Linkages between soil carbon, soil fertility and nitrogen fixation in Acacia senegal plantations of varying age in Sudan

Introduction

Over the last decades sub-Saharan Africa has experienced severe land degradation and food security challenges linked to loss of soil fertility and soil organic matter (SOM), recurrent drought and increasing population (Nkonya et al., 2015). While soil water availability is the main limitation on primary productivity in drylands, nutrient deficiencies, particularly nitrogen (N), phosphorus (P) and potassium (K), are other important causes (FAO, 2004; Lal, 2004a). SOM plays an important role in maintaining adequate nutrients and moisture levels (Tiessen, Cuevas & Chacon, 1994; Lal, 2004b) and soil fertility management practices that increase SOM contents have been adopted in many drylands in order to enhance crop productivity (FAO, 2004; Koohafkan & Stewart, 2008). The use of a fallow period is a well-known practice in these areas, allowing the soil to restore its SOM content and so recover from years of cultivation (Sanchez, 1999). However, the area of land put under fallow and the duration of the fallow period have been reduced as a result of increasing population pressure (Kaya, Hildebrand & Nair, 2000; FAO, 2004). Other practices aimed at reversing land degradation have focused on the role of trees, particularly N₂-fixing species, in maintaining soil fertility and protecting the soil from wind and water erosion (FAO, 2001, 2004). The deeper roots of trees play an important role in mineral nutrient recycling, enabling mineral nutrients to be taken up from deeper soil layers and making them available to ground vegetation via litterfall—so-called ‘nutrient uplift’ (Scholes, 1990; Ludwig et al., 2004).

Sub-Saharan drylands are characterized by woodland savanna with trees and shrubs forming an open canopy with varying proportions of grasses (Bourlière & Hadley, 1983; Torello-Raventosa et al., 2013). The importance of the facilitative mechanisms (relative to competition) of trees in tree-grass systems has been reported to be greater in drier savanna (Dohn et al., 2013; Moustakas et al., 2013). The positive effects of trees and shrubs on ground vegetation have been attributed to the effect of shade, improvement in soil moisture conditions, and increased nutrients contents under tree canopies (Belsky et al., 1993; Hagos & Smit, 2005; Blaser et al., 2013). Fire in savanna is typical, although varying in frequency and intensity, and generally results in a loss of C and N from the ecosystem (Pellegrini et al., 2015). However, fire may have little effect on soil total N and soil organic carbon (SOC) because of the superficial nature of the fires (Coetsee, Bond & February, 2010; Coetsee, Jacobs & Govender, 2012). Savanna ecosystems are also subject to grazing and browsing, the effects of which on ecosystem biogeochemistry and nutrient fluxes are complex and variable, but maybe significant (Holdo et al., 2007). In open ecosystems, such as savannas, herbivores may bring in significant quantities of nutrients, particularly N and P, in the form of dung and urine (Holdo et al., 2007).

N₂ fixation can increase soil N contents (Ludwig et al., 2004; Blaser et al., 2013). However, N₂ fixation has a high P requirement (Vitousek et al., 2002; Binkley, Senock & Cromack, 2003), which is low in dryland soils due to P adsorption either by iron oxide (Dregne, 1976) or calcium (Lajtha & Schlesinger, 1988). The abundance of stable N isotopes (δ¹⁵N) in leaves and, to a lesser extent, soils can be used to assess N₂ fixation and indicate patterns of ecosystem N cycling (Boddey et al., 2000; Aranibar et al., 2004; Peri et al., 2012). Low foliar and soil δ¹⁵N values indicate biological N₂ fixation (Schulze et al., 1999; Robinson, 2001), while the enrichment of soil ¹⁵N can be attributed to SOM reprocessing by microorganisms (Aranibar et al., 2004; Swap et al., 2004).

