1 Introduction

Savannas cover about a fifth of the earth’s land surface, and they are characterized by the coexistence of two types of vegetation: trees and grasses (Baudena et al., 2015). Type of vegetation (trees and grasses) is an important component influencing soil physical properties and processes in savanna ecosystems (Holdo & Mack, 2014; Ludwig et al., 2004). It has been established, for example, that savanna tree canopies diminish understory solar radiation and temperature and enhance soil moisture availability, with consequences for herbaceous composition and diversity (Holdo & Mack, 2014). Savannas, particularly in Africa, are also dominated by grazing herbivores, which impose their own effects on soils and plants, thus potentially enhancing or counteracting tree canopy effects (Heggenes et al., 2018; Malongweni & van Tol, 2022). Holdo and Mack (2014) found that when herbivores aggregate under trees, canopy–herbivore interactions may result. During canopy–herbivore interactions, animals exert pressure on the ground through the action of trampling, pawing, and wallowing. Pressure exerted on the soil via the trampling action of grazing herbivores may increase the proportion of bare soil, thus causing soil compaction (Heggenes et al., 2018). In turn, soil compaction alters the soil structure and hydrology by changing many aspects of the soil such as strength, gas, water, and heat (Gürsoy, 2021; Holdo & Mack, 2014). Experimental studies have shown that the soil compaction results in an increase in the soil strength, bulk density, volumetric water contents, and field capacity while decrease in total porosity, soil aeration, water infiltration rate, and saturated hydraulic conductivity (Heggenes et al., 2018; Ludwig et al., 2004). A key remaining challenge is to understand the relative contributions and interactions of tree canopies and herbivores to soil functional heterogeneity (Holdo & Mack, 2014).

In addition to the mechanical impact of trampling herbivores, fire is an important driver of soil properties and tree-grass dynamics in savanna ecosystems (Chamane, 2012; Gibson & Hulbert, 1987; Smit et al., 2010; Van Langevelde et al., 2003). Smit et al. (2010) found that fire alters vegetation biomass and structure through direct killing or by reducing grass and woody vegetation height, stem density, and biomass and, to a lesser extent, by altering vegetation composition of the herbaceous layer. Baudena et al. (2015) claim that C4 grass biomass enhances fire spread in open ecosystems due to its high flammability. At the same time, grasses benefit from fire because they recover faster than trees and benefit from the open spaces after fire, thus creating a positive feedback mechanism that enhances savanna formation and presence. According to Malongweni and van Tol (2022), fire-affected soils are also associated with the loss of soil organic carbon. The combustion of soil organic carbon during the burning process leads to major changes in soil physical and hydraulic properties and processes (MacKenzie et al., 2004; Mehdi et al., 2012). These changes include the formation of a water-repellent layer, disintegration of stable soil aggregates, reduced infiltration, development of preferential flow, decreased hydraulic conductivity, increased surface crusting and surface runoff, and a loss of biodiversity (Gürsoy, 2021; Holdo & Mack, 2014; Ludwig et al., 2004; Sandhage-Hofmann et al., 2021). Conversely, Malongweni and van Tol (2022) and Holdo and Mack (2014) argue that fire improves soil fertility and aeration, thus promoting water and nutrient infiltration, and aggregate stability, while minimizing compaction and runoff. This is because fire tends to chemically convert nutrients bound in dead plant tissues and the soil surface to more available forms. Moreover, herbivores are attracted to recently burned area and they deposit their dung and urine in those burnt areas (Holdo & Mack, 2014). Animal excreta (dung and urine) is an excellent source of nutrients known to improve the physical, chemical, biological, and hydraulic properties of soils (Malongweni & van Tol, 2022; Tuomi et al., 2021; van Coller et al., 2013).

Since fire, herbivores, and vegetation type highly influence soil properties and processes, numerous studies have been conducted on them (Baudena et al., 2015; Chamane, 2012; Holdo & Mack, 2014; Smit et al., 2010; Van Langevelde et al., 2003). However, much of the literature has focussed on effects of fire and herbivores on savanna soils either in isolation or in relation to vegetation distribution. An understanding of the overall interactive effects of fire, herbivores, and vegetation type on soil physical properties is often not the focus of investigation and has been less studied. Assessment of the long-term effect of fire, herbivores, and vegetation type on soil physical properties could not only guide improvement to fire and herbivore grazing management practices but may also provide information that will serve as a guideline towards maintaining or improving soil biodiversity and soil health in savanna ecosystems. Therefore, this study is aimed at determining the influence of 20 years of fire, herbivores, vegetation type, and their interaction on selected soil physical properties in a semi-arid savanna ecosystem, Kruger National Park (KNP), South Africa. The authors of this paper recently published an article investigating the 20-year impact of fire, herbivory, vegetation type, and their interaction on soil biochemistry (Malongweni & van Tol, 2022); hence, this one only focuses on soil physical properties.

2 Methodology

2.1 Description of study area

The study site was located in the Nkuhlu large-scale long-term exclosures (24°59′10″ S, 31°46′24.6″ E) of the Kruger National Park (Fig. 1). The area receives approximately 561 mm of rain annually which means it is a semi-arid subtropical savanna. The average annual temperature ranges from 5.6 °C in winter to 32.6 °C in summer (Siebert & Eckhardt, 2008). The most common woody species include Combretum apiculatum, Grewia bicolor, G. flavescens, Dichrostachys cinerea, Euclea divinorum, Terminalia prunioides, Spirostachys africana, Vachellia (Acacia) grandicornuta, and Senegalia (Acacia) nigrescens (Gertenbach, 1983).

Fig. 1
figure 1

Location of study area (red dot) within the Kruger National Park in South Africa and map of the Nkuhlu exclosures indicating the locations of the sample plots [adopted from Le Roux et al., (2017), Malongweni and van Tol (2022)]

Full size image

The size of the exclosures is 139 ha consisting of 3 different herbivory treatments of (1) full exclosure: a 70-ha fully fenced area designed to exclude the entry of herbivores, (2) partial exclosure: a partially fenced area of 44 ha where large and tall animals cannot enter, but all other herbivores are allowed, and (3) open access: a 25-ha unfenced area where all animals are permitted to enter. The herbivory treatments are further divided into two fire treatments of burnt and unburnt plots to give rise to a total of 6 treatment combinations, respectively (Fig. 1).

2.2 Soil sampling and experimental design

As you move down the terrain of the Nkuhlu exclosure variations in the soil profile are apparent (Scogings et al., 2011; Siebert & Eckhardt, 2008). These variations occur with the changes in slope gradient. The crest has shallow sandy soil directly overlying weathering rock, whereas sandy to sandy loam soil can be found on the foot-slope. The foot-slope is sometimes referred to as the sodic zone. According to Siebert and Eckhardt (2008) and Scogings et al. (2011) this is because it is characterized by deep, sodium-rich duplex soils. The shallow sandy soil on the crest can be classified as Leptosol in World Reference Base (WRB) or Mispah and shallow Glenrosa soils in the South African system (Soil Classification Working Group, 2018). The focus of this paper was restricted to the shallow soils of the crest.

There were two selected sampling positions within the exclosures, the crest and foot-slope. Soil samples were collected and analysed for soil physical properties including bulk density, penetration resistance, aggregate stability, and size distribution, respectively. For soil samples intended for bulk density, aggregate stability, and size distribution analysis, five sub-samples were randomly collected to a depth of 10 cm from eight points underneath tree canopies and in open grassland zones. The sub-samples were combined into a single composite sample for each sampling point to give a total of 80 samples collected from KNP. Soil penetration resistance is on the other hand a highly variable parameter, and many measurements are required to obtain a reliable value (Bayat et al., 2017). For that reason, we took 12 penetration tests per observation site.

The study consisted of three herbivory treatments (open access, partial, and full exclosure), two fire treatments (burnt and unburnt plots), and two vegetation types (under tree canopies and open grassland zones) resulting in a 3 × 2 × 2 factorial design which was unbalanced because the open access area had no burn plots. The collected soil samples were air-dried at room temperature for about a week and then passed through a 2-mm sieve obtaining uniform particle size for subsequent analysis.

