Wednesday, July 23, 2014

Soil Erosion, Climate Change and Global Food Security: Challenges and Strategies. Part 3.


This is the third part of a much longer article published in the journal Science Progress, and which may be found here: http://stl.publisher.ingentaconnect.com/content/stl/sciprg/2014/00000097/00000002/art00001


5. Climate change
.

As noted later, the effects of climate change must be considered in the context of soil erosion and land degradation: in particular a more vigorous hydrological cycle is anticipated as a result of rising global temperatures, with much harder rainfall in some regions19. Hence, in the absence of mitigating measures, the future rate of soil erosion can be expected to increase on the global scale. There are a number of prevailing factors at play, the most significant being the erosive force of rainfall; however, we must also consider the following: (a) the changing moisture regime might alter patterns of biomass growth, and affect the canopy layer; (b) the latter may amend both the plant residue decomposition rates (since the underlying processes are driven by temperature and by moisture and are connected with the activity of soil microbes) in addition to the biomass production rates; (c) in consequence of varying rainfall and evapotranspiration rates - which changes infiltration and runoff ratios - the soil moisture content might be affected; (d) a fall in the SOM content may weaken the structure of some soils, rendering them more susceptible to erosion, and the amount of runoff could be increased as a result of surface sealing and crusting; (e) non-erosive winter snowfall might turn into erosive rainfall as the winter temperatures increase; (f) when permafrosts melt, a previously non-erodible soil state can be converted to a highly erodible form; and (g) land use changes, e.g. to grow more cereals, as the global climate and local weather patterns change, may lead to further erosion.

It has been estimated that a ca 1.7% change in soil erosion is likely for each 1% change in total precipitation resulting from climate change20. It is common practice to deal with a loss in soil fertility from erosion by applying greater quantities of artificial fertilizers, which actually compounds the problem [since the soil food web (Figure 4) dies back further2] and incurs yet more water and soil pollution, as opposed to simply giving the land sufficient time to regenerate naturally. The relationship between the mycorrhizal fungi in the soil and, for example, phosphate fertilizer can be thought of as pseudo-addictive, since the fungi act symbiotically with the roots of plants, delivering to them phosphorus (and other nutrients) drawn from the soil, in return for carbohydrate bestowed to the fungi from the plant (formed by photosynthesis). The over-application of artificial phosphorus discourages the growth of the fungi, with the result that the plants become increasingly dependent on artificial phosphorus inputs.

6. Monitoring, measuring, and modelling erosion.

To model erosion accurately is difficult, because of the complexity of the detailed processes of erosion, which involve aspects of climatology, hydrology, geology, chemistry, and physics, etc. Due to the non-linear nature of erosion models, they tend to be difficult to use numerically, and it is accordingly difficult or impossible to make predictions about large scale events on the basis of results taken from plots on much smaller areas. Erosion models are either process-based or empirically based. The former models are physically based and provide a mathematical description of the processes of detachment, transport, and deposition: by solving the equations that describe them, estimates are obtained of soil loss and sediment yields that occur from a given land surface area. The science of erosion is not sufficiently developed that some input of empirical data can be avoided. The fundamental difference between process-based and other types of erosion models is that the sediment continuity equation is used, as is discussed later. Empirical models relate management and environmental factors directly to soil loss and/or sedimentary yields on the basis of statistical methods21. A detailed review22 of process-based and empirical erosion models has been published, including a discussion of conceptual models, which are a kind of intermediary stage between the process-based and entirely empirical models. The most usual model that is employed to assess the degree of erosion and to permit an outlook toward conservation strategies is the Universal Soil Loss Equation (USLE), which remains under improvement and development.

6.1 The USLE.

The USLE was developed using a comprehensive range of data taken from erosion plot and rainfall simulator experiments: in all, over 10,000 plot-years worth of actual data, taken from 50 different locations in 24 U.S. states was used to calibrate the input parameters23. The USLE contains six factors, according to which the long-term average annual soil loss (A) is estimated, which are: the rainfall erosivity factor (R), the soil erodibility factor (K), the topographic factors (L and S) representing length and slope, and the cropping management factors (C and P). The equation has the simple linear structure:

A = RKLSCP

The unit plot concept is important in the context of the USLE, and is defined as the standard plot condition to determine the soil's erodibility, i.e. when the LS factor = 1 (slope = 9% and length = 72.6 feet) where the plot is fallow and tillage is up and down slope and no conservation practices are applied (CP = 1). Under these conditions:

K = A/R

Wischmeier et al.24 have devised a more straightforward means for the estimation of K, the soil erodibility factor, which involves the particle size of the soil, its SOM content, soil structure and profile permeability. If sufficient information is available, K can be approximated from a nomograph. By knowing the length and gradient of the slope, the LS factors can be determined from a slope effect chart. The cropping management factor (C) and conservation practices factor (P) are determined empirically from plot data, and are described in soil loss ratios [i.e. (C or P with) /(C or P without)]. Erosion is measured and analysed using e.g. the micro-erosion meter (MEM) and the traversing micro-erosion meter (TMEM). The MEM has been used successfully to measure bedrock erosion in a range of ecosystems across the globe, and is able to determine both terrestrial and oceanic erosion. There is also the Revised Universal Soil Loss Equation (RUSLE), which is an extension of the USLE, and other related variants. A highly informative and practical description of the use of the USLE can be found at: http://www.asu.edu/clas/shesc/projects/medland/documents/soilerosion.pdf

7. Validity of universal soil loss estimates.

Pimental et al. have asserted25 that globally, soil erosion rates are lowest in the U.S. and Europe, “averaging about 17 metric tons ha-1 year-1”. Boardman26 has investigated the origin of this figure and concluded that it is actually derived from an uncritical extrapolation of data taken from just 12 experimental plots in three small areas of central Belgium, reported by Bollinne in his Ph.D thesis at the University of Liege. [A range of soil loss of 10─25 t ha-1 year-1 (which “average” to ca 17 t ha-1 year-1) has been claimed from Bollinne’s work25]. Hence it is of some interest to know how much erosion is actually occurring across Europe, according to a more extensive compilation of measurements, taken over a larger and more representative area of the continent. Indeed, the available data indicate that the process is quite variable in its extent26, both in space and time, and ranges from (assuming a soil density of 1.4 g cm-3, since some data are quoted in units of m3 ha-1 year-1) < 1 t ha-1 year-1, over a 90 km2 area in southern Sweden, to ca 16 t ha-1 year-1 in northern France, but taken from just 33 small catchments. Arden-Clark and Evans quote27 that erosion rates in the U.K. are 1─20 t ha-1 year-1, while noting that those in the higher part of the range are rare and localised events, and typical rates of erosion are 1─2 t ha-1 year-1. Measurements made over a 709 km2 total area of localities in England and Wales produce a mean28 of 3.2 t ha-1 year-1.

Ryszowski29 reckoned the soil erosion rate in Poland to be 0.52 t ha-1 year-1. Hence, overall, it might appear that the average value of 17 t ha-1 year-1 quoted25 by Pimental et al. represents a considerable overestimate of the rate of soil erosion in Europe. Crosson too has taken issue with the figure30, this time as applied to soil erosion in the U.S. According to the USDA in 1989, an average of 17 t ha-1 year-1 is moved from U.S. croplands as a result of the combined effect of wind and water erosion; however, Crosson cites more recent data showing that by 1992, the rate had fallen to 13 t ha-1 year-1, and refers to a paper by Lal and Stewart31 in which it is stated that, annually, some 36 billion tons (metric tons = tonnes is meant) of soil are eroded worldwide: 10 billion tons from natural phenomena and 26 billion tons as a result of human activities. Crosson notes that Lal and Stewart cite a paper by Brown32 as a source for the 26 billion ton estimate, but this is based on erosion measurements for the United States, which it would appear are higher than those for Europe. At any rate the 36 billion ton figure is considerably less than the 75 billion metric ton value assessed elsewhere to be removed by wind and water erosion, and mostly from agricultural land.

In their response to Crosson, Pimental et al. assert33 that since the United States has about 11% of the world’s arable land (and approximately the same amount of pasture land), and an estimated total soil loss of 4.5 billion tons per year, assuming that the rest of the world suffers similar rates of soil loss, a total global soil loss of 40 billion tons per year is indicated. They further stress that the rates of soil erosion in Asia, Africa and South America are about double those in the U.S., and hence the 75 billion ton annual global loss of soil appears reasonable. They defend too, their contention that some 80% of the world’s agricultural land has been degraded, though Crosson criticised30 this figure as a more than three-fold overestimate, based on the GLASOD34 [Global Assessment of Soil Degradation] study by Oldman et al. which reports that about 1.03 billion ha of agricultural land has suffered moderate-to-strong erosion as a result of wind and water, or less than 25% of the roughly 4.5 billion ha global land-base, under crops, pasture and range. In a later paper, Trimble and Crosson point out the considerable disparity in quoted values for the annual soil losses through erosion from croplands (442 million acres = 179 million ha) across the United States, which vary from 2 billion to 6.8 billion tons, which suggests an average lost soil depth of ca 1─3 mm35. Some are of the opinion that the recent rates of soil erosion are as high as those in the 1930s, though others disagree35.