The biogeochemical cycles of C, N and P are often closely related (Finzi et al., 2011) and C:N:P stoichiometry is commonly used to provide an insight into the nature of nutrient limitations in ecosystems (Jobbágy & Jackson, 2001; Bui & Henderson, 2013). Soil C:N and C:P ratios are useful indicators of the state of SOM decomposition and N and P availability (Batjes, 1996; Tian et al., 2010), and foliar N:P ratios have been used to assess plant nutrient limitations (Ludwig et al., 2004; Sitters, Edwards & Olde Venterink, 2013; Blaser et al., 2014).

Acacia senegal (L.) Willd. (the new scientific name is Senegalia senegal (L.) Britton.) is a highly drought-resistant tree native to Sudan and Sahel zone of Africa (Obeid & Seif El Din, 1970). Although the new name has been used in a number of recent publications, we have retained the use of the old name, A. senegal, for reasons of consistency with our previous two related articles and with literature in general, and because of the local importance of the old name. A. senegal provides a wide variety of benefits, such as fodder for animals, fuelwood, charcoal and gum arabic (Barbier, 1992). Gum arabic is an exudate collected from A. senegal trees and widely used as an emulsifier in confectionary and beverages, photography, pharmaceutical and other manufacturing industries (Barbier, 2000). This tree is also known to be capable of N₂ fixation under different soil types and climatic conditions (Raddad et al., 2005; Gray et al., 2013). The influence of A. senegal on soil physiochemical properties in arid and semi-arid areas of Africa has been documented in a number of studies (Deans et al., 1999; Githae, Gachene & Njoka, 2011). In Sudan, particular attention has been given to SOC and N contents under A. senegal in the north Kordofan region (Jakubaschk, 2002; Olsson & Ardö, 2002; Ardö & Olsson, 2004; Abaker et al., 2016) and on the influence of inter-cropping systems with A. senegal on soil properties of sandy and clay soils (Raddad et al., 2006; El Tahir et al., 2009).

The aims of our study were to determine the effects of A. senegal plantation age on: (1) soil N (total), P (total and available) and K (total and exchangeable) concentrations, stocks and accretion rates; (2) potential N₂ fixation using foliar δ¹⁵N values; and (3) acacia leaf, ground vegetation N:P ratios and soil C:N:P stoichiometry in order to indicate nutrient limitations, imbalances and cycling in these ecosystems. We hypothesized that soil N, P and K concentrations and stocks would be positively correlated with SOC and increase with planation age, further indicating the benefits of maintaining tree cover in these semi-arid environments. This paper complements two previous papers dealing with effects of A. senegal plantation age on SOC stocks (Abaker et al., 2016) and on soil moisture and water balance (Abaker, Berninger & Starr, 2018). These two studies were carried out at the same sites as in this study.

Materials and Methods

Study sites

We conducted our research at two sites in western Sudan: El Demokeya forest reserve (13°16′ N, 30°29′ E, 560 m a.s.l.), an experimental site managed for gum arabic research, and El Hemaira forest (13°19′ N 30°10′ E, 570 m a.s.l.) owned and managed by farmers for gum arabic production (Fig. 1). At both sites there was an area of open grassland which was taken to serve as a control against which the plantations of differing age were compared. Photographs showing the plantations and grasslands at the two sites during the rainy season are given in Supplementary Material S1.

The long-term mean annual rainfall and temperature for both sites is 318 mm and 27.3 °C. The soils at both sites are classified as Cambic Arenosols (FAO) (≥90% sand). The topography is very gently sloping eastwards at El Demokeya and flat at El Hemaira and the hydrology similar at the two sites. Water balance modelled runoff from the grasslands was 32 and 95 mm for 2011 and 2012 respectively, zero for the plantations in 2011 and 63 mm in 2012 at both sites (Abaker, Berninger & Starr, 2018). Drainage was higher in 2011 than in 2012, and somewhat less at El Hemaira (ranging from 0 to123 mm) than at El Demokeya (ranging from 25 to 128 mm). The vegetation at both sites falls within the low rainfall woodland savanna type (Ayoub, 1998; FAO, 2006). Main components of the ground vegetation at both sites were grasses such as Cenchrus biflorus, Aristidia pallida and Eragrostis tremula, and some herbs, including Geigeria alata, Justicia kotschyi, Trianthema pentandra and Acanthus spp. A complete list of ground vegetation species found at the two sites in given in Supplementary Material S2. Although site-specific information about grazing and frequency of fire at the two sites is unavailable, it is known that there is over-grazing by sheep and browsing by camels, even within the forest reserve at El Demokeya. Additional information about the study sites and sampling have been described in Abaker et al. (2016).