2.3 Data collection

2.3.1 Bulk density

The bulk density of fine aggregates was determined using the core method (Blake, 1965). In brief, soil cores were used for collecting undisturbed soil samples from the field. Undisturbed core samples were randomly collected from underneath the tree canopies and in the open grassland zones. The undisturbed soil cores were then removed and trimmed to the end with a knife to yield a core whose volume can easily be calculated from its length and diameter. Afterwards, the core rings filled with soil will be oven-dried at 105 °C for 24 h and then later weighed. Bulk density was calculated and presented in g/cm3 using Eq. (1).

$$ {\text{Soil bulk density}} = \frac{{{ }M_{s} }}{{\pi r^{2} h }} $$
(1)

where Ms (g) is the weight of the oven dry soil contained in the core rings, r (cm) is the radius of the core ring, and h (cm) is the height of the core ring.

2.3.2 Penetration resistance (mm)

Surface penetration resistance was measured using a Gilson HM-500 pocket penetrometer. According to Graesch et al. (2015) a penetrometer is an instrument used to test the compaction level of the soil. Penetration resistance is used to assess the physical quality of the soils. It indicates the presence or absence of problems related to soil compaction which has a direct relationship with soil quality.

$$ {\text{kPa value}} = \frac{{{\text{kg}}}}{{{\text{cm}}^{2} }}{\text{value }} \times 98.0665\;{\text{kg}} $$
(2)

To quantify surface penetration resistance, the pocket penetrometer's penetration piston was pushed vertically downwards into the soil surface. As the soil surface resists penetration, the piston compresses a calibrated stainless-steel spring, forcing the indicating ring to slide forward on the piston. At the point when the piston no longer penetrated the soil, the user stopped exerting downward pressure and recorded the measure (tsf2 or kg/cm2) as indicated by the new location of the indicating ring against a piston-mounted scale. Equation (2) was used for expressing penetration resistance in kPa.

2.3.3 Aggregate stability and size distribution

The water stable aggregates of the soil were assessed using the wet sieving method (Youker & McGuinness, 1957). In brief, approximately 50 g of soil clods from each treatment was placed on the topmost of the nested sieves with diameters equalling to 4, 2, 1, 0.5, and 0.25 mm, respectively. The sieves containing clods were then placed in a sieve holder and gently transferred into an aggregate analysis machine which had a pan filled with water. The machine moved the sieves vertically for about 30 min at a rate of 30 cycles per minute at approximately 3 cm. The clods were dunked in water inside the pan as the sieves were moving up and down. The aggregates retained on each sieve were oven-dried at a temperature of 105 °C for 24 h. The parameter chosen to assess the impact of herbivore and fire on aggregate stability was the MWD (mean weight diameter), and it was calculated using the equation described below.

$$ {\text{MWD}} = \mathop \sum \limits_{i = 1}^{n} x_{i} w_{i} $$
(3)

where n is the total number of sieves and \(x_{i}\) and \(w_{i}\) are mean weight (g) and mean diameter (mm), respectively.

Aggregate size class for each sieve was calculated using the following formula:

$$ R_{n} = \frac{{W_{n} }}{T} \times S_{i} $$
(4)

where Rn is the soil aggregate content of each size sieve (aggregate size class) (%), Wn is the weight of the soil aggregate of each sieve size (g), T is the total weight of soil, and Si is the size of each sieve.

2.4 Statistical analyses

Response variables were assessed for conformity to assumptions of normality and homogeneity of variance. Log and logit transformations were used to improve normality and homogeneity of variance. The data were then put through ANOVA to test for significance. Fischer’s protected least squares difference was used to separate the significant means at α = 0.05 using JMP version 16 Pro (SAS Institute Inc., 2021).

3 Results

3.1 Bulk density

The results of the study revealed a significant main effect of vegetation type on bulk density in both the crest and foot-slope, with areas under tree canopies having higher bulk density in open grassland zones than under tree canopies (p < 0.05; Table 1). On the crest, exclosure and fire did not independently cause statistical changes in bulk density nor were there any significant two-way interactions between exclosure, fire, and vegetation type (Table 1). Unlike on the crest, exclosure had a significant main effect on soil bulk density in the foot-slope of the Nkuhlu exclosures, with the full exclosure having the lowest bulk density and the open access area in conjunction with the partial exclosure having the highest (Table 1). In the foot-slope, it was also noted that there were also significant two-way interactions between exclosure and vegetation type, such that absence of both large herbivores and tree canopies resulted in drastic reductions in bulk density.

