commutata Bess. (IV = 32.5; p < 0.01). Their relative abundances in years of low and high precipitation were, respectively, 13 and 87, 71 and 29, 65 and 35, 0 and 100, 74 and 26, and 91 and 9.

4. Discussion

The effects of a single time of litter removal or addition on the short-term effects on vegetation expressed by ANPP, ground cover and plant height were examined. Litter removal or addition were both expected to affect the microclimate of the soil [26], which were evaluated by measuring the plant response. While both experiments would have impacted the soil microclimate, the response of the plant community could be expected to differ because in one case litter was removed from a community that had been protected from grazing for 11 years while in the second experiment, litter was added to a grazed site. Nevertheless, the same major species were present at both sites and in similar composition as estimated by canopy cover. Therefore, the primary distinction between them was the mass of litter and their total ground cover.

As in all arid regions, the principal driver of production and species dynamics in the Desert Steppe is precipitation, which in itself is hardly noteworthy, while the effect of the season that litter was applied is obscured by the fact that it included treatments ranging from a control to heavy litter removal or addition. It has been shown that increasing precipitation under the treatment of added litter significantly increased grassland biomass [27], and the same conclusion was obtained in semi-arid grasslands, where the increase in litter and precipitation significantly increased community composition [28]. Therefore, it is the interactions of precipitation or season with litter application treatments that are the most revealing. We anticipated that the effect that season had on litter treatments would be expressed through litter mass, whether left in situ or added, on snow capture over winter [13], or by its influence on the loss of soil moisture in either winter or spring. The effect of season was most demonstrable with Salsola collina where only litter addition in fall increased its mass and cover, and only in years of high precipitation. Nevertheless, this does not explain the lack of a significant interaction of season with the litter treatment.

The presence of litter in grassland serves to moderate the soil environment and its removal exposes it to greater extremes of temperature, evaporation and, therefore, soil moisture [16]. And soil moisture is the main factor affecting the community structure of desert grassland [29]. Consequently, the community response to litter mass is emulated by the precipitation regime, which can provide some clues to how the community might respond to a more favorable moisture regime. Nevertheless, the opportunity for litter to modify the soil environment is dependent on its mass and the occurrence of soil moisture. In the present study, the mass of litter present was very small (14.6 g·m⁻²) compared with the Typical Steppe in Inner Mongolia (140 g·m⁻² [18]) or the Mixed Prairie in Canada (1171 g·m⁻² [15]). Willms et al. (1993) found that litter had no effect on ANPP in a year with above-average precipitation and had a decreased effect in a year with under-average precipitation; while in both studies by Willms et al. (1993) and Wang et al. (2011) after litter removal, soil heat units were the greatest.

Therefore, there are several factors that may have mitigated the treatment differences in the present study. Perhaps the more likely factor is that the effect of a small litter mass is below a threshold effect that can be detected. In the study by Willms et al. (1993), the year with the least amount of rainfall had 227 mm which was marginally greater than the average of 203 mm in the present study. In the previous study, the coefficient describing the relationship of litter mass to ANPP was 0.114. When applied to the average litter in the control of the present study (14.6 g·m⁻²), the effect on yields would be about 2 g·m⁻². Another factor that would mitigate the effect of litter is that water conservation is irrelevant because a majority of annual precipitation (85%) occurs in the growing season (May to September) when the rainfall would be quickly used by plants.

Our inability to detect differences in litter mass between removal or addition treatments after the first growing season is not only an indication of the high variability among plots and the relatively reduced effectiveness of litter in the Desert Steppe but is also exacerbated by a high rate of litter decomposition that would further reduce treatment differences. While an average of 69 g·m⁻² were removed from the high removal treatment, which is an estimate of the total amount of litter in that site, only about 15 g·m⁻² were sampled in the control plots the next summer, which suggests a loss of about 78%, presumably the result of weathering and herbivory. The dry matter losses on the addition site were considerably greater, amounting to virtually all measurable litter. Willms et al. (1996) reported average losses of 24%, 43%, and 56% from fall to spring from ungrazed, moderately grazed, and heavily grazed fescue grassland, community, respectively, suggesting that losses were determined by the species contributing to the biomass [9]. In the Desert Steppe, a large proportion of biomass consists of annual plants, which are highly susceptible to weathering and fragmentation. This would be particularly pronounced where litter had been added because it had already been detached and disturbed by raking and application.