Results

All nutrient concentrations were generally higher in the plantations than in the grasslands, increased with plantations age and decreased with depth (Fig. 2). Concentrations of SOC, N, total P and K_ex in the top (0–10 cm) layer were significantly (p ≤ 0.05) higher in the oldest plantations at both sites compared to the grassland plots. Soil concentrations of total N, P, P_av and K_ex also significantly depended on SOC concentrations (Fig. 3). The strongest dependence was for N (R² = 0.90) and the weakest was for total K (R² = 0.11). The correlations between SOC and N concentrations were significant for all layers at both of the sites (Table 1). The correlations between SOC and total P concentrations were significant for all layers at El Hemaira but in the case of El Demokeya the correlation was significant only for the top layer. The correlations between SOC and total K concentrations were stronger at El Hemaira than at El Demokeya. In the case of P_av and K_ex, significant correlations with SOC were associated with the upper layers.

Nutrient stocks in the soil of the plantations were generally greater than those in the grassland and increased with plantation age (Table 2). As the fixed depth SOC stock values showed better relationships with SOC concentrations and with plantation age than did ESM SOC stock values, only the fixed depth stock SOC and nutrient values are presented in Table 2 and handled further. However, the ESM SOC and nutrient stock values are presented in Supplementary Material S3. At El Demokeya SOC, N, total P and K_ex stocks were significantly higher in the oldest plantation than those in the grassland, but not P_av and total K stocks. At El Hemaira SOC, N and K_ex stocks were also significantly higher in the oldest plantation than in the grassland. K_ex stocks in the 15-year-old plantation were also significantly higher than in the grassland. Assuming that the significant difference between grassland and the oldest plantation SOC, N and total P stocks represents the addition of these elements brought about by the effect of the plantation, the average under canopy accretion rates of SOC and N at El Demokeya would be respectively 12.9 and 2.0 g m⁻² yr⁻¹. At El Hemaira, the corresponding SOC and N accretion rates would be 27.8 and 3.0 g m⁻² yr⁻¹. The total P accretion rate at El Demokeya would be 0.5 g m⁻² yr⁻¹ (as the difference in total P stocks between the grassland and oldest plantation at El Hemaira was not significant, the accretion rate is considered zero).

Grassland ground vegetation and soil δ¹⁵N values were generally lower than corresponding plantation δ¹⁵N values. Acacia foliar δ¹⁵N values were higher than ground vegetation values, but neither showed a difference related to plantation age (Table 3). The number of ground vegetation samples was too small to allow for significance testing. Soil δ¹⁵N values increased with plantation age and decreased with depth at both study sites, but these trends were not significant (p > 0.05). Plantation soil δ¹⁵N values were significantly correlated to soil C:N ratios, but the relationship for grasslands was clearly different (Fig. 4).

Soil C:N:P ratios did not significantly differ with depth and the N:P ratios only showed significant differences between plantation age for the 0–10 cm layer (Table 4). The 0–10 cm soil layer C:N and C:P ratios did not show significant differences with age at either of the sites. At El Demokeya, the 0–10 m soil layer N:P ratio in the 24-year-old plantation was significantly (p < 0.05) higher than those in the grassland and 15-year-old plantation. At El Hemaira, the 0–10 cm soil layer N:P ratio in 20-year-old plantation was significantly greater (p < 0.05) than those in the grassland and 7-year-old plantation. Acacia leaf nutrient concentrations and N:P ratios did not show significant differences related to plantations age at either of the sites (Table 5). There were too few ground vegetation samples for statistical testing of nutrient concentrations and ratios.

