Spatial dynamics of biological soil crusts: bush canopies, litter and burial in Kalahari rangelands

Berkeley, A., Thomas, A.D. and Dougill, A.J.

Proposed Journal – Journal of Arid Environments (??)

Too long for African J. of Ecology => target for an associated Dougill and Thomas paper with our data from Mabuha / Tshane

Abstract

Introduction

The Kalahari is the vast, semi-arid savanna landscape that comprises much of Botswana (Thomas et al., 2000). Livestock farming in the Kalahari is typified by the use of boreholes that provide groundwater reserves to cattle. Intensive grazing pressure around these waterpoints, has led to widespread concerns over rangeland degradation (e.g. Moleele & Perkins, 1998; Dougill et al., 1999; Moleele et al., 2002), notably over the increased dominance of woody bush species (Moleele, 1998). This process, referred to as bush encroachment, has been linked to spatial heterogeneity of soil resources, capable of facilitating a reorganization of the community into so-called ‘islands of fertility’ (Titus et al., 2002) that can contribute to the competitive advantage of encroaching bush species (Schlesinger et al., 1990; Dougill & Thomas, 2004). This paper aims to improve understanding of the mechanisms controlling relations between the encroaching bush cover and sub canopy soil biochemical characteristics that will control future ecological changes in Kalahari rangelands.

One component of the Kalahari system that has been largely overlooked in past research are biological soil crusts, comprising cyanobacteria, green algae, lichens, mosses, microfungi and other bacteria (USGS, 2001). Biological soil crusts are present in all arid and semi-arid regions (Belnap & Lange, 2003). The ecological roles of these crusts include; increasing soil surface stability by binding erodible soil particles into less vulnerable soil aggregates, thus decreasing erosion by wind and water (Eldridge & Leys, 2003); fixing atmospheric nitrogen (Aranibar et al., 2003), which is vital as nitrogen, after water, is the resource most limiting to primary productivity in drylands (Belnap, 2002); also, some communities of biological crust have a considerable photosynthetic element and so sequester soil organic carbon (Zaady et al., 2000).

Although crusts are usually associated with finer grain soils, Dougill & Thomas (2004) documented a biological soil crust cover of between 19 - 40 % at a range of sites on Kalahari sand soils. Fundamental to understanding the ecological significance of biological soil crusts in the Kalahari is a comprehension of their spatial distribution. Several factors are recognised as influencing crust distribution and development, especially substrate character, vegetation type and cover, and disturbance levels. Thomas et al. (2002) have documented the differences in biological crust cover for several substrate types in the Southern Kalahari. However, the relationship between crust cover, vegetation cover and disturbance regime remains uncertain.

It has been demonstrated that plants growing in crusted soils may exhibit enhanced nutrient levels, as compared to those growing on non-crusted surfaces (Belnap, 2002). However, it is also reported that vegetation cover and biotic crust cover are negatively related due to the effects of competition for light and moisture (Malam Issa et al., 1999), and nutrients (Harper & Belnap, 2001). It is generally accepted that trampling, as a result of continuous grazing, destroys biologically crusted surfaces (e.g. Eldridge, 1998). It follows that, in areas of intense grazing such as around Kalahari boreholes, the spatial distribution of biological soil crusts will be limited. However the hypothesis that crust cover will increase with distance from borehole (i.e. with decreasing disturbance) is yet to be examined and may be complicated by the increase in bush cover away from waterpoints (Ward et al., 2000).

Zaady & Bouskila (2002) describe disturbances as the key factors in determining crust development in areas where physical conditions are relatively constant. Given the spatial homogeneity of the Kalahari, in terms of altitude, relief and surface water (Thomas & Shaw, 1993), it is reasonable to impart a significant role to grazing disturbances in determining the distribution of biological soil crusts. In this context bush canopies may represent quasi-discrete environments, in which the response of crusts to local disturbance regimes is altered. This phenomenon is yet to be tested with direct reference to disturbance intensity (i.e., with disturbance as the independent variable), but could be vital in controlling response of the Kalahari ecosystem to grazing related disturbance and to the relative abundance of grasses and shrubs. That biological soil crusts may develop differentially within these sub-canopy habitats would have important implications in terms of the spatial heterogeneity of resources, ecosystem resilience and long-term ecological stability of rangelands.

