Woody plant encroachment has been increasing worldwide for centuries at varying rates across different continents. Moreover, this has led to an alarming negative effect on the grassland-savanna biome (Sala and Maestre, 2014; Stevens et al., 2016; Archer et al., 2017; García Criado et al., 2020). Not fully acknowledging the role of human activities has led to a significant loss of the grassland-savanna biome (Ripley et al., 2022). Ecosystem degradation due to woody plant encroachment is a widespread issue in savanna ecosystems and requires thorough scientific investigation by rangeland ecologists (Belay et al., 2013; Ellis et al., 2019). The rise of woody plant encroachment in open savannas decreases grazing carrying capacity and subsequently affects grazing animals (Kimaro et al., 2019; Hare et al., 2020). This is because increased woody plant encroachment is typically characterized by thorny and unpalatable species, which are not favored by browsing animals (Kgosikoma and Mogotsi, 2013). Thus, in some areas, the savanna biome has transformed into a thicket-like biome due to the encroachment of woody plant species (Kimaro et al., 2019). The rapid encroachment of woody plants into savanna rangelands has been reported as a significant indirect threat to profitable beef enterprises based on natural rangelands (Dalle et al., 2006; Lunt et al., 2010).

Several studies on woody plant encroachment have previously focused on biotic and abiotic factors, either individually or in combination, to understand the main contributors to this phenomenon (Holdrege et al., 2021). The areas most impacted by the encroachment of woody plant species are rangelands primarily utilized as grazing lands for livestock and/or game animals, owing to their grass cover and tree species that are acceptable for grazing and browsing animals (Kgosikoma and Mogotsi, 2013). Consequently, the quality of rangelands has deteriorated because of overgrazing, inadequate management practices, and the impacts of climate change (Ratajczak et al., 2011). Alterations in vegetation structure and species composition impact the sustainability of livestock production and ecosystem services (Mussa et al., 2016). Additionally, rangeland degradation due to bush encroachment seems to weaken ecosystem integrity and reduce grazing capacity (Sankaran et al., 2008). Once 75% of the total of the savanna rangeland is covered by woody plant species, that rangeland is declared bush encroached (Oba et al., 2000; Bond and Midgley, 2001; Kellner et al., 2022).

Encroaching woody plant species may include native trees and/or shrubs that inhabit grasslands and savanna grazing areas (Kgosikoma and Mogotsi, 2013). In South Africa, bush encroachment has reached a scale that demands scientific intervention (O’Connor and Chamane, 2012). Kraaij and Ward (2006) reported that the extent of bush encroachment in South Africa has reached an alarming level, with approximately 20 million hectares of land declared as bush encroached by woody plant species. Additionally, Higgins and Scheiter (2012) and Moncrieff et al. (2014) reported similar findings, indicating that bush encroachment is one of the most recognized rangeland issues in South Africa, particularly in the savanna regions. This problem is likely to escalate further over time, depending on the measures taken to reverse its aggressive inversion in savanna regions. As part of restoration programs, new technologies have been integrated with traditional rangeland techniques used in the long-term monitoring of these areas. Hudak and Wessman (1998) demonstrated a technique for mapping bush densities across a savanna landscape using historical aerial photographs from 1955, 1970, and 1984 to understand the factors that could be driving bush density in savanna rangelands.

However, restoring and rehabilitating areas affected by bush encroachment require advanced technologies to monitor changes in vegetation, especially given the impact that bush encroachment has on savanna rangelands (Maphanga et al., 2022). Currently, rangeland assessment and monitoring primarily rely on comparing the existing species composition to an expected successional end-state defined by the ecological condition of the site (D’Odorico et al., 2012). For instance, historically, rangeland assessment and monitoring have primarily relied on fieldwork-based surveys, which are often time-consuming, costly, and occasionally inaccurate (Mutanga et al., 2016). These methods frequently require specialized technicians to perform laboratory analyses of plant and soil chemicals in order to assess forage quality. (Zhao et al., 2016; Zhang et al., 2016). Therefore, an integrated assessment that utilizes satellite remote sensing techniques along with fieldwork methods is essential (Jindo et al., 2021). These techniques aim to offer significant guidelines for monitoring savanna rangelands affected by woody plant species and to ensure sustainable management that enhances savanna rangeland productivity. Therefore, it is essential to have skills in remote sensing tools and ecology in order to provide valuable support to policymakers. This enhances the savanna’s carrying capacity and investigates the dynamics of woody plant species and ecosystem properties in a cost-effective manner.