Discussion

In this study we aimed to determine whether the previously reported increase in SOC contents with plantation age at the two sites (Abaker et al., 2016) would also result in higher nutrient (N, P and K) concentrations and stocks, which would further support the importance of maintaining or increasing tree cover in the region. In an earlier paper, we showed that the increases in SOC with plantation age at the two sites resulted in increased available water capacities which then had an effect on the water balance of the plantations (Abaker, Berninger & Starr, 2018). Because of the pseudoreplicated design of our study, general patterns about plantation age effects may not be strictly inferred. However, given the inevitable within site variation in site conditions, the climate, soil type and topography were uniform across each site and the replicate three plots for each treatment (grassland and plantation age) were located so as otherwise to be as similar and comparable as possible. Unfortunately, documented information about land-use prior to the establishment of the plantations at the two sites was not available. However, from discussions with local staff, the A. senegal plantations were established on areas of homogenous abandoned grassland.

Recognising the potential limitations imposed by the pseudoreplicated design, the significant dependence of nutrient concentrations on SOC and the significantly higher N and K_ex stocks in the oldest plantations compared to the grasslands support our initial hypothesis that soil N, P and K are linked to SOC and are in agreement with results reported from other studies. For example, in A. tortilis savanna woodlands in northern Tanzania Ludwig et al. (2004) found increases in SOM, N, P and P_av concentrations with tree growth stage (grassland, under small and large trees), and Deans et al. (1999) working with A. senegal in Senegal found that N and K_ex, but not P concentrations, increased with plantation age. In both these studies, the soil refers to the surface layer (0–10 cm). This layer had the highest SOC contents and would therefore be expected to be the most affected by the plantations. Furthermore, in the study by Deans et al. (1999), soil concentrations of N, P and K_ex were all significantly correlated to loss-on-ignition contents, i.e. SOC contents. El Tahir et al. (2009) working at El Demokeya site, reported a SOC stock value of 738 g m⁻² for 0–30 cm layer and for total N, P_av and K_ex values of 118, 2.5 and 29 g m⁻², respectively. We were unable to take into account the effect of fire and grazing on soil SOC and N stocks at our study sites. However, fire has generally been found not to result in a loss of soil total N and SOC because of the superficial nature of the fires (Coetsee, Bond & February, 2010; Coetsee, Jacobs & Govender, 2012). The effect of grazing at our study sites is discussed below in relation to soil N stocks.

Compared to the grasslands, the higher N, P_av and K_ex concentrations observed in the upper soil layer of the plantations indicates a significant effect of acacia trees on ecosystem nutrient cycling, at least at our study sites. The higher concentrations in the surface layer was particularly obvious in the older plantations and can be explained by ‘nutrient uplift’ by the deeper roots of the acacia trees (Scholes, 1990; Ludwig et al., 2004). Mubarak, Abdalla & Nortcliff (2012) also concluded that tree litter input is a significant source of P and K in southern Kordofan soils and the presence of trees has been shown to contribute to the general maintenance of soil fertility in the Sahel (Wezel, Rajot & Herbrig, 2000; Schlecht et al., 2006).

The higher N concentrations in the surface soils of the plantations may be thought to be due to N₂ fixation as acacia species are considered to be N₂ fixing (Ludwig et al., 2004; Raddad et al., 2005; Boutton & Liao, 2010). Although, A. senegal has been reported to be a N₂ fixer (Raddad et al., 2005; Isaac et al., 2011; Githae et al., 2013; Gray et al., 2013) the high δ¹⁵N values we observed for acacia leaves (>6‰) would indicate that A. senegal did not fix N₂ or is very limited in our sites. If there had been significant N₂ fixation in the plantations then one would expect foliar δ¹⁵N values to be closer to 0‰ (Robinson, 2001; Aranibar et al., 2004; Nardoto et al., 2014). Nevertheless, our acacia foliage δ¹⁵N values are in agreement with the findings of other studies conducted in arid environments. For example, Aranibar et al. (2004) observed that Acacia leaves had δ¹⁵N values similar to non-legume species and even higher than known N₂-fixing species in a study carried in the Kalahari Desert. Pate et al. (1998) reported a mean δ¹⁵N value of 9.10‰ for Acacia species in arid Australia, which was identical to those of non-fixing woody species, suggesting little or absence of N fixation. In a study carried out in A. tortilis savanna woodlands in Kenya, Belsky et al. (1993) concluded that that N₂ fixation was not an important contributor of N to the soil. N₂ fixation by legume trees in drylands has been show to vary considerably, even within the same species (Nygren et al., 2012). For example, N₂ fixation by A. senegal growing on clay soil in Sudan was shown to vary from 29 to 48 kg N ha⁻¹ (Raddad et al., 2005).