Table 1 The main and interactive effect of herbivores, fire, and vegetation (veg.) type on soil bulk density and penetration resistance within the crest and foot-slope of the Nkuhlu exclosures
Full size table

The three-way interaction between exclosure, fire, and vegetation type had a significant impact on soil bulk density (Fig. 2A, B). Except for unburnt plots in the full exclosure situated on the crest, three-way interaction between exclosure, fire, and vegetation type resulted in higher bulk density in open grassland zones than under tree canopies (Fig. 2A). Moreover, unburnt plots in the full exclosure had significantly lower bulk density than unburnt plots. On the foot-slope, open grassland zones in the partial exclosure and open access area had the highest bulk density, whereas tree canopies in the full and partial exclosure had the lowest bulk density under both burnt and unburnt conditions (Fig. 2B).

Fig. 2
figure 2

Main differences for the significant 3-way interactive effect of herbivory, fire, and vegetation type on bulk density (A crest, B foot-slope) and penetration resistance (C crest, D foot-slope)

Full size image

3.2 Penetration resistance

The results of the study indicated a significant (p < 0.05) main effect of herbivores and vegetation type on soil penetration resistance (Table 1), whereby tree canopies and herbivore exclusion resulted in lower penetration resistance than open grassland zones and exclosures where herbivores reside. On the foot-slope, it was also noticed that unburnt plots had high penetration resistance relative to burnt plots. Moreover, resistance under tree canopies was significantly higher than in open grassland zones. The two-way interaction between exclosure and vegetation type had a significant impact on the soil’s ability to resist penetration in both the crest and foot-slope. However, the interaction between exclosure and fire caused significant changes in soil penetration resistance only within the foot-slope.

On the crest, we observed that the absence of herbivores in the full exclosure resulted in no changes in penetration resistance amongst all treatments (Fig. 2C). Conversely, the presence of herbivores in the partial exclosure and control site caused a decrease in penetration resistance under tree canopies. However, there were no statistical differences between burnt and unburnt plots. Exclosure, fire, and vegetation type also had a significant interactive effect on soil penetration resistance within the foot-slope (Table 1). With the exceptions of the unburnt plots of the full exclosure, the three-way interaction between exclosure, fire, and vegetation type resulted in lower resistance under tree canopies than in open grassland zones. In the burnt plot of the full exclosure, there were no significant differences in penetration resistance between areas under tree canopies and open grassland zones (Fig. 2D).

3.3 Aggregate size fractions and MWD

Data regarding aggregate size fractions and mean weight diameter showed an inconsistent trend for the influence of exclosure, fire, and vegetation type on the distribution of water stable aggregates in both the crest and foot-slope (Table 2). On the crest, it was observed that areas under tree canopies in unburnt plots of the partial exclosure had the highest mean weight diameter of water stable aggregates. Open grassland zones of the open access area where large herbivores were permitted had the lowest MWD than all the other exclosures. Within the open access area situated within the crest, it can also be observed that areas under tree canopies had higher MWD than open grassland zones. For areas under tree canopies in the partial exclosure, burnt plots had significantly less MWD than unburnt plots.

Table 2 Effect of herbivores, fire, and vegetation (veg.) type on aggregate size fractions and mean weight diameter (MWD) of water stable soil aggregates
Full size table

Within the foot-slope, the open grassland zone the unburnt plot in the partial exclosure had significantly low proportion of large-sized (2–1 mm fraction size) and medium-sized aggregates (0.5–0.35 mm fraction size) aggregates than areas under tree canopies (Table 2). Areas under tree canopies in the unburnt plot of the partial exclosure and open access area had the highest mean weight diameter of water stable aggregates, whereas the unburnt plot on the full exclosure had the lowest mean weight diameter of water stable aggregates than all the other treatments.

4 Discussion

4.1 Herbivores

The results of our study revealed that the most prominent soil changes influenced by herbivores in the crest and foot-slope of the Nkuhlu exclosures were penetration resistance, aggregate size fractions, and mean weight diameter of water stable soil aggregates. The presence of herbivores caused significant changes in bulk density only within the foot-slope and not the crest. The foot-slope of this study area is sodic by nature, and according to Khomo and Rogers (2005), sodic areas are attractive to herbivores because of more nutritious vegetation than surrounding areas, predator vigilance, dietary salts or anti-acidosis minerals, and the presence of water.