On the basis of on-farm productivity effects, it was concluded that should the rates of erosion that prevailed in 1982 continue for the next 100 years, the crop yields (output per hectare) would be reduced only by around 2─4%, and hence increased federal funding to reduce the erosion is not justified. Most of the estimations made of soil erosion are based on models, particularly the Universal Soil Loss Equation (USLE) although this itself has been developed and “calibrated” using more than 10,000 plot-years worth of actual data, taken from 50 different locations in 24 U.S. states. Its successor, the Revised Universal Soil Loss Equation (RUSLE), has been improved using further experimental data and in two studies of measured soil loss rates, taken from over 1,700 plot-years worth of data on 205 research plots in 20 locations across the U.S., it was shown that the USLE and RUSLE actually predict rates of soil loss reasonably well, even for post-1960 conditions23. A limitation of USLE and RUSLE is that they predict the amount of soil moved on a field, which is not necessarily the same as the amount of soil physically removed from the field. To estimate the latter, the sediment delivery ratio (SDR) is determined. The SDR is given by the amount of sediment delivered from an area divided by the gross erosion of that same area. Expressed as a percentage, SDR reflects how efficient the watershed is in moving soil particles from an area where erosion is occurring, to the location at which the sediment yield is measured. The model assumes that a relatively small amount of eroded soil leaves a field or stream basin, and some sediment is presumed to be deposited by wind on the field, or along streams as alluvium.

However, it is often assumed that the USLE measures soil actually removed from the land, and the variance in SDR according to particular conditions is not taken account of. As an example, in the 1970s, the sediment delivery to streams from a 3 km2 area in Wisconsin amounted to just 8% of the USLE prediction35, with the remainder thought to be stored as colluvium (loose, unconsolidated sediments that have been deposited at the bottom of hillslopes by either rainwash, sheetwash, slow continuous downslope creep, or a combination of these processes). In contrast, in the 1930s, the sediment delivered was 123% of the USLE value, due to the then frequent effect of gullying downslope from agricultural fields. Also in the 1930s, it was common that the skies over the eastern United States were filled with enormous clouds from the Dust Bowls, which moved out over the Atlantic Ocean, due to severe wind erosion. It is thought that, more recently, the erosion process amounts more to local redistribution than wholesale loss. To predict the latter phenomenon, the Wind Erosion Equation (WEE) has been used, and as with the USLE there is a mass discontinuity problem: namely that even though soil may be eroded from one area, most of the particles are simply moved onto other fields, and so the net soil loss may be overestimated. It has been claimed that large areas of the U.S. have annual erosion rates of >25 t ha-1, and yet the sediment yields were often in the range 0.5─2.0 tonnes ha-1, including a contribution from significant erosion of the banks and stream channels35. Thus, the total sediment delivered to streams has been reckoned at 2.7─4.0 billion tons, but the actual amount measured at this destination is nearer 0.5 billion tons, with the inference that a large quantity of the sediment must be stored in the watershed. Alternatively, as Trimble and Crosson have averred35, the rate of soil removal may be far less than is apparent using the USLE, calling for more field studies and monitoring of sediment mass, a view endorsed by Nearing23.

Cerdan et al. have used erosion plot data to compile a comprehensive database with which to investigate the rates and spatial variations in sheet and rill erosion across Europe36. It was demonstrated that land use is overwhelming in its influence on erosion rates, which may be greater by an order of magnitude on conventionally tilled arable land, as compared with those surfaces that are permanently covered by vegetation. The erosion rates tend to be lower in the Mediterranean, due to the protective effect of the rocks in the stony soils there. However, because these soils are already very thin, any further loss of soil could be highly detrimental. The average rate of sheet and rill erosion was determined to be ca 1.24 t ha-1 year-1 for the entire area covered by the CORINE database (essentially the whole of Europe - both eastern and western – plus Turkey), and 3.6 t ha-1 year-1 for arable land. Evidently, there are “hot spots” for erosion, but these are masked by quoting a figure for an average erosion rate for Europe. A review22 has been published of the USLE and its family of related models, as applied to the determination of event soil loss, and runoff. It is concluded that the predictive power of the method works well in some locations but poorly in others. One problem in the use of the USLE to predict event erosion is that it was originally designed to model long-term average soil loss, and the event rainfall-runoff factor contained in it does not consider runoff explicitly.