The study focuses on assessing woody vegetation to analyze the effects and dynamics of bush encroachment in savanna rangelands. This is to examine the changes in vegetation that lead to bush encroachment and its relationship with understory establishment. Subsequently, the data from this study are also valuable for validating the predictions of existing models that forecast savanna rangeland productivity after bush encroachment. Moreover, despite the negative ecological impacts of bush encroachment, there have been very few studies examining how woody plant encroachment affects the successful development of the understory woody plant layer and its relationship with mature trees (Archer et al., 2017). Thus, this study aims to highlight the intraspecific interactions among woody plant species in the Marikana Thornveld.

The study was conducted at the Roodeplaat Experimental Farm (25°36′S, 29°2’08′E, altitude 1,182 m) of the Agricultural Research Council (ARC), which encompasses over 2,067 hectares of natural vegetation outside Pretoria, South Africa ( Figure 1 ). The average rainfall is 687 mm, with the most rainfall occurring during austral summer (November to February). The daily maximum summer temperature ranges from 20°C to 29°C, while the winter temperature can range from 2°C to 16°C. The soil types in the study area consist of vertisols, ferralsols, and luvisols. The vegetation type of the study site is Marikana Thornveld, characterized by open Vachellia karroo and Senegalia caffra veld types (Mucina and Rutherford, 2006). The farm is primarily dominated by the tree species Vachellia karroo, Senegalia caffra, Ziziphus mucronata, and Combretum zeyheri. The most dominant grass species include Panicum maximum, Setaria sphacelata, Digitaria eriantha, Themeda triandra, Heteropogon contortus, and several weak annual species such as Aristida congesta subspecies congesta, and subspecies barbicollis (van Rooyen, 1983). On the farm, three sites were designated to evaluate the composition of woody species and the changing patterns of these species on the farm. All three sites were marked within heavily encroached areas where the density of woody plants exceeded the acceptable threshold of 2,500 plants per hectare, displaying a similar species composition. Invasive alien plant species are found in high densities in localized areas, especially along drainage lines. The common invasive plants in the study area include Tagetes minuta, Bidens pilosa, Lantana camara, and Pontederia crassipes (Rutherford and Powrie, 2012). The study site has only been grazed and browsed by a very small number of antelopes, such as Tragelaphus strepsiceros.

A total of 9,028 woody plant individuals were recorded, including 158 seedlings, 161 saplings, 159 shrubs, and 154 mature woody plants, in three sites measuring 320 m × 223 m during the 2020–2022 woody plant species survey ( Table 2 ). The relative abundance of adult trees (mature trees and shrub plant species) accounted for 16% of the variation in the abundance of young woody plant species (saplings and seedlings), with r² = 0.16, p ≤ 0.001 ( Figure 3B ). However, when assessing the relationship between seedling abundance and mature trees at the subplot level, the results indicated a weak correlation between seedling abundance and the abundance of mature woody plant species ( Figure 3A ). Generally, the slope of the regression line and the correlation coefficient for seedling abundance had no impact on mature tree species abundance ( Figure 3A ). However, when compared to adult trees (mature trees and shrub density) and young trees (sapling and seedling density), as shown in Figure 3B , there was a slight increase in the relationship among the stages of tree growth. The number of mature trees in the Marikana Thornveld represented only 4.75% of the seedling abundance (r² = 0.0475, df = 134, p < 0.0001). However, the regression line fitted for the log-transformed data was significantly less than 1, with a b value of 0.1535 representing the negative effect of mature woody plants on seedling abundance. The slope of the regression line was significantly less than 1, indicating that the number of seedlings per mature tree declines with increasing reproductive mature tree abundance ( Figure 3A ).

The slope of the regression line for the log-transformed data for woody plant density was significantly less than 1, with b values of 0.58 (r² = 0.58), 3.23 (r² = 0.61), and 0.64 (r² = 0.39) for per capita seedling abundance for mature tree, sapling, and shrub abundance, respectively, at a 95% confidence interval ( Figure 4 ). This indicated that as the abundance of mature trees, saplings, and shrubs increased, there was a general decline in seedling abundance ( Figures 4a, c ). However, the b value shown in Figure 4b was greater than 1 (b = 3.23, r² = 0.61), indicating a positive effect of per capita seedling abundance on the abundance of saplings’ woody plant species. Furthermore, there was a positive correlation between seedling abundance and mature tree, sapling, and shrub abundance, with R values of 0.75, 0.78, and 0.68, respectively ( Figure 4 ). Figure 5 demonstrates a very poor relationship between seedling density and mature tree, shrub, and sapling density. The results in Figure 5 illustrate that the assessed relationship exists at a site level. However, the overall findings indicate that mature tree, shrub, and sapling density has no effect on seedling establishment.