The exclusion of herbivores in the full exclosure caused significant reductions in penetration resistance in both the crest and mid-slope. This may be a consequence of soil compaction due to the trampling, pawing, and wallowing action by large herbivores (Malongweni & van Tol, 2022). The partial exclosure and open access area are often associated with trampled compact soils, and according to Tuomi et al. (2021) and van Coller et al. (2013), compaction reduces the soils’ capacity to retain water.

In all the sampling regions, the control site where heavy and tall animals were permitted had the lowest number of water stable soil macroaggregates (≥ 1 mm fraction size). Similar observations have been reported by Holdo and Mack (2014) who found that an increase in herbivore size and population density leads to soil compaction. Herbivore-mediated soil compaction can obstruct the integration of small aggregates into larger aggregate units. This is because compaction destroys the network of tunnels and pores used by soil microorganisms primary responsible for organic matter production (Chamane, 2012; Malongweni & van Tol, 2022Vermeire et al., 2021). According to Brady and Weil (2008), organic matter is an important agent responsible for binding soil mineral particles together creating an aggregate hierarchy.

4.2 Fire

Fire had a significant effect on soil penetration resistance within the foot-slope. Lowest resistance was observed in burnt plots than unburnt plots. Burnt plots had low penetration resistance because fire improves soil aeration through ash production. Bodí et al. (2011) found that ash resulting from burning vegetation is highly porous and tends to reduce compaction; this in turn minimizes penetration resistance by improving aeration. This claim contradicts the findings of many researchers. For instance, a study conducted by Scott (1993) in the fynbos catchments of South Africa investigated the impact of fire on soil infiltration rates. He concluded that fires negatively impacted soils aeration by decreasing soil infiltration. Similar studies conducted locally in African savannas found that fire resulted in crusted and compacted soil surfaces which reduced infiltration (Chamane, 2012; Mills & Fey, 2005; Vermeire et al., 2021).

This may be due to the clogging of soil pores by fire ash, thereby decreasing porosity and increasing penetration resistance (Strydom et al., 2013). Also, according to Malongweni and van Tol (2022), fire tends to grazers. Trampling action by grazing herbivores results in the development of a hard crust on the soil surface. Surface sealing and crusts greatly impede the transport of water and oxygen (hydraulic properties) into the soil profile, thus promoting runoff and erosion (Strydom et al., 2013).

In both the crest and foot-slope, there was an inconsistent trend on the effect of fire towards aggregate size fractions and mean weight diameter of water stable soil aggregates. The influence of fires on large-sized aggregates (> 2 mm fraction size) and MWD was considered negligible within the foot-slope. On the crest, it was noted that burning had a negative effect on water stable aggregates. All the burnt plots had significantly higher proportion of meso (1–0.5 mm fraction size) and micro-aggregates (0.5–0.25 mm, < 0.25 mm fraction size). This agrees with Heydari et al. (2017), who reported negative effects of wildfire on water content, water repellence, bulk density, and stable aggregates.

4.3 Vegetation type

We observed that tree canopies had significantly lower soil bulk density and penetration resistance than open grassland zones in both the crest and foot-slope. Tesfaye and Lemma (2019) reported similar findings, and they claim that this may be due to organic matter build up under the canopy from litter fall and higher turnover of fine roots closest to the tree. The study by Tesfaye and Lemma (2019) concluded that the accumulation of litterfall under the canopy buffered the soil against rain drop impact, wind erosion, and associated compaction as evidenced by the lower bulk density. From agricultural literature, it is known that penetration resistance is positively correlated with bulk density and that bulk density and penetration resistance are major indicators of soil compaction. Therefore, low penetration resistance for samples collected under tree canopies is also due to canopy cover since crown cover reduces compaction and the rate of erosion by protecting the soil against raindrop-induced splashing of particles. Low penetration resistance under tree canopies may also be caused by tree shade. Shading provides shelter from sunlight and reduces soil temperature and evapotranspiration in below-crown environments. Such reductions in water loss are thought to minimize soil compaction. Kumar et al. (2012) found that in dry soil conditions, soil penetration resistance is much higher than wet conditions because soil water acts as a lubricant for soil particles.