When runoff is measured or estimated reasonably accurately, the prediction may be improved, but in incorporating the direct runoff in the rainfall-runoff factor an impact on some of the other factors used in the model is incurred. Parsons et al. have attempted to predict the travel distance of different particle sizes to provide a model for the erosive impact of specific rainfall/runoff events37: this is of particular relevance to the European situation, in terms of the off-site impact of runoff and erosion. The model only considers simple cases of erosion under a spatially uniform rainfall and on slopes of uniform infiltration and gradient, but it is independent of measurement area since it rests upon the notions of entrainment and travel distances of particles, and so on sediment flux. The results resolve the “paradox” of sediment-delivery ratio, and some of the recent discussion over the validity of erosion rates made from USLE erosion plots; potentially, erosion rates can also be reconciled with the known life-spans of continents. It is concluded that some of the accepted estimates of erosion rates are fallacious, and those that are based on measurement of inter-rill erosion (short runoff plot) are likely to be much larger than reality (for entire hillsides). That noted, the significance of travel distance increases the importance of rill and gully erosion, and of the movement of fine particles to which nutrients are pollutants adhere preferentially. It is clear that far more actual experimental data must be garnered - taken on a range of scales, from small plots to hill-slopes and catchments – in order to evaluate the predictions of the model, and indeed to provide a true image of global soil erosion and land degradation.


8. Measures and realities of global soil erosion.

Results from remote sensing measurements show that the total amount of biomass over the earth’s surface has increased3,4. This may be a result of carbon dioxide fertilization, where photosynthesis and hence plant growth is stimulated by elevated levels of CO2 in the atmosphere. Globally, the amount of biomass was measured to increase by 3.8% during the years 1981─2003. However, it was also determined that 24% of the global land area suffered some degree of degradation during the same period. Hence, there are regions of “greening”, while elsewhere “browning” has occurred. According to a recent study made by the Australian Commonwealth Scientific and Industrial Research Organisation (CSIRO) and the Australian National University, in some of the earth’s driest regions (arid regions of Australia, North America, the Middle East and Africa) foliage has increased by 11% during the past three decades, as a consequence of CO2-fertilization38. Due to different definitions and terms, as noted earlier, there is a wide variance in the estimates of both the degree and rate of global land degradation. Most of the policy decisions over land have been made on the basis of two major sources: the extent of global desertification by Dregne and Chou39, and that of global land degradation by the International Soil Reference and Information Centre by Oldeman et al.34, termed Global Assessment of Human-induced Soil Degradation (GLASOD).

The two sets of figures are not strictly comparable, because while Dregne and Chou considered only dry areas they also included the status of vegetation on the rangeland. The terminology also differs between the two studies: Dregne and Chou used the terms38 slight, moderate, severe and very severe to denote the degree of degradation, while Oldeman et al. used the terms34 light, moderate, strong and extreme, and hence the degree of correspondence between the levels of designation in the two studies is uncertain. Oldeman et al. tried to separate natural degradation from that which had been caused by human activities. Eswaran and Reich40 attempted to determine the vulnerability of land to degradation and desertification, but only considered arid, semi-arid, and sub-humid regions, according to the definition of UNEP; their estimates of water erosion also include humid areas. According to GLASOD34, of the different erosion mechanisms, it is water erosion that is the most important, and afflicts some 1,094 million ha (56%) of the total area that is impacted upon by human-induced soil degradation. Globally, wind erosion affects 548 million ha (38%) of the terrain that is degraded. Chemical soil deterioration affects 239 million ha (12%) of the total, and physical soil deterioration occurs over 83 million ha (4%). The most important subtype of displacement of soil material is through the influence of water or wind. An area of 920 million ha is affected by water erosion (365 million ha in Asia, and 205 million ha in Africa), and 454 million ha by wind erosion. The principal chemical deterioration of soils involves the loss of nutrients and this affects 135 million ha worldwide, of which 68 million is in South America. Salinization follows next in order of its impact, and afflicts some 76 million ha globally, of which 53 million ha is in Asia. An area of 22 million ha is affected by pollution, of which 9 million ha is located in Europe. The most significant subtype of physical soil deterioration is compaction, and occurs over an area of 68 million ha, of which 33 million ha is in Europe, and 8 million ha is in Africa.