The loss of litter after treatment means that any apparent effectiveness will also degrade over time, and its effect on production will depend on the amount of moisture in the soil when the temperature becomes favorable for growth in spring [30]. The presence of litter can affect snow capture as well as water conservation in spring and summer. However, on average, over 80% of precipitation occurred during the growing season from May to September and very little as snow, which would diminish this potential.

The short-term relationship between litter quantity and yield is related to its water retention capacity, while evidence [7] shows that species composition may also be affected. where litter is removed over a longer period of time. Few effects of litter during the first growing season after treatment, and no lasting effect after that, thereafter, indicated a high degree of resilience of plant communities and dominant individual species. Repeated treatment of litter may produce detectable changes in species composition by allowing plants to adapt to new soil moisture, but this possibility is hypothetical. A grazing study in desert steppes suggested that this could happen, but it hypothesized that grazing and litter removal would have a similar effect in arid environments. Li et al. (2008) reported no effects of grazing or litter mass on the dominant species or ANPP in a Desert Steppe after 4 to 6 years of treatment even though litter mass declined from 66 to 16 g·m² with heavy grazing intensity from an ungrazed control [31]. An examination of the canopy cover of individual species confirms their relative stability in relation to litter mass with only a few exceptions that include both perennial and annual species in both litter removal and addition treatments.

Measures of the canopy cover of individual species is a more rapid estimate of abundance than ANPP and subject to fewer potential errors during the harvesting to weighing stages because it is measured in situ. Therefore it provides useful information on the individual species’ response to disturbances, which litter removal and addition represent. We observed very few examples of species responding to these treatments but they indicate that the annual forb, Neopallasia pectinata, tended to favor greater masses of litter while perennial species favored less litter. The effect of the proportion of different species in the litter is not excluded [4]. Litter may provide Neopallasia pectinata a more favorable environment for seed germination in years of low precipitation, which was facilitated with litter addition in one experiment. The litter addition experiment included the potential process of introducing seed with the litter, which may have increased the cover of Neopallasia pectinata. However, this was not observed with Salsola collina, a more abundant annual forb, which exhibited an insignificant increase of 0.5% with litter addition. As succession progresses, litter accumulation can alter seed dispersal patterns, which may have implications for the long-term dynamics of plant populations [32].

The smaller canopy cover of the perennial species, Stipa breviflora, Convolvulus ammannii and Krascheninnikovia ceratoides to reduced litter mass may indicate sensitivity to shading, although that explanation was not tested. It seems supported by the observation that the canopy cover of Stipa breviflora increased only with fall litter removal, when the presence of litter would have a greater impact over a longer period, and litter addition reduced the cover of Convolvulus ammannii only in years of low precipitation when the plant was subject to additional stress. Similarly, the greater plant height of Stipa breviflora to heavy litter removal in dry years may have been in response to the same processes, although the explanation of its height response to litter addition is obscured by the complexity of a 3-way interaction.

The greatest proportion of plant types found in this community is annual forbs, consisting primarily of Salsola collina and Neopallasia pectinata, which demonstrate the greatest fluctuation in relation to precipitation and, to a considerably smaller degree, the litter treatments. The presence of litter changes soil available water [33]. On the other hand, the ANPP of grass expressed stable production regardless of precipitation, reflecting a highly drought-resistant trait, and only the cover of individual species responded to precipitation. The effect of the changes resulted in a greater proportion of perennial to annual species with low precipitation in the litter removal experiment but a lower proportion in the addition experiment. The differences in trends between the two experiments appeared to be dictated by the relative mass of grass and forbs and the contrasting response of Salsola collina and Neopallasia pectinata to moisture regimes, where Salsola collina increased and Neopallasia pectinata decreased with greater precipitation.

5. Conclusions

The vegetation of the Desert Steppe is resilient to moisture stress with perennial grasses, as a group, being unresponsive to precipitation and litter mass. The greatest response came from the annual forb, Salsola collina, which was primarily responsible for taking advantage of increased precipitation and explained for fluctuations in total biomass. While litter affected the canopy cover of some species, it had no effect on the biomass of grass or forbs, or their total. Therefore, litter management as a strategy for enhancing production on the Desert Steppe is ineffective although the abundance (canopy cover) of selected species may be affected. The study did not address other factors that could be indirectly affected by the presence of litter or sheep grazing. The fact that ANPP at the site with litter addition was only about 60% of that at the site with litter removal suggests that other factors, such as a smaller proportion of Salsola, may be responsible for the difference. Nevertheless, the community responses to the litter treatments were similar at both sites even though the effects were unremarkable.