SOIL AND GRASSLAND EFFECTS ON GRASSLAND NET PHOTOSYNTHESIS IN GANNAN, CHINA, FROM 2000 TO 2018

What is the ideal soil-grass combination for maximum photosynthesis? In this study, we investigated how soil and grassland types affect photosynthesis in the grasslands of Gannan, China. We divided the grasslands of Gannan into 166 study sites, each with a unique soil-grass combination by intersecting the soil and ecoregion maps using ArcGIS. We obtained 19 years of data on the Net Photosynthesis (PsnNet) of grasslands in the area from 2000 to 2018 and then divided them into “growing season” (June to September) and “non-growing season” (October top May). Between 2000 and 2018, PsnNet of grasslands showed a gradually increasing trend. The effect of soil type on PsnNet was not significant during the growing season. However, it was highly significant during the non-growing season. Among the soil types, grasses that grew in Mollic, Gelic, and Haplic soils had the highest rate of photosynthesis. The difference in PsnNet among the various grass types was highly significant during both seasons. However, Tropical and Subtropical succulent evergreen broad-leaf shrubs and Temperate meadows had the highest rate of photosynthesis. Additionally, there was a highly significant difference in PsnNet among the various soil-grass interactions. In the growing season, TStEgBLS growing Eutric soils had the highest PsnNet. However, SaDBLS growing in Gelic soils had the the highest.


Introduction
The role of grasslands in the energy and matter exchange between the land's surface and the atmosphere cannot be overemphasized. They play a crucial role in the assimilation of atmospheric carbon dioxide through photosynthesis. Light energy is a key factor in this process. Generally, photosynthesis that occurs in the presence of light is called photorespiration. However, dark or mitochondrial photosynthesis can occur with or without light energy. When photorespiration and dark photosynthesis are subtracted from true photosynthesis, net photosynthesis is obtained [1]. Photosynthesis is essential for the reduction of atmospheric carbon, thereby slowing global warming.
Several strategies have been devised to absorb the fast-growing amounts of atmospheric carbon. One such strategy is afforestation and reforestation [2][3][4][5]. Research shows that grasslands have a high potential to sequester more carbon in the soil [6]. However, they face problems of shrinkage and degradation [7,8] owing to overgrazing and poor The elevation ranges between 1172m (about the height of the Burj Khalifa, the tallest building in the world) and 4777m (about half the cruising altitude of a commercial jet) above sea level. Temperatures are between −30.6 °C and 28.9 °C with annual means between 1 °C and 3 °C. The annual precipitation varies from 400 to 800 mm (about the length of a baseball bat), with potential evapotranspiration of 1200-1350 mm (about half the height of a regulation basketball hoop) [19]. High-elevation areas experience cold summers. However, areas with low elevations experience warm summers. There is snow cover in the high mountains during the winter season.
The area is characterized by nineteen soil types. These were further classified into ten major groups: Mollic, Haplic, Gelic, Eutric, Calcic, Dystric, Luvic, Terric, Fibric, and Calcaric soils (Table 1). There are fourteen grassland types including Tropical and Subtropical Succulent Evergreen Broad-Leaf Shrubs, Temperate Deciduous Broad-Leaf Shrubs, Sub-Alpine Deciduous Broad-Leaf Shrubs, Sub-Alpine Hard-Leaf Evergreen Broad-leaf Shrubs, Alpine Cushion, Temperate Meadows, Marsh Meadows, Tibetan Caragana & Dwarf Grass Desert, Thorn weed, Halophyte, Alpine Meadows, Temperate Typical, and Alpine Sparse [20,21]. The most dominant include Temperate Typical, Alpine Meadow, Typical Meadow, and Temperate Deciduous (Table 2). Grasslands cover 65% of the total land surface. . It is a repository of the spatial distribution of ecosystem classifications freely available in vector format. It comprises the spatial distribution patterns of various ecosystems across the country, such as forests, grasslands, shrubs, farmland, towns, deserts, bare land, and other ecosystems.

Soil data
We obtained soil data from the Harmonized World Soil Database Version 1.2 (HWSD V1.2) [21] in a vector format. This version was compiled from: the European Soil Database (ESDB), the soil map of China, a number of regional SOTER databases, and the Soil Map of the World [21]. This was a collaborative among the ISRIC-World Soil Information, the FAO, the European Soil Bureau Network, and the Institute of Soil Science, Chinese Academy of Sciences. The Soil Map of China was based on data from the Second National Soil Survey of China in 1995 and distributed by the Institute of Soil Science in Nanjing [22].