GLASOD categorises four degrees of soil degradation. “Light” refers to a somewhat reduced productivity of the terrain, but which is manageable in local farming systems, applies to 38% of all degraded soils (749 million ha). “Moderate” requires improvements which are often more than can be achieved by local farmers in developing nations, and accounts for 46% of the Earth’s degraded soils. Thus 910 million ha of the Earth’s surface has a greatly reduced productivity: >340 million ha of these moderately degraded lands are in Asia and >190 million ha in Africa. There are 296 million ha globally of “strongly degraded” soils (124 million ha in Africa, and 108 million ha in Asia), and it is these that it is not possible to reclaim at the farm level and which may therefore be regarded as lost land. These terrains can only be recovered through major engineering work and/or international assistance. Finally, soils that are “extremely degraded” are regarded as irreclaimable and beyond restoration, amounting to a global total of 9 million ha (>5 million ha in Africa).

GLASOD is not without its critics41, and indeed its authors were well aware of, and the first to indicate, its limitations: principally that it was based on the perceptions of experts, rather than being a direct measure of land degradation. More recently3,4, methods of remote sensing have been applied to determine the extent of global land degradation: LADA (Land Degradation Assessment in Drylands). These aim to determine the degree and trends of land degradation in drylands, degradation hotspots and bright spots (both actual and potential), using changes in net primary productivity (NPP) as a proxy measure of land degradation. [Net primary productivity (NPP) is defined as the net flux of carbon from the atmosphere into green plants per unit time. NPP refers to a rate process, i.e. the amount of vegetable matter produced (net primary production) per day, week, or year]. The most heavily degraded regions are identified to be Africa: south of the equator (13% of the global degrading area and 18% of lost global net NPP), South-East Asia (6% of the degrading area and 14% of lost NPP), south China (5% of the degrading area and 5% of lost NPP), north-central Australia and the western slopes of the Great Dividing Range (5% of the degrading area and 4% of lost NPP), the Pampas (3.5% of degrading area and 3% of lost NPP) and swaths of the high-latitude forest belt in Siberia and North America, directly affecting the livelihoods of the 1.5 billion people who live there. The results indicate that 24% of the total global land surface has suffered degradation during the past quarter century, and may be compared with the 15% of the world’s soil (not land) being degraded, according to the GLASOD study.

Much of the degradation identified by GLASOD33 does not overlap with the areas newly highlighted by LADA, demonstrating that land degradation is both cumulative and global. The authors stress that land-use changes which reduce NDVI [remotely sensed Normalised Difference Vegetative Index (Figure 6)], e.g. from forest to cropland of lower biological productivity, or an increase in grazing pressure, may or may not be accompanied by soil erosion, salinity or other symptoms of land degradation that are of concern to soil scientists. They note further that while long-term trends in NDVI derivatives are only broad indicators of land degradation, taken as a proxy, the NDVI/NPP trend is able to yield a benchmark that is globally consistent and to illuminate regions in which biologically significant changes are occurring. Thus attention may be directed to where investigation and action at the ground level is required, i.e. to potential “hot spots” of land degradation and/or erosion.

[Fig 6]

Montgomery42 has made a global compilation of studies which confirms the long held contention that the erosion rates from conventionally ploughed soils are 1─2 orders of magnitude greater than the background rates of soil production, of erosion under native vegetation, and long-term geological erosion. He concludes that on a global basis, hill-slope soil production and erosion evolve to balance geologic and climate forcing, whereas agriculture based on conventional ploughing increases the rates of erosion to unsustainable levels. At a rate close to 1 mm/year of soil loss (amounting to around 14 t ha-1 year-1), net erosion rates in conventionally ploughed fields can erode a typical hill-slope profile on a timescale of major civilizations, whereas no-till methods of farming cause rates of erosion that are nearer to those of natural creation rates of soil, and hence might set the cornerstone of a system of sustainable agriculture.

No comments: