Nutrients in an African Savanna: the consequences of supply heterogeneity for plants and animals

In savannas, trees and grasses co-exist and share resources such as water and nutrients. The ratio between the tree and grass components (i.e., vegetation structure) importantly controls productivity, animal assemblages and earth-atmosphere feedbacks. As the structure of savanna vegetation is inherently unstable and easily disturbed, finding out how the vegetation structure is controlled is of great importance for rangeland management and conservation. Currently four factors are believed to control the vegetation structure in savannas; namely, water, nutrients, herbivory and fire. While the water and fire factors have been intensely studied, the roles of nutrients and herbivores as factors are less well known. Improving our understanding of the role of nutrients in savannas is relevant, because it is increasingly realised that global change also alter the availability of nutrients, e.g., increased nitrogen deposition. How savanna systems respond to nutrient perturbations is uncertain. Changes in soil nutrient availability are also likely to feed back to changes in forage quality, which may influence large herbivore use and vegetation impact patterns, thus indirectly influencing vegetation structure. Moreover, it is increasingly realised that not only changes in the availability of nutrients influence plants and subsequently herbivores, but also how nutrients become spatially available. In fact, new ideas suggest that differences in the scale of spatial resource heterogeneity may control how resources are partitioned between co-existing species differing in size, e.g., large trees coexisting with small grasses in savanna systems. To test how changes in nutrient availability and spatial configurations influence savanna systems, several field experiments were conducted in a semi-arid savanna in South Africa. In the study area we found that nutrient (N, P and K) availability negatively affected tree (Colophospermum mopane) seedling establishment in fertilization experiments. Also, trees failed to re-colonize nutrient-rich kraal sites that were abandoned almost half a century ago. In dry savannas, it is currently believed that the success of tree seedling establishment exercises large control over the relative dominance of trees, thus an increase in nutrient availability may feed back to a structurally more open vegetation state. Different explanations may account for constrained tree seedling establishment under fertile soil conditions. We tested the hypothesis that the intensification of grass competition reduces tree seedling recruitment in fertile environments. In controlled competition experiments it was shown that negative nutrient effects on tree seedlings only occur when seedlings were competing with grasses in mixtures. Furthermore, we found that adding both water and nutrients to tree seedling-grass plant mixtures cancelled the negative effects of added nutrients on tree seedlings. Thus the suppressing effect of increased nutrient availability on seedlings appears to operate indirectly through the pre-empting of soil water resources by vigorous herbaceous growth under fertile conditions. Since woody seedlings are vulnerable to water stress, increased nutrient-induced water stress translates into higher mortality rates and suppressed growth of tree seedlings in fertile areas. In support, transplanted tree seedling mortality increased during a mid-season drought as local fertilizer concentration increased. In conclusion, intensified herbaceous competition under fertile soil conditions appears to be a viable mechanism explaining poor tree seedling recruitment under fertile soil conditions. Savannas and nutrients While establishing tree seedlings suffer under fertile conditions, our data suggest that established (mature) tree growth benefitted from an increase in nutrient availability, especially following an increase in N availability. With regards to increased atmospheric N deposition, we predict that tree cover may initially expand following nutrient enrichment in dry savannas, although tree cover responses may be insensitive to current levels of N deposition. However, in the long-term tree cover is expected to decline, because of constrained tree recruitment that appears to be more sensitive to small increases in N availability. The relative availability of nutrients such as N and P may also influence how resources are partitioned between co-existing trees and grasses. In a field experiment we found evidence that trees were relatively more limited by N than P availability. An East African study has shown that grasses underneath trees are more P than N limited and we found evidence that the competiveness of sub-canopy grasses in our study increased when only P was supplied. The relative availability of N vs. P may therefore offer an additional axis governing resource partitioning between trees (non N-fixing) and grasses in savanna systems. This supports the idea that organisms with a high growth potential, apparently grasses in savannas, have relatively high P requirements to sustain rapid protein synthesis, while slower growing organisms such as trees are more limited by the availability of N. The availability of soil nutrients strongly mediated where large herbivores concentrated their impact in the