According to both Fisher’s alpha values and estimates of diversity based on rarefaction, tree species diversity was lower in the shrub layer compared to all other growth stages. However, the seedling layer recorded the highest Fisher’s alpha index. This indicates that, even though the number of individuals in the seedling layer was low, the ratio of the total number of individuals to the number of species exceeded 1, which reflects a completeness of taxa and individuals. Although species evenness did not differ significantly, a noticeable decline was observed in the shrub layer ( Table 2 ).

H’ and evenness (J) were used to evaluate the diversity of woody species at the study site across various growth stages. The results in Table 3 also showed that the shrub and sapling layer had high species diversity compared to that of mature trees and the seedling layer, with 5.05 and 5.12, and 4.89 and 4.75, respectively. Compared to adult trees, the seedling layer showed moderate to low species richness and evenness (Rutherford and Powrie, 2013) ( Table 3 ).

Woody plant density exceeding 2,400 TE ha^-1 or having woody plant cover greater than or equal to 40% of the total area is considered bush encroached (Stevens et al., 2017) and has the potential to pose significant environmental risks, particularly to the composition and structure of natural vegetation (Tiawoun et al., 2022). Other research results indicate that an area with a woody plant density exceeding 2,500 TE ha^-1 reflects a moderately encroached condition (Richter et al., 2009; Abate et al., 2012). Furthermore, our findings are supported by Gemedo-Dalle et al. (2006), who indicated that an area with a woody plant density exceeding the 2,400 ha^-1 threshold is considered bush encroached. In our study, the total density of woody plants across all species and growth stages surpassed 2,400 TE ha^-1, indicating that the Marikana Thornveld is significantly encroached upon (O’Connor et al., 2014). The study results indicated a high density of woody plant species, primarily Ehretia rigida at 7,233 plants per hectare, followed by Dichrostachys cinerea with 6,633 plants per hectare and Combretum zeyheri with 6,133 plants per hectare. Therefore, the elevated density values of these woody plant species suggest that their ecological impact in the Marikana Thornveld is significantly problematic. Ayele et al. (2024) reported similar results in Church Forest, documenting a total of 5,923 saplings and 6,136 seedlings per hectare, along with 83 and 66 woody plant species recorded, respectively. The data on species abundance and density varied significantly by growth stage and individual woody plant species. These differences may stem from various factors, including the environmental adaptations of specific species, population pressure, and management practices, such as the underutilization of woody shrub species by browsing animals (Tiawoun et al., 2022a).

Additionally, when we grouped identified woody plant species by their ecological growth stages, our results showed the highest density of woody plants for saplings at 9,978 ha^-1, followed by shrub species at 6,256 ha^-1, mature trees at 3,322 ha^-1, and seedling species at 3,136 ha^-1. Consequently, the high abundance of saplings and shrub species clearly indicates that the Marikana Thornveld is in a state of regenerating woody vegetation. Thus, as a component of bush management, applying fire is frequently suggested for controlling bush, particularly in the shrub and sapling layer (Balch et al., 2013). However, for successful burning to control the understory bush layer, at least ≥4,000 kg ha^-1 of dry herbaceous material is required to ensure the efficacy of burning (Trollope 2004). Ward (2005) explains the possibilities that lead to occurrences where these changes are realized when resource allocation and certain management practices, such as fire, interact. Furthermore, numerous studies have linked the encroachment of woody plant species to changing fire regimes, overgrazing, soil moisture, nutrients, and global climate change (Ward, 2005; Bond, 2008; van Auken, 2009; Stevens et al., 2016). The prolonged absence of fire promotes woody plant encroachment by providing a secure environment for the development of woody plant seedlings and saplings into fire-resistant growth stages (Trollope, 1980; Bond et al., 2003). Therefore, it is not surprising that our results show a high density of woody plant species, particularly shrubs and saplings. This is due to the fact that prescribed fire applications at our site have been recorded since 2010, resulting in more than 10 years without fire as a control tool for bush encroachment. Additionally, the rapid increase in woody plant species has negative effects that lead to a decline in herbaceous plant species (Angassa and Oba, 2008; Devine et al., 2017) due to competition for available resources among grasses and woody species (Strohbach et al., 2015). Woody plants and herbaceous vegetation coexist; however, once the threshold of 2,500 woody plants per hectare is exceeded, the resulting changes completely suppress grass growth and establishment (Sankaran et al., 2005; Tiawoun et al., 2022b). This creates a hostile environment for herbaceous plants to access water and nutrients in the soil profile (Boldrin et al., 2022).