As demonstrated above, it is probable that the roles of vegetation and disturbance on biological crust distribution are not mutually independent of one another. The aim of this study is to describe the distribution of biological soil crusts at grazed Kalahari study sites in terms of the overlapping domains of vegetation and disturbance. In order to address this aim the following objectives were chosen: (1) To deductively test models that suggest that there are species-specific, sub-canopy protection impacts on the form and characteristics of biological soil crusts; (2) To concurrently make an inductive survey of the association of both litter and sediment burial, with the spatial distribution of biological soil crusts, that are potential explanatory variables which have received little attention within the literature.

Research Design and Site Selection

This study aims to ascertain knowledge on the existence of relationships, principally between bush species cover, disturbance and biological crust cover. In order to test the model proposed by Thomas et al. (2002) regarding the protective capacity of encroaching bush species, two explanatory variables – bush species (as well as bush interspace as a control; these variables will be described as sub-habitats hereinafter) and disturbance – were sought. Analysis of crust distribution in the bush interspaces was necessary to make sound inferences on the additional role of bush canopies on crust development. In order to clarify that any differences in crust cover between these habitats can be attributed to disturbance, it was necessary to record them at differing levels of disturbance. Consequently, several sites were chosen for data collection, at which crust distribution was recorded within the discrete habitats of bush sub-canopy and interspace.

Research was undertaken during July 2003 on communal grazing lands adjacent to Berrybush Farm, near Tshabong, Southern Kgalagadi District, Botswana. Four sites, at different settings around a borehole, were selected for data collection. Given that disturbance characteristics are likely to be very variable, disturbance was quantified at each site using a disturbance index rather than the proxy of distance from borehole, as used in other studies (e.g. Moleele & Perkins, 1998). The closest and furthest sites, with respect to the borehole, correspond to the ‘sacrificial zone’ (Site 1) and ‘un-encroached zone’ (Site 4) of the piosphere model described by Moleele et al. (2002), with the intermediate sites representing the ‘bush encroached’ (Site 2) and ‘mixed’ (Site 3) zones respectively (Figure 1).

In addition to site disturbance, and site sub-habitat as independent variables, litter was used as a further explanatory variable of crust development. This was achieved by concurrently recording in situ litter cover within the crust cover survey. Although it is somewhat unexplored in the literature, and thus essentially unknown, it is possible to attach a priori hypotheses to the nature of crust response to litter. Litter may smother biotic crust and prevent photosynthesis, or, alternatively, may only shade crust and provide a moister habitat more conducive to crust development. Either way, the null hypothesis of no correlation makes this variable deductively testable. In addition, there are no well-established theories on the occurrence of buried crust, and only anecdotal references to the process appear in the literature (Belnap, 2002; Belnap & Gillette, 1998; Harper & Belnap, 2001).

Given that biological crust cover has been shown to reduce sediment entrainment (Belnap & Gillette, 1998; Eldridge & Leys, 2003), it is reasonable to assume that the magnitude of sediment redistribution at a given site is inversely proportional to the amount of surface crusted (i.e. in proportion to the ‘unconsolidated cover’). It is not possible to place an absolute value on sediment redistribution based on the amount of substrate which is unconsolidated, since the degree of entrainment and transport may be site specific, based on factors such as sediment grain size, local wind regime and vegetation. However, the actual extent of redistribution, will be proportional to the area of ground that is unconsolidated. In simple terms: a site with, say, only 5 % biotic crust cover (and therefore 95 % unconsolidated) will have more mobile sediment than a site will 60 % of the substrate crusted (40 % unconsolidated). Furthermore, it is reasonable to suggest that the probability that mobilised sediment will settle upon an area of biotic crust is equal to the area of biotic crust covered. That is, sediment blown across a site with 90% biotic crust cover has a 90% chance of being deposited upon, and thus burying, biotic crust. That the occurrence of crust burial is proportional to both crust cover and the amount of ground which is unconsolidated, can be written in mathematical terms as:

Cburied = kC(100 – C).

where Cburied is the amount of crust buried, C is the percentage of ground crusted (the sum 100 – c representing the percentage area uncrusted, or unconsolidated), and k is the constant of proportionality which, in this case, may describe the combined influences of climate, grain size, and vegetation.

This model predicts maximum values for crust burial at those sites where crust cover and unconsolidated substrate share a mutual maximum (i.e. ~ 50% each), and minimum values of buried crust where the crust cover is either too high (too little unconsolidated substrate for reworking), or too low (probability of burial too low). So it seems, theoretically at least, that the process of crust burial is a trade off between sufficient crust cover to be buried and sufficient unconsolidated substrate to supply the material for burial. Tthe aim here is to provide a basic, inductive description of buried crust distribution. In this respect the incidence of buried crust was added to that of regular biotic crust as one of the dependant variables. However, it is hoped that the proposed model may provide a starting point from which to interpret the results.

Data Collection

Quantification of level of disturbance

At each site, disturbance levels were quantified using cattle track and dung frequency (as per Dougill and Thomas, 2004). At each site, a 50m x 50m grid was established. The grid was crossed at 10m intervals in two, perpendicular, directions. Cattle tracks and dung were counted along each of these gridlines, cattle tracks being defined as well established ‘routes’, and dung laying only within 0.5m either side of each gridline counted. The 0.5m value is arbitrary and for the sake of consistency only. Values of dung indicate ‘sitings’ as opposed to total fragments.

Assessment of biological crust cover in interspaces

Crust cover data were estimated using a 0.5 m x 0.5 m quadrat at intervals of 10 m within a 50 m x 50 m grid. Percentage cover was estimated for each successionary stage of biological soil crust (according to the morphological classification system of Dougill and Thomas, 2004), buried crust, unconsolidated soil, litter and grass within five 0.5 m x 0.5 m quadrats.

Assessment of crust cover beneath bush canopies

The two most common bush encroaching species (Reed & Dougill, 2002) at the study area were selected for sampling, the thorny Acacia mellifera and the non-thorny Grewia flava. The sampling regime was simple – every bush within the aforementioned 50m x 50m quadrat was studied. The canopy dimensions were measured taking the longest diameter on each bush, and then the perpendicular diameter. To measure crust cover, several 0.5m x 0.5m quadrat estimates were taken adjacent to one another along a line extending from the bowl to the canopy edge in two directions – north and south – so as to account for any orientation controlled differences in crust cover. Within each quadrat, crust cover (and morphological type as per classification of Dougill and Thomas, 2004) was estimated, as well as buried crust, unconsolidated substrate and litter.

Results

Bush canopies and biological crust cover

Table 1 summarises the main results obtained across all sites and sub-habitats. In order to test the hypothesis that Acacia mellifera sub-canopies exhibit enhanced crust cover, analyses were required in two contexts; between sites and between sub-habitats (Figure 2). One-way ANOVA showed that there is a statistically significant difference between sites for the crust cover in interspace sub-habitat (F3, 140 = 42.683, p < 0.01), rejecting the null hypothesis of no disturbance-mediated impact on crust development across this zone. A Bonferroni adjustment demonstrated significant differences between Sites 2 and 3 (Site 2, the least disturbed site, having a crust cover significantly greater than Site 3; p < 0.01), and Site 3 having significantly greater crust cover than both Sites 1 and 4 (p< 0.01), between which there was no significant difference. Similarly, crust cover beneath the canopy of Grewia flava was differed significantly between sites (F3, 252 = 27.837, p < 0.01). Within this sub-habitat, values of crust cover at Sites 2 and 3 were statistically taken from the same population, as were Sites 1 and 4; the former pair nevertheless exhibiting significantly greater crust cover than the latter (p < 0.01). However, beneath Acacia mellifera, there was no statistically significant difference in crust cover between sites (F3, 504 = 1.862, p = 0.135; see Figure 2, top). This infers that Acacia mellifera equalizes the effects of local disturbance by protecting the sub-canopy soil from disturbance, whereas the two other sub-habitats show statistically significant variations across the disturbance gradient.

A further vindication of this appears when analysing between-sub-habitats at each respective site. At Site 1 (Dung count = 7.2, the site most intensely disturbed), interspace and Grewia flava crust cover were seen to be statistically indistinguishable from each other, although ANOVA detected a significant difference between the three sub-habitats (F2, 266 = 33.045, p < 0.01). The difference occurred with Acacia mellifera showing significantly higher crust cover than the other sub-habitats (p < 0.01; see Figure 2, bottom). At Site 2 (Dung count = 2.1, the site least intensely disturbed), ANOVA revealed no difference in sub-habitat crust cover (F2, 225 = 0.449, p = 0.639). At Site 3 significant sub-habitat-differences for Site 3 were found (F2, 204 = 3.939, p < 0.05) with Grewia flava sub-canopies enjoying a statistically significant higher share of crust cover, anomalous to our model. At Site 4, the sites differ significantly (F2, 201 = 10.364, p < 0.01), with Acacia mellifera displaying significantly greater crust cover than Grewia flava (p < 0.05) and the interspaces (p < 0.01).