The ecosystems across the world are affected by diverse impacts of climate change. Climate change will also affect soil seed bank as it is determined by the plants that are above the ground. Soil seed bank is the repository for the next generation of plants; these are the seeds on the soil surface and within the soil (Hu et al., 2019). Soil seed bank is very important in grassland productivity as it is the basis for survival, population settlement, expansion, and reproduction (Milberg and Hansson, 1993). History and the current situation of a community is reflected in the soil seedbank (Thompson, 2000; Kalamees et al., 2012). Any factors contributing to vegetation productivity or reduction in biodiversity and grassland degradation can affect the size and the composition of the soil seed bank (Dreber and Esler, 2011; Báldi et al., 2013). The distribution of rainfall also affects seed production and release (Gutterman, 2012). Disturbance related to environmental changes influence the overall heterogeneity in the soil seed bank (Solomon et al., 2006; Kassahun et al., 2009).

By allowing a part of their offspring into a dormancy and therefore forming seed bank, plants commonly delay their reproduction (Baskin and Baskin, 1998). A reduced mean annual number of offspring is because of delayed reproduction which is an apparent maladministration (Aikio et al., 2002). Soil seed bank is an important aspect of the grassland as it determines its current and future condition. Seed recruitment from the soil seed bank is restricted to soil parameters and periods with favorable conditions for seed germination (Solomon et al., 2006). Soil water is one of the important variables in this regard (Snyman, 2004), along with soil pH (Henig-Server et al., 1996), light (Trabaud and Renard, 1999), and heat exposure (Snyman, 2005).

According to del Cacho et al. (2012), simulated climate experiments provide the best approach to study the processes that determine the community changes at stand level. Most farmers rely on the natural rangeland for all-year-round grazing of their livestock, and therefore the status of the soil seed bank is important to understand how climate change will affect the soil seed bank to prepare for the future. This will also assist farmers to understand the measures that can be undertaken to restore the rangeland. Understanding the deterioration of the rangelands will also assist on how they can be restored (Solomon et al., 2006). Climate manipulation may affect soil seed bank when the seed input is not depleted by the germination due to unfavorable conditions imposed by climate modification (del Cacho et al., 2012). Grassland researchers and scientists predict a reduction in cover and productivity due to summer droughts (Hufkens et al., 2016; Basto et al., 2018). However, some studies on climate manipulations show that these responses will defer from community to community and soil properties (Fridley et al., 2011). Furthermore, some components such as soil seed bank have not been incorporated enough in studies that evaluate the impact of climate change on rangeland productivity. Most rangeland studies in Africa focus only on the aboveground vegetation community and ignore the soil seed bank which also requires attention. Therefore, this study was conducted to determine whether the rainfall interception and resting period have a significant effect on the soil seed bank or not. We examined the forbs and grass composition and density in four different rainfall interceptions (0%, 15%, 30%, and 60%) and two resting periods (70 and 90 days). Resting period was employed to determine the period required to allow seed dispersal when the veld is rested.

Data for seedling emergence was analyzed using the general linear model procedure (Proc GLM) of SAS (2008). The rainfall interceptions and the resting periods were the fixed effects, while the blocks (plots arranged in rows) were the random effects. Seedling emergence was the number of seedlings that emerged in each tray. The model was designed to determine whether rainfall interception and resting period have a significant effect on soil seed bank or not. The seedlings were separated into non-grasses (forbs) and grasses. Grasses and forbs were response variables, while rainfall interception and resting period were response variables while rainfall interception and resting period were the studied factors.. The statistical were conducted separate for each group. Densities per tray and per square meter were analyzed for forbs, grasses, and total number of seedlings. Raw data was tested for normality and where it fails log transformed data were statistically anlaysed. However, the actual means were presented in Tables and Figures.

Species diversity was calculated using Simpson’s index (Yadav and Misshra, 2013):

where D = diversity, n _{i =} total number of plant species of each individual species, and N = the total number of all plant species

Species richness was calculated using Margalef’s index (Margalef, 1958):

where S = total number of species, N = total number of individuals in the sample, and ln = natural logarithm

A total of 26 plant species, of which seven were grass species and 19 were forbs, were recorded in the soil seedbank in 2018, while there were 25 plant species in 2019 with six grass species, 18 forbs, and one sedge ( Tables 1 , 2 ), respectively. This means that grasses were mainly affected by rainfall interception than forbs. Emergence in terms of functional groups (grasses and forbs) differed from treatment to treatment, with 60% rainfall interception treatment having the lowest grass seed abundance in both years ( Table 3 ),while forbs seed abundance was highest in both years ( Table 4 ). On the other hand, 90-day resting period had higher grass abundance than 70-day resting period in 2019, while the resting period had no significant effect on grass soil seed bank abundance in 2018 ( Table 3 ). This could be due to the fact that 70 days’ harvest took place before the grasses dropped their seeds, while 90 days’ harvest took place after they dropped their seeds. This resulted in 70-day resting period having less grass soil seed bank than 90-day resting period, while resting period had no significant effect on grass soil seed bank in 2018. These results concur with those reported by Beier et al. (2004) when it comes to grasses, as they reported that soil seed bank fell at drought treatments compared to control treatments due to the probability of indirect water deficit. The site has not been grazed for over 60 years, and the plots have been subjected to rainfall interception and cutting for 5 years. These treatments have had an impact on the above-mentioned vegetation, which has led to a decrease in grass productivity on high rainfall interception treatment (60%), while that of forbs increased with the increase in rainfall interception ( Table 4 ). This is due to the fact that most forbs have a deep root system than grasses. These forbs dropped their seeds which, in turn, contributed largely to the soil seed bank. According to Akinola et al. (1998a, 1998b), rainfall manipulation scarcely modifies the abundance soil seed bank.

Water availability limits plant growth in the ecosystem (Larcher, 2000), difficulties in plant growth and survival could be exacerbated by drought (Llorens et al., 2004; Prieto et al., 2009), and ultimately seed production will be affected (del Cacho et al., 2012). However, our rainfall interception treatments did not completely exclude soil moisture but rather reduced its availability. As a result, seedlings were able to emerge after rainfall events, and it could affect the full plant reproductive cycle of producing seeds (del Cacho et al., 2012). The experiment may result in cumulative effects if continued over several years, resulting in a significant decrease in soil seed bank of grasses and more forbs at high rainfall interception compared to control treatments. This also concurs with the results that were reported by Akinola et al. (1998a), who argued that annual species, which are mostly forbs, respond faster to climate manipulation.

Many previous studies have reported large persistent seed banks on rangelands (Milberg and Hansson, 1993; Edwards and Crawley, 1999), where in most cases below-ground species composition differs from that aboveground (Snyman, 2004). In our study, our results showed lower densities of soil seed bank especially at the 0% and 15% rainfall interception treatments ( Figure 3A ). The high seedling density of forbs at 60% rainfall interception in both years correlated with high rainfall interception. This is also in agreement with what was reported by Snyman (2004) who reported a slightly larger soil seed bank production at the degraded rangeland. Few decreaser grass species such as D. eriantha and T. triandra were recorded at low rainfall interception than high rainfall interception treatments. The absolute soil seed bank composition can be important in determining the future condition of the rangeland and dictating which species will play part in successional processes (Kinucan and Smeins, 1992; Snyman, 2004). Rangeland degradation has a potential to reduce the soil seed bank and therefore change its species composition (O’Connor et al., 2001; Ingram, 2003); however, opposing trends have been observed (O’Connor and Bredenkamp, 1997; Snayman, 2004), and this is what has happened in our study as the total densities per square meter have increased at 60% rainfall interception in 2018 and at 15% rainfall interception in 2019.