Contrary to expectations, data regarding soil aggregates were highly variable and did not follow any specific trend. However, in some cases, particularly in the control site, soil samples collected under tree canopies in the crest had significantly higher microaggregates (< 0.25 mm fraction size) and MWD of water stable aggregates than open grassland zones. In the foot-slope, the area under tree canopies had higher proportion of water stable meso-aggregates (1–0.5 mm fraction size) than open grassland zones. The higher aggregate size fractions and MWD values below trees could be attributed to the presence of more fine and deep roots. Roots enmesh fine particles into stable micro-aggregates by root secretions and physical entanglement, thus resulting in the formation of more stable aggregates (Strydom et al., 2013). The supply of decomposable organic residues by tree leaves also increased soil aggregation and stabilization (Holdo & Mack, 2014; Vermeire et al., 2021).

4.4 Herbivore, fire, and vegetation type

Bulk density was influenced by the interactive effect of herbivores, fire, and vegetation type. With the exceptions of the unburnt plot on the full exclosure within the crest, soil samples collected under tree canopies in the burnt plots of all the exclosures had the lowest bulk density than all other treatments. This is because fire in conjunction with trees tends to attract grazers which deposit their drop**s, dung, and urine beneath canopies. Animal excreta (dung and urine) decreases soil bulk density by increasing organic matter decomposition.

Soil penetration resistance was also largely influenced by herbivory, burning, and vegetation type. Within the crest and foot-slope, the presence of herbivores in the open access area and partial exclosure caused an increase in penetration resistance for samples collected in grassland zones. Holdo and Mack (2014) found that herbivores tend to exert pressure on open grassland zones through the action of trampling, pawing, and wallowing. Pressure exerted on the soil via the trampling action of grazing herbivores increased the proportion of bare soil, thus causing soil compaction (Fig. 3). Fire, herbivores, nor vegetation type influenced penetration resistance in the full exclosure, and this might be a consequence of herbivore exclusion.

Fig. 3
figure 3

Schematic diagram of the main and interactive ecological effect of fire and herbivores on savanna soil and vegetation in Nkuhlu exclosure of the Kruger National Park, South Africa.

Full size image

In the crest, soil samples collected under tree canopies in the unburnt plot of the partial exclosure had significantly higher MWD of water stable aggregates than all other treatments, while the open access area had the least under open grassland zones. Low aggregate stability may be caused by the presence of large and tall herbivores. The open access area is often associated with overgrazed, trampled vegetation, and according to Tuomi et al. (2021) and van Coller et al. (2013), the presence of large animals in the open access area, particularly elephants, causes soil compaction, decreased soil porosity, increased bulk densities, and decreased aggregate stability. One other prominent damage caused by large herbivores is defoliation, leading to a decrease in vegetation height and structural complexity (Fig. 3).

Against all odds, samples collected under tree canopies of burnt plots in the full exclosure within the foot-slope had the lowest MWD of water stable aggregates than all the other treatments. We expected stability to be greater in areas where there was canopy cover and herbivores were absent. This is because according to Parwada and van Tol (2018), a lower degree of soil disturbance improves aggregate stability by binding soil particles into more stable macroaggregates.

5 Conclusion and recommendations

This study indicated interesting feedbacks on the main and interactive ecological impact of herbivory, fire on selected soil physical properties and processes within the crest and foot-slope of the Nkuhlu exclosures. Burning and large mammalian grazers seemed to greatly influence changes in soil physical properties between open grassland zones and areas under tree canopies. These changes had a huge impact on bulk density, penetration resistance, aggregate size distribution, and mean weight diameter of water stable aggregates which may ultimately cause soil compaction, land degradation, and thus loss of biodiversity. Exclosure resulted in significant changes in bulk density in the foot-slope than the crest, with the exclusion of herbivores, resulting in major reductions in bulk density. Therefore, careful consideration should be made when making decisions in the management of fire and exclosures. Further research studies should incorporate a range of fire frequencies and intensities, as well as herbivore densities and abundancies. These need to be more thoroughly determined in order to quantify the impacts of soil crusting and water repellence, resulting from herbivory and fire on plant available water in savannas.