Arid landscape dynamics along a precipitation gradient: addressing vegetation - landscape structure - resource interactions at different time scales

This research is entitled ‘Arid landscape dynamics along a precipitation gradient: addressing vegetation – landscape structure – resource interactions at different time scales’ with as subtitle ‘A case study for the Northern Negev Desert of Israel’. Landscape dynamics describes the interactions and feedbacks among landscape structure, resource flows and organisms. This study focuses on the Northern Negev Desert of Israel, a semi-arid to arid rock desert with local loess and sand cover. Climate and humans are important driving factors of landscape dynamics here. Semi-arid and arid regions worldwide, are vulnerable to land degradation and desertification. A profound knowledge of the processes in these regions can help to avert land degradation and desertification. The objective of this thesis is to increase the knowledge on landscape dynamics and its drivers in semi-arid and arid regions by field and model studies in the Northern Negev Desert. This study can contribute to a sustainable future for the inhabitants of these areas. • Chapter 1 is the introduction of this thesis and discusses among others the four studied catchments along a precipitation gradient: Lehavim receives at average 280 mm precipitation per year, Sayeret Shaked 200 mm yr-1, Halluqim 93 mm yr-1 and Avdat 87 mm yr-1. Of the surface of Lehavim 53% is covered by vegetation and 15% by bedrock outcrops. The catchment is intensively grazed by livestock. Sayeret Shaked is covered by a thick layer of homogeneous loess. Vegetation cover is dense (62%). The catchment is taken out of grazing since 1987. In Halluqim only 20% of the surface is covered by vegetation. The catchment is very rocky, as bedrock crops out at 42% of the surface. Avdat, located close by, is much less rocky (7%). Here 22% of the surface is covered by vegetation. Both catchments are extensively grazed. The thesis can be separated in three parts. In the first part the relationships between landscape structure and vegetation in the four catchments is studied by statistical analyses. This part gives insight in the landscape dynamics along a precipitation gradient and provides a system framework for the remainder of the thesis. The second part focuses on simulating water and sediment dynamics in the catchments using the landscape evolution model LAPSUS. The model is adapted to a semi-arid and arid climate, and vegetation cover is incorporated. The interactions between resource flows and vegetation is studied by model simulations. In the third part the system knowledge and modelling framework are applied at a longer time scale. Firstly the history of a valley fill is reconstructed by field observations, after which this valley fill is simulated with LAPSUS. Additionally the effect of land use on the valley fill development is tested by model simulations. Part 1: System framework • In chapter 2 the controls on functional surface cover types are studied in the four catchments along the precipitation gradient. First, four functional surface cover types are selected, based on their unique functionality in terms of water use and redistribution: shrubs, Asphodelus ramosus, other herbaceous plants and surface crusts (biological and physical). Percentage of surface cover of these functional surface cover types is estimated, and of bedrock outcrops and loose surface stones. Additionally, data is collected on soil depth, relative elevation, insolation, slope, profile curvature and plan curvature. Relations between functional surface cover types and landscape structure variables are analyzed with descriptive statistics, factor analyses and linear regressions. The landscape structure variables bedrock outcrop, relative elevation, soil depth and surface stones explain most of the cover variance in the catchments. In catchments with many bedrock outcrops, the occurrence of functional surface cover types is best explained by the landscape structure variables. In catchments with homogeneous soils reaching beyond the root zone, biological interactions between functional surface cover types are more important. Along the precipitation gradient the explanatory power of the biological variables decreases with decreasing precipitation, while the explanatory power of landscape structure variables appears unrelated. Only in homogeneous semi-arid catchments can regular vegetation patterns develop, in arid and heterogeneous catchments irregular vegetation patterns dominate. Part 2: Model framework • In chapter 3 the process of water redistribution at catchment scale is studied with the landscape evolution and erosion model LAPSUS. LAPSUS, formerly applied in Mediterranean regions, is modified to deal with the arid climate of the Northern Negev Desert of Israel. Daily model runs are used instead of yearly model runs, and the infiltration module is adapted to better represent the spatial diversity in water availability in an arid catchment. The model is calibrated for Halluqim and Avdat. First, a sensitivity analysis of the modified LAPSUS was done. Especially pore volume of the soil appears to have a strong influence on the modelling results. Second, the capability of LAPSUS to deal with varying surface characteristics was assessed by comparing the simulated water redistribution patterns in the two catchments with field data. Simulation results demonstrate that the catchments respond very different to precipitation. Water redistribution is larger in the dominantly bedrock-covered Halluqim compared to the dominantly sedimentcovered catchment of Avdat. Consequently, Halluqim has more positions with water accumulation than Avdat, and can sustain a larger vegetation cover including Mediterranean species. Finally the modelled infiltration patterns are spatially compared with vegetation cover in the catchments. The results indicate that there is a broad agreement between infiltration and vegetation patterns, but locally there is a strong mismatch indicating that part of the involved processes are still missing in the model. • In chapter 4 the interactions between resource flows and vegetation is studied and simulated in the loess-covered catchment of Sayeret Shaked. In semi-arid areas vegetation is scarce and occurs often as individual shrubs on raised mounds. The formation process of these mounds is still debated. In this chapter the hypothesis that shrub mounds are formed in part of the Northern Negev Desert by erosion and sedimentation is tested. Height and diameter of shrub canopy and shrub mounds are measured and micro-morphological techniques are used to reconstruct the formation process of the shrub mounds. The results suggests that shrub mounds are formed by accumulation of atmospheric dust and sedimentation of eroded material in the vicinity of the shrub, as well as by erosion of the surrounding crust. Model simulations are done for single events and longer time scale (100 years). In the simulations, mound formation appears most prominent at low shrub density and large shrub canopy diameter. Positive and negative feedbacks between shrubs and resource redistribution results in a meta-stable landscape. Long-term model simulations of the current climate indicates that initially formation rate of shrub mounds is high, but stabilized at lower rates. In dryer and wetter climates mound formation is unlikely to happen, as respectively too little or too many resources are redistributed, causing a stable or highly erosive landscape. Mound simulation with LAPSUS is successful and simulated shrub mounds resemble the actual shrub mounds in Sayeret Shaked. Consequently the model may prove to be valuable for the modelling of ecohydrological landscape processes in semi-arid areas. Part 3: Long-term application • In chapter 5 the interactions between climate change, human occupation and semi-arid landscape dynamics are studied to increase the insight in the effect of climate change and human land use. A Late Quaternary valley fill in the catchment of Sayeret Shaked is studied. The aggradation and incision history is reconstructed based on a transect study. The reconstructed valley fill is put in a temporal framework by correlation with local climate records and optically simulated luminescence and potsherd dates. Two Late Pleistocene and four Holocene aggradation and incision cycles are recognized, of which three in the last 2000 years. Contradictory to the expected positive relation between amplitude of climate fluctuations and cycles of aggradation and incision, the Late Holocene cycles are stronger than those in the Late Pleistocene and Early to Middle Holocene. The most significant cycle coincides with the rise and fall of the Byzantine Empire and appears related to the higher pressure on the landscape due to human occupation during that time. Human activity appears to have a strongly amplifying effect on aggradation and incision phases, which are initially triggered by climate fluctuations. This amplifying effect occurs only when human occupation crosses a threshold and triggers destabilization of the landscape. It causes collapse of the ecosystem and increases sediment redistribution. • Chapter 6 aims to quantify the effect of humans on semi-arid catchments, by reconstructing the infill history of Sayeret Shaked using LAPSUS. First, the infill history of Sayeret Shaked between about 800 BC and 800 AD is simulated. Second, three land use scenarios are tested to quantify the effect of extensive grazing, intensive grazing and intensive grazing combined with rainfed agriculture. Especially intensive grazing combined with rainfed agriculture leads to strong landscape dynamics. Extensive grazing causes almost no landscape dynamics, resulting in an almost stable landscape. The results seem to indicate that this catchment is formed by coevolution of human and natural induced processes. Rainfed agriculture leads to valley aggradation by tillage translocation, whereas intensive livestock grazing causes gully incision by increased slope runoff. Humans appear to be the main driven factor of landscape dynamics in this semi-arid catchment, much more than climate fluctuations. Only a short time period of strong human land use can irreversibly alter the development trajectory of a catchment. It is thus of high importance to manage the land sustainable, both in the present and future, to avoid further degradation of drylands. • In chapter 7 the results of the different chapters are combined and the most important conclusions discussed. The four catchments display very different landscape dynamics, caused by a high variation in climate, land use and landscape structure. In Lehavim and Halluqim the landscape dynamics is strongly influenced by the landscape structure, because bedrock outcrops regulate positions for vegetation grow and stimulate water redistribution. In Sayeret Shaked water redistribution depends mainly on biological surface cover. In Sayeret Shaked interactions between shrub and crust patches can, under a more intensive grazing regime, lead to regular vegetation patterns. When grazing pressure is released the herbaceous plant coverage recovers, as is happening today. Avdat is a divers catchment, with steep rock outcrop, a flat plateau and a loess covered wide gully. Though the whole catchment is characterized by a high aridity, each zone experiences different landscape dynamics. At a larger spatial scale, in the whole Northern Negev Desert, the most relevant interactions and feedbacks between landscape structure, resource flows and organisms are related to water availability and redistribution as well. Since the Late Holocene, the main driving factor of landscape dynamics is human land use, especially tillage and intensive livestock grazing. Climate fluctuations seem to have much less influence on the region. The influence of humans, even confined in a small period in time, strongly affects landscape development in the whole Northern Negev Desert, causing co-evolution and formation of cultural landscapes.