Net Photosynthesis (PsnNet)
We downloaded the MOD17A2HGF Version 6 Net Photosynthesis (PsnNet500) product from lpdaac.usgs.gov [23]. The gap-filled MOD17A2HGF is an improved version of MOD17, which has as input, the cleaned poor-quality 8-day Leaf Area Index and Fraction of Photosynthetically Active Radiation. MOD17A2HGF is a cumulative 8-day composite of values based on the radiation use efficiency concept at a resolution of 500m. The PsnNet (kg C day-1) values were calculated from gross primary production (GPP) and maintenance respiration as: where GPP is the initial daily total of photosynthesis and daily net photosynthesis (PsnNet), and Leaf_MR and Froot_MR are leaf and fine-root respiration respectively. [23,24].

Data Preparation
The soil and vegetation maps in vector format were intersected using the ArcGIS intersect tool to produce a map of 166 sites. Each site is a unique combination of soil and grassland types as shown in Table 3. Using QGIS, the area of each site was computed in square meters. The resulting map was used as input for the lpdaac Application for Extracting and Exploring Analysis-Ready Samples (AppEARS) data tool to extract the PsnNet values for each site. This is a cumulative 8-day composite of values within each month. Therefore, for each site, the average net photosynthesis at 500m resolution for each month was calculated as: AvPsnNet500(M x ) = (1 st 8days + 2 nd 8days +…+n th 8days)/n (2) where AvPsnNet500(M x ) is the average net photosynthesis (kgC/8day) in month M x at 500m resolution.
By proportion, we obtained the mean PsnNet for each site in any given month as: where AvPsnNetS ID is the PsnNet (kgC/8day) at site S ID in any given month, and A SID is the area of site SID (sqm).
We split the entire data into growing season (June to September) and non-growing season (October to May).
For the same site, we computed the percentage change in PsnNet between the growing and non-growing seasons for the various soil-grass interactions as: where %CPsnNet is the percentage change in net photosynthesis between the growing and non-growing seasons, PsnNet GS is the net photosynthesis in the growing season, and PsnNet NGS is the net photosynthesis in the non-growing season.
We employed a split-plot design with a mixed-effects (random and fixed) model to analyze the effects of soil and grassland as well as their interaction on PsnNet. Grass types were manipulated at the subplot level and the soil type was manipulated at the plot level. Grass and soil types were treated as random effects. "Block" was set as a fixed factor. The response variable was PsnNet (KgC/8day).

Statistical Analyses
All statistical analyses were performed using the RStudio Version 1.4.1096. ArcGIS version 10.16 and QGIS version 3.16.2 were used for all map-related tasks. The summary statistics were computed using Microsoft Excel 2016. In all cases, P ≤ 0.05, 0.01, and 0.001 were considered significant, moderately, and highly significant respectively. Linear mixed-effects model (LME) analysis was performed to test the possible dependency of grassland photosynthesis on soil and vegetation types. A post hoc test for pair-wise comparison of means was conducted using the Tukey-LSD test.   Figure 3 shows the spatial distribution of net photosynthesis in Gannan between 2000 and 2018. The values represent the average photosynthesis at 500m resolution for the respective range of years. These values were calculated using ArcGIS zonal statistics tool. The spatial distribution indicated that net photosynthesis decreased from the southeast to the southwest. The southeastern region is dominated by Tropical succulent evergreen broadleaf and Subalpine hard-leaf evergreen broad-leaf shrubs [19]. These are perennial woody plants with persistent woody stems above the ground. The southwestern region is dominated by Alpine meadows. In this area, elevation may reach over 4700m above sea level. During winter, the area is usually covered by snow and the summers are also cold. Water runoff is also a common phenomenon in the alpine regions as compared to the subalpine regions. As a result, the subalpine regions are well hydrated year-round. This is why spatial location has a highly significant effect on net photosynthesis during the non-growing season. However, in the growing season, location (block) had a significant effect on net photosynthesis (Table 4). This could also be attributed to the fact that some soil types have higher water retention capacity, thereby supporting plant growth even in the absence of rain, especially in the subalpine regions.  Alpine sparse (AS) had the lowest rate of photosynthesis followed by Marsh Meadows (MM). As shown in Table 4, grassland type had a highly significant effect on net photosynthesis in both the growing (Table 4A) and non-growing (Table 4B) seasons.

Effects of Grassland type on Productivity.
On a monthly scale, Temperate meadows (TM) perform best between June and August, dropping, however, below Tropical and Subtropical succulent evergreen broad-leaf Shrubs (TStSEgBLS) which maintain the highest rate of photosynthesis until mid-May. Throughout the year TM and TStSEgBLS maintained the highest PsnNet ( Figure 5).

Effect of Soil type on PsnNet
We were also interested in determining which soil types supported which grassland types. Figure 6 shows the PsnNet distribution according to the soil type in which particular grasses grow. The values are the net photosynthesis by the coverage area of the respective soil-grass type interaction. Grasses that grow in Mollic, Gelic, Haplic, and Eutric soils have the highest net photosynthesis, with those growing in Fibric soils having the least. ( Figures. 6A and B). In the growing season, soil type had no significant effect on net photosynthesis (Table  5A). However, in the non-growing season, it had a highly significant effect. (Table 5B) From the grouping analyses, it was confirmed that soil type does not cause much variation in net photosynthesis during the growing season ( Figure 7A). Nonetheless, there was an obvious variation during the non-growing season ( Figure 7B Treatments with the same letter are not significantly different. In the growing season, Terric, Eutric, and Haplic form one group, thus they are not significantly different; they form one group. Also, Gleyic, Calcaric, Luvic, Gelic, Mollic, Fibric Calcic, and Dystric are not significantly different; they form another group ( Figure 7A). This confirms that there was no significant variation among the soil types in the growing season. However, there was a large variation among the soil types in the non-growing season; 8 groups, as opposed to only two in the growing season, exist in the non-growing season ( Figure 7B).

Impact of Soil and Grassland Types Interaction on Net Photosynthesis
The interaction between soil and grassland types was highly significant in both the growing and non-growing seasons (Tables 6A and B). Table 6. ANOVA of Soil-Grass interaction effect on PsnNet (a) Growing Season (b) Non-growing Season. Significance codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.   Shrubs and Eutric soils was the most productive in the non-growing season ( Figure   8B).

ANOVA (Growing season)
As mentioned above, we observed that the percentage changes in PsnNet for all soil-grass interactions between the two seasons were not proportional. As illustrated in Figure 9, about 40% of the photosynthetic activities of Temperate deciduous broad-leaf growing in Calcaric soils remain during the non-growing season. However, less than 10% remains in the Haplic soils.

Discussion
The soils and grasslands of Gannan are not evenly distributed. For instance, Eutric and Calcic soils alone take up over 69% of the total land area (Table 1). Nonetheless, the other soil types support a good variety of grasses. For example, Gelic soils, which cover only 0.15% of the total area, support 6 different grassland types (Table 1, Fig. 8). In addition, Alpine meadows and Subalpine deciduous broad-leaf shrubs alone cover over 83% of the total land area (Table 2). However, at 500m resolution, they do not have the highest photosynthetic activities in Gannan. On the contrary, Tropical and subtropical succulent evergreen broad leaves and temperate meadows, which occupy only 0.081% of the total area, have the highest photosynthetic activities at a 500m resolution, followed by Subalpine Deciduous Broad-Leaf Shrubs (Table 2, Figure 4). These are mainly located in the subalpine regions. Especially for Temperate Meadows and Tropical and Subtropical Deciduous Broad Leaf Shrubs, photosynthesis is year-round. This indicates that if these high productivity grasses had a larger coverage area, grassland productivity would increase enormously.
Notwithstanding, the grasslands of Gannan are showing improving vegetation health (Fig 2A ), thanks to Temperate meadows and Tropical and Subtropical succulent evergreen broad-leaf shrubs which show increasing trends for the study period (Fig. 4). This could be attributed to the various grassland reclamation projects by the Chinese gov-ernment over the years such as the "Retire Livestock and Restore Pastures" [25]. Generally, the Alpine regions have the lowest amount of photosynthesis as compared with the subalpine regions. This could be explained by the fact that there is occasional water runoff to the subalpine regions thereby keeping them wet at the expense of the alpine regions. Consequently, the subalpine regions have year-round soil water to support plant growth. Also, temperatures in the alpine regions are extremely low in the winters. As a result, soil water becomes frozen, thereby inhibiting plant growth [26][27][28]. This explains why location (block) has a significant effect in the non-growing season. (Table 5B). Additionally, some soil types have higher water retention capacities than others, and this may further contribute to disparities in photosynthesis.
We observe from the grouping analyses (Fig. 7) that a greater diversity exists among the soil types in the non-growing season than there is in the growing season. The soil effect on net photosynthesis in the growing season (Table 5A) is not significant. We conjecture that in the growing season, there are favorable weather and environmental conditions for all soil types to support plant life. There is sufficient rainfall for maximum production [29,30]. In this period, all soils exhibit their best and maximum agronomic properties [9]. However, soil effect on net photosynthesis is highly significant in the non-growing season (Table 5B). Again, this is because, in the non-growing season, harsh weather conditions may turn the odds against soil types that are highly sensitive to harsh weather conditions. In other words, different soils respond differently to different weather conditions, and hence their support for plant life may be affected differently [31,32]. It is important to note that Mollic, Gelic, and Haplic soils which interaction with grass triggers the highest photosynthetic activities only take up a little over 12% of the total area. Nevertheless, Eutric soils, which occupy over 40% of the total area are only fourth to the aforementioned soil types. (Fig 6B, Table 1). This is useful, as policymakers may have to pay special attention to these extremes.
Additionally, the variation of soil nutrients between seasons could also be a major factor. The role of soil nutrients in photosynthesis cannot be overestimated. For example, 3nutrient deficiency may lead to low production of chlorophyll [33]. Even though the annual and seasonal losses of the main nutrient elements are small according to Alías et al., [10] our findings show that soil effects on photosynthesis are different between the growing and non-growing seasons. If Alías et al. are right in their conclusion, we hypothesize that the small seasonal changes in the concentration of main nutrients are enough to make a difference. These variations are caused mainly by temperature and precipitation as observed by Patel, M.P.; Gami, B.; Patel B. [34]. In concordance with findings by other researchers, they concluded that with the change in precipitation and temperature regimes, physical properties and trace element concentration vary significantly [32, 34 and 35, 36]. Fatubarin and Olojugba also found that soil pH and available phosphorus were higher in the dry season than in the rainy season [29]. Additionally, Gentili et al observed that high pH values affect growth adversely [30], observing that plants grown at pH7 were shorter and developed leaves at a slower rate than those grown at pH5 and pH6. At the same time, soil organic matter and total nitrogen (essential for plant growth) were low, but nitrogen content increased at the peak of the rainy season due to nitrogen fixation [31]. Other researchers also confirm soil properties vary with seasons [37,38]. In a nutshell, the amount of soil nutrients changes with seasons.
Directly linked to the seasonal variation of soil nutrients is the activities of soil microbes as observed by. Xue et al. They showed that soil microbial communities directly affect soil functionality by cycling soil nutrients and carbon storage [39]. They concluded that microbial communities vary in space and time, between soil types and under different land management. These variations, they claimed are governed by soil properties. As has been shown, soil properties vary with season. This implies that different soils support microbial growth to different degrees in different seasons, which may alter the main soil nutrients from season to season. This seasonal variation of nutrients and environmental factors could cause seasonal variation in photosynthesis. For example, the net photosynthesis of Alpine Cushion in Calcic and Mollic soils are equal in the non-growing season.
However, it increased more in Calcic soils than it did in Mollic soils in the growing season ( Figure 8). Furthermore, the net photosynthesis of Subalpine Deciduous Broad-Leaf Shrubs in Luvic, Molllic, and Terric soils are equal in the non-growing season. Nonetheless, photosynthesis in Luvic soils lags those in Mollic and Terric soils in the growing season (Fig 8). A similar phenomenon could be observed across the various soil-grass types interactions.
Another important reason is the variations in phonological response to seasonal changes among the various grass types. Chang et al [40] confirm this in their research on European grasslands. They established that there is a significant phenology shift, particularly in an earlier winter-spring onset of grass growth over Europe. This phenology shift differs among grasses, thereby affecting their productivity curves. Also Digrado et al. [32] confirm that plants respond differently under stressed conditions observing that decreased photosynthetic performance was associated with high respiration flux. Their study revealed that variation in the photosynthetic performance of a grassland ecosystem responded to environmental constraints. This can be confirmed by the fact that there is an unequal percentage change in productivity from the growing season to the non-growing season among the various soil-grass combinations (Fig 9). This is illustrated by [18] showing that the Leaf Area Index (LAI) exhibited large seasonal change. It is therefore clear that soil and grassland types all contribute to the seasonal variation of photosynthesis in Gannan. Consequently, the interaction effect between soil and grassland types is highly significant in both seasons.

Conclusions
In this study, we have established that the effect of soil on grassland photosynthesis varies between seasons (growing and non-growing seasons). It is highly significant in the non-growing season but not significant in the growing season. However, the effects of grassland type on grassland photosynthesis are highly significant in both seasons. In addition, the interaction of the soil-grass type has a highly significant effect on net photosynthesis. Grassland health in Gannan is gradually trending up but only prominent in areas dominated by Temperate Meadows and Subalpine Deciduous Broad-leaf Shrubs. We hope that our findings would be informative to policymakers in selecting not only which grassland type to focus on, but also which soil type to improve. We recommend that Temperate Meadows be given special attention since their production is high, second only to Tropical and Subtropical Succulent Evergreen Broad-leaf Shrubs. In addition, areas occupied by Eutric soils in Gannan should be given special management care. In future grassland restoration projects, areas occupied by Mollic, Gelic, and Haplic soils should be a priority as their interaction with grasses triggers the highest photosynthetic activities. This would in turn accelerate the absorption of atmospheric carbon, thereby mitigating global warming.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Figure S1: title, Table S1: title, Video S1: title. ated during the study. Please refer to suggested Data Availability Statements in section "MDPI Research Data Policies" at https://www.mdpi.com/ethics. You might choose to exclude this statement if the study did not report any data.