Our soil N accretion rates in the plantations appear high but are comparable to those reported by Blaser et al. (2014) of 1.3–2.0 g N m⁻² yr⁻¹ (for 0–10 cm layer) in Zambian savanna. However, the vegetation at their site was dominated by the N₂-fixing shrub Dichrostachys cinerea. As deposition loads of N in the Sahel are about 0.3–0.7 g N m⁻² yr⁻¹ (Delon et al., 2010), our high N accretion rates cannot be explained by deposition. The paradox between the accumulation of soil N in the absence of N₂ fixation and sufficient N deposition in humid tropical forests has been identified in several studies (see Hedin et al., 2009) and has been explained by heterotrophic N₂ fixation by free-living bacteria decomposing litter and SOM (Vitousek & Hobbie, 2000) or by canopy epiphytic N₂ fixation (Hedin et al., 2009). However, the rates of such N₂ fixation are low and could not explain our high soil N accretion rates. A possible source of our observed high soil N accretion rates could be from grazing animal excretion. The two study sites are not fenced and seasonal pastoral and nomadic grazing (mainly sheep) and browsing (camels), although varying, takes place throughout the study area (Poussart, Ardö & Olsson, 2004). Bigger trees (older plantations) may be expected to provide increased shading and ground vegetation for grazing and browsing. Animals entering the plantations may therefore have added N to the soil in the form of animal excretion derived from grazing outside and in excess of grazing removals from inside the study sites. Studies on elk and bison in north-temperate grassland indicate that herbivore excretion can add significant amounts of N to the soil (Frank et al., 1994). However, data on land-use history and animal herbivory at the two sites is not available and therefore this animal excretion N explanation is only speculative.

It has been shown that N-fixing trees accumulate large amount of N-rich litterfall during the first years of establishment, however once N availability has built up in the soil, N fixation may be ceased or inhibited (Khanna, 1998; Boddey et al., 2000; Hedin et al., 2009) and the older trees/plantations become more dependent on litterfall and N recycling (Deans et al., 2003). The relatively high and increasing trend in soil δ¹⁵N with plantation age at our sites may be an indication of greater microbiological processing of SOM and a more open N cycle (ammonia volatilization and denitrification during the wet season) resulting in an enrichment of ¹⁵N (Aranibar et al., 2004; Swap et al., 2004; Hobbie & Ouimette, 2009). The negative relationship observed between soil δ¹⁵N and soil C:N ratios in the plantations is consistent with the notion that low soil C:N ratios in arid environments promote greater N gaseous losses (Austin & Vitousek, 1998; Aranibar et al., 2004; Saiz et al., 2016). The vegetation present at a given site exerts a large influence on SOM dynamics not only because of the quantity and quality of organic matter returning to the soil (Saiz et al., 2015), but also because of its impact on soil hydrological conditions (Abaker, Berninger & Starr, 2018). In this regard, trees growing on coarse-textured soils in semi-arid regions may promote the maintenance of soil water conditions suitable for the activities of SOM decomposers through the interception and funnelling of rainfall by their canopies and the reduction in soil water evaporation by shading (Bargués Tobella et al., 2014; Ilstedt et al., 2016). Two recent works have shown potentially faster SOM decomposition rates at locations dominated by trees compared to those dominated by grass vegetation in mixed C₃/C₄ systems occurring on coarse-textured soils (Saiz et al., 2015; 2016). These vegetation-related factors may be responsible for the higher SOC and nutrient contents observed in our acacia plantations. The higher soil δ¹⁵N values observed with plantation age is further evidence of SOM decomposition processes being comparatively more dynamic under the direct influence of trees.

Cyanobacteria associated with the formation of cryptogamic soil crusts have been shown to be a significant pathway to fix atmospheric N₂ in arid environments, but their development diminishes with vegetation cover (Aranibar et al., 2004; Wang et al., 2013). Therefore, N fixation by cyanobacterial soil crusts (which may be expected to be more strongly developed in the grasslands) may explain the low soil δ¹⁵N values observed in grassland sites. However, we have no information on the presence and development of such cyanobacterial soil crusts at our sites, but in any case annual N fixation rates associated with cyanobacterial soil crusts are very low (Aranibar et al., 2003).

The decreasing rather than increasing trend in soil δ¹⁵N with depth observed in both the grasslands and plantations is somewhat unusual (Hobbie & Ouimette, 2009), but it has also been observed in an arid, sandy site in the Kalahari (Wang et al., 2013). The variation in soil δ¹⁵N values with depth are the result of multiple interacting factors, which include N inputs by plant and cryptogamic crusts, vertical transport processes (i.e. leaching, fungal immobilization and bioturbation), soil moisture conditions, and isotopically fractionating processes (e.g. ammonia volatilization and denitrification) (Hobbie & Ouimette, 2009; Wang et al., 2013; Saiz et al., 2016). However, as the N contents in our soils are very low resulting in a low analytical signal for ¹⁵N, our soil δ¹⁵N results should be interpreted with caution.

Soil C:N ratios often decrease with soil depth as a result of the SOM being older and more decomposed and therefore relatively enriched in N compared to SOC (Batjes, 1996; Tian et al., 2010). However, there was no consistent trend in C:N ratios with depth in either the grasslands or the plantations at our sites, which may be explained by gaseous losses of N as indicated by the soil δ¹⁵N values and discussed above. The significantly lower soil (0–10 cm) N:P ratios in the grasslands than in the oldest plantations at our sites would indicate N limitation in the grasslands. The N:P ratios of the ground vegetation were on the lower side of values presented for savanna grasses by Ludwig et al. (2004) and Sitters, Edwards & Olde Venterink (2013). Ludwig et al. (2004) considered low grass N:P ratios from open grasslands to indicate N limitation and higher values for grasses sampled from under the canopy of trees to indicate P-limiting conditions for the grasses. Sitters, Edwards & Olde Venterink (2013) similarly concluded that their observed increase in grass N:P ratios with tree density indicated a shift towards P-limiting conditions for the ground vegetation.

Conclusion

The concentrations of all studied nutrients were relatively low but directly and significantly correlated to SOC, were highest in the topsoil and increased with plantation age at our sites. Although these results are specific to our study sites, we consider these results support our hypothesis that soil N, P and K contents in the Sahel region are strongly controlled by SOM (SOC) contents. Although A. senegal is known to be capable of N₂ fixation and may have occurred when the trees were young, current foliar δ¹⁵N values did not indicate ongoing N₂ fixation in the plantations. The soil N accretion rates observed in the plantations were unlikely to be due to N deposition but may be related to inputs of excreted N brought into the area annually by grazing and browsing animals. The relatively high surface soil N contents in the plantations at our sites were considered to be the result of litterfall and recycling. The higher total and plant available contents of P and K in the soil surface of the plantations may be an indication of ‘nutrient uplift’ by the deeper roots of the acacia trees. Soil N:P ratios indicated N limitation in the grasslands and a trend towards P-limitation in the plantations. Our results support the notion that an increase in SOM (SOC) contents related to the retention and preferably planting of trees in the Sahel region would not only increase carbon sequestration, but also significantly improve soil fertility.

Supplemental Information