Both of these reported woody plant species encroachment thresholds (2,400 plants ha^-1 and 2,500 plants ha^-1) (Gemedo-Dalle et al., 2006) were lower than the density values reported in our research findings. The results for both saplings and shrub species showed high woody density values, which could be due to the fact that woody plants of their height can evade natural disturbances such as trampling and competition (shading, nutrient availability, etc) (Muturi et al., 2013). Our research findings indicate that all the surveyed sites experienced bush encroachment, with several dominant woody species including Ehretia rigida, Dichrostachys cinerea, Gymnosporia buxifolia, Combretum zeyheri, Pappea capensis, Lantana camara, and Vachellia robusta. These findings align with the research conducted by Malan et al. (2021) on wood plant densities, particularly concerning the two most encroaching species: Senegalia mellifera and Dichrostachys cinerea. Consequently, the increase in woody plant density leads to greater complexity in ecosystem functions and a transition from a simple single-stratum to a multiple-stratum ecosystem (Archer, 1998).

The abundance of adult plants represented approximately a quarter or less of the total seedling abundance in the Marikana Thornveld. Seedling abundance showed no positive relationship with adult tree abundance; consequently, adult tree abundance had a negative effect on the number of seedlings per reproductive adult in the Marikana Thornveld. The poor representation of seedlings beneath the adult tree layer can be attributed to seed predation and seedling mortality (Holík and Janík, 2022). The abundance of seedlings and saplings in African savannah ecosystems is the primary factor contributing to bush thickening (Chidumayo, 2013).

The species diversity index includes both species richness and the abundance of the identified woody species. The diversity of woody species plays a crucial role in conserving woody plant species as it guides ecological processes, management efforts, and the protection of specific species (Wakjira, 2006). The species diversity in the Marikana Thornveld, Roodeplaat, South Africa, demonstrated a wide range of woody plant species. This veld type includes mature trees, shrubs, saplings, and seedlings. There were differences in the Shannon–Wiener diversity index between mature trees and seedlings, with values of H′ = 4.89 and H′ = 4.75, respectively, and between shrubs and saplings, with H′ = 5.05 and H′ = 5.12, respectively. Lower diversity values were noted for seedlings, possibly due to seed predation by birds and/or insects before germination. Consequently, there is low or no establishment of various new tree species in the seedling layer (Abunie and Dalle, 2018). Moreover, low species diversity in the seedling layer may result from poor seed viability in the soil bank. However, greater woody species diversity was noted in the shrubs and saplings compared to the mature and seedling layers (Wassie et al., 2010). Only shrub species exhibited low species evenness, which was due to a high diversity component (Rutherford and Powrie, 2013) ( Table 3 ).

Tree species grow and develop until they reach a stage of reproduction, and as they increase in size, both the DBH classes and woody density also change (Kebede et al., 2016). In all growth stages observed in this study across different density classes, a J-shaped pattern emerged. This pattern can be attributed to several factors, such as anthropogenic activities like logging, senescence, diseases, and fire suppression, particularly affecting adult trees (Tebabala et al., 2024). However, for seedlings and saplings, the J-shaped pattern can be attributed to factors such as trampling, competition for sunlight, and heavy browsing by herbivores. Additionally, regenerating ecosystems can develop varying density-class distributions due to changes in specific characteristics of the secondary ecosystem environment (Yigeremu et al., 2022).

DBH class increased with larger DBH size classes across all woody plants in the four growth stages. This pattern indicates a normal population and a healthy ecosystem environment (Gurmessa et al., 2012). Additionally, Senbeta et al. (2014) reported that this shape demonstrates an ecosystem with a positive potential to regenerate and recruit normally. Therefore, the DBH classes for all woody species across different growth stages suggest that the ecosystem environment still has remnants to support regeneration, even as bush encroachment accelerates.