landscape. Both grazers and browsers responded positively to fertilization, apparently via the control that soil nutrient availability has on forage quality. Tree and grass leaf N and P concentrations increased and condensed tannin concentrations in trees decreased following fertilization. Under high local soil nutrient concentrations the vegetation biomass was in some instances reduced below control biomass by herbivores, indicating that top-down herbivore effects potentially override bottom-up nutrient effects under fertile conditions. In ecology, it is increasingly realized that it is not only the availability of nutrients, but also how nutrients become spatially available that matters. Data from a large field experiment where a gradient in the scale of nutrient patchiness (i.e., patch grains sizes 2 x 2 m, 10 x 10 m or 50 x 50 m) was created, suggested that the scale of nutrient patchiness controlled the partitioning of resources between co-existing trees and grasses. For the same local fertilizer concentration, tree leaf quality was unresponsive to fertilization in small patches, but responded in the larger patches. Grass leaf quality increased with local fertilizer concentration regardless of patch size. The differential responses of trees and grasses to scale differences subsequently modulated the responses of the browser and grazer guilds. For the same high local fertilizer concentration, grazers responded to both fine and coarse scales of nutrient patchiness, while browsers responded only to the coarse scale of nutrient patchiness. In turn, the selective grazing in the fine scale of nutrient patchiness treatment, apparently stimulated tree growth. In the coarser scale of patchiness treatment both browsers and grazer impact intensified locally. Thus the scale of nutrient patchiness controlled nutrient partitioning between trees and grasses, which was apparently closely tracked by the large herbivore assemblage, resulting in differential local impacts on the tree and grass layers. Apart from local effects, the scale of nutrient patchiness may also regulate the use and quality of forage resources at larger (e.g., landscape) scales. In the large field fertilizer experiment, calculations suggest that the total herbaceous off-take by grazers peaked where the same fertilizer amount (15 kg N plot-1) was spread over the whole plot surface area rather than concentrated in 10 x 10 m patches. Thus, how nutrients are distributed in an area controls secondary productivity and where herbivores concentrate their impact. The scale related patterns observed in the large fertilizer experiment may have been reinforced by plastic responses (e.g., fine root proliferation in nutrient-rich soil) to heterogeneous nutrient supplies of trees in the coarse scale treatments and grasses in the fine scale treatments, respectively. To test this, the same nutrient amount was supplied at two different scales of patchiness to focal trees with their associated grasses. Two years after fertilization, large-scale fertilized trees showed increased shoot growth and increased leaf N concentrations compared to small-scale fertilized trees receiving the same N amount. Conversely, trees in a small-scale configuration fertilized with P showed negative responses compared to large-scale counterparts. These results suggest that differences in the local scale of nutrient patchiness also influence how nutrients are partitioned between co-existing trees and grasses. Herbivores not only respond to nutrient heterogeneity, but may also create spatial heterogeneity in nutrient availability, which, in turn, may influence the vegetation structure of savannas. We studied the current soil nutrient status, tree and grass biomass patterns and large herbivore use of nine former livestock holding pen areas (kraals) in a semi-arid, nutrient poor savanna. These were contrasted with nearby control sites located in the surrounding landscape. The kraals, formerly enriched by livestock dung and urine, were abandoned around 1970 and since then wildlife replaced livestock in these parts. We found that around 40 years later, kraal soil had elevated concentrations of inorganic N, extractable P, K, Ca and Mg compared to control sites, which resulted in high quality forage in kraal sites. Trees also failed to invade these sites, thus kraals remained as structurally open patches in the otherwise dense savanna. Evidently, wild large herbivores maintain the high nutrient status of kraal sites, probably by importing nutrients into these sites and by accelerating local nutrient cycling. In turn, the increased local nutrient availability prevents tree seedlings from establishing under these fertile conditions. Finally, this study provided evidence that changes in the availability of nutrients influenced the success of woody seedling establishment, which may feed back to changes in the relative proportions of trees and grasses in dry savannas. Furthermore, this study supports the new idea that the scale of resource heterogeneity influences how resources are partitioned between co-existing trees and grasses, which, in turn, modulated browser vs. grazer use and impact patterns on the vegetation. In conclusion, this study provides new information on nutrient-plant-herbivore interactions in a dry savanna with potentially important implications for the management of dry savannas in general.