In other studies where there were no grazers involved such as this one, it was found that perennial grasses increase, and this results in a rapid decline in species diversity (Sykora et al., 1990; Hill et al., 1992; Snyman, 2004). It is important to take different methods used in data collection when comparing soil seed bank studies such as this one, including depth, sample size, number of germination cycles, seasons, and whether there are grazing animals involved. Soil seed bank densities from our treatments (282.74, 208.60, 260.32, and 503.41 total seedlings per square meter for 0%, 15%, 30%, and 60% rainfall interception, respectively) and these values were lower than the previous reported results reported outside South Africa, but within the South African results. In a study that was conducted by Ruyle et al. (1988), they recorded 800 m⁻² in southern Arizona and 2,252 to 4,409 seeds per square meter in Texas (Kinucan and Smeins, 1992). On the other hand, our results are within the South African reported results where Adams (1996) recorded 200 seeds per square meter in KwaZulu-Natal and O’Connor (1997) who recorded 350 seeds per square meter of Themeda triandra in the soil seed bank.

The 60% rainfall interception treatment had more bare patches than the rest of the treatment plots compared to other rainfall interception treatments. Seed predators such as ants also play a pivotal role in predation and seed dispersal; on the other hand, the vegetation determines the composition of ant communities and structure (Retana and Cerdá, 2000; Arnan et al., 2006). These ants were more abundant at the 60% than 0%, 15%, and 30% rainfall interceptions. This is in agreement with the results that were reported by Arnan et al. (2006) and Rodrigo and Retana (2006) who reported that ants were more abundant under low vegetation cover as well as Wolff and Debussche (1999) who reported that ants disappear with the increase in vegetation cover. According to del Cacho et al. (2012), ants’ activities may increase in plots with larger bare patches due to water deficit that hampers the regeneration of vegetation, leading to increased seed removal. However, it is also possible that the ants may also bring the seeds from outside the plots, but this would be minimal as there were no visual observations of high densities of ants outside the 60% rainfall interception treatments. However, these bare patches reduced the densities of grass seedlings and increased those of forbs. There might be reasons for these termites to prefer grass seeds over those of forb plants, e.g., size.

Basto et al. (2018) reported that grass seeds in calcareous grassland soils are not resistant to long-term summer drought. This may also justify why there were no major differences between the aboveground and the soil seedbank as the grass seeds might have died as the experiment has been in existence for more than 5 years. The decline in grass soil seed bank varies with soil water content and grass species, while other species increase germination as the soil moisture decreases (Gazanchian et al., 2006; Rohollahi et al., 2015). Reduced soil moisture also influences the availability of soil nutrients that are required by plants (Köhler et al., 2001)—for example, some species growing in dry grassland soils have smaller nitrogen uptake (Durand et al., 2010), and the nitrogen may accumulate in soil and have possible effects on soil seed bank (Basto et al., 2015a) and soil pH (Carroll et al., 2003), which may also result in seeds being persistent in grassland soils (Basto et al., 2015b). Even 1-year drought events may have long-term effects that are unpredictable on grasslands (Haddad et al., 2002).

Our results show that rainfall reduction has a negative effect on grass soil seed bank, while it improves the forbs soil seed bank. These results also show that rainfall reduction may eventually affect seed storage in the soil, which is the one that plays a pivotal role in grassland plant dynamics. The results from this study paint a picture of how the rangelands will look like during and after the events of reduced precipitation. The most palatable species such as T. triandra and D. eriantha were fewer in the soil seed bank from the highest rainfall interception. The reason for this could be that these grasses did not produce seeds at this treatment as it was visually observed during data collection. Future rainfall reductions may lead to grasslands being dominated by forbs over grasses with larger bare patches, which will lead to soil erosion and land degradation. These forbs are probably suppressed by the grasses at other rainfall interception treatments and take advantage as there are less grasses at 60% rainfall interception. According to Snyman (2004), most researchers recommend 80-mm depth when collecting soil seed bank; however, we collected up to 20 mm as the density of the seed bank declines with depth (Fenner and Thompson, 2005), with the majority of the seeds being found in the first 5 cm of the topsoil (Valbuena and Trabaud, 2001). However, we would encourage more studies of this nature to be conducted taking into consideration the effect of time lags between soil seed bank collections. This calls for alternative interventions that can be taken into consideration should the droughts be more persistent—for example, reseeding of rangelands with naturally occurring grasses in a specific area.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

The manuscript presents research on plants and does not require ethical approval for their study.

TM: Writing – original draft, Funding acquisition. AH: Supervision, Conceptualization, Project administration, Writing – review & editing. ET: Supervision, Writing – review & editing.