Interrill and rill erodibility in the northern Andean Highlands

Abstract

There is a lack of quantitative information describing the physical processes causing soil erosion in the Andean Highlands, especially those related to interrill and rill erodibility factors. To assess how susceptible are soils to erosion in this region, field measurements of interrill (Ki) and rill (Kr) erodibility factors were evaluated. These values were compared against two equations used by the Water Erosion Prediction Project (WEPP), and also compared against the Universal Soil Loss Equation (USLE) erodibility factor. Ki observed in situ ranged from 1.9 to 56 × 10⁵ kg s m^− 4 whereas Kr ranged from 0.3 to 14 × 10^− 3 s m^− 1. Sand, clay, silt, very fine sand and organic matter fractions were determined in order to apply WEPP and USLE procedures. Most of the evaluated soils had low erodibility values. However, the estimated USLE K values were in the low range of erodibility values. Stepwise multiple regression analyses were applied to ascertain the influence of the independent soil parameters on the Ki and Kr values. After this, we yield two empirical equations to estimate Ki and Kr under this Andean Highlands conditions. Ki was estimated using as predictors silt and very fine sand, while Kr used as predictors clay, very fine sand and organic matter content. Relationship among Ki, Kr and K are described for the Highland Andean soils.

Introduction

More than 10 million rural inhabitants reside in the Mountainous region of the major Andean countries where moderate and severe soil erosion occurs (Zimmerer, 1993) However, quantitative studies on soil erosion as well as the knowledge on water and erosion processes are scarce in the Andes (Stroosnijder, 1997) compared with other areas in the world, especially those related to soil erodibility (Víctora et al., 2001, Zehetner and Miller, 2006). Published erosion rates are around 48 Mg ha^− 1 year^− 1 in Colombia (Ashby, 1985); 0 to 836 Mg ha^− 1 year^− 1 in Ecuador (Harden, 1988); 10 to 70 Mg ha^− 1 year^− 1 in Peru (Low, 1967) and 114 to 173 Mg ha^− 1 year^− 1 in Bolivia (Zimmerer, 1991). Current studies try to give a better approximation of the estimated erosion rates, because time- and scale-dependent aspects of soil loss and sediment transfer make comprehensive measurements difficult.

Inbar and Llerena (2000) determined sediment yield quantitatively from abandoned terrace areas in Central Andes of Peru. However, no calculation of erodibility values has been done in these plot studies. Sánchez et al. (2002) made a comparative study of soil erosion in the Venezuelan Andes. Soil losses were quantified by using erosion plots in areas covered by four types of vegetation (apple trees, pasture, natural forest, and horticultural crop in rotation). The lowest soil loss rated was associated to the natural forest, with an average value of 0.43 mg ha^− 1 year^− 1 and the highest occurred with horticultural crops in rotation, with an average value of 15 Mg ha^− 1 year^− 1. They calculated the soil erodibility factor (K) of the Universal Soil Loss Equation (USLE—Wischmeier and Smith, 1978) based on the relation between aggregation, textural class and organic matter content of the topsoil. The K factor values were 0.030, 0.045, 0.032 and 0.038 Mg ha^− 1/MJ ha^− 1 mm h^− 1 for natural forest, horticultural crops, pastures and apple plantation treatments, respectively. Zehetner and Miller (2006) studied the erodibility and runoff-infiltration characteristics along an altitudinal Entisols–Inceptisols–Andisols sequence in the Andes of northern Ecuador. Using disturbed soil samples packed into small pans and placed on a 9% slope, simulated rainstorm with varying intensities was applied for a duration of 30 min. During the simulated event, runoff and eroded sediment were collected in 5-min intervals and measured by weight before and after drying. They calculated the interrill erodibility with the original WEPP interrill equation (Flanagan and Nearing, 1995). Ki ranged from 0.5 to 7.9 × 10⁵ kg s m^− 4.

Natural rainfall represents natural conditions at a given place; however, data acquisition is difficult due to the lack of control of the spatial and temporal distribution of rainfall intensity (Moore et al., 1983). A more cost-effective alternative is to use rainfall simulators to apply controlled rainstorms to small plots (Kamphorst, 1987, Esteves et al., 2000). Portable rainfall simulators used on small plots give sufficient flexibility to study a variety of processes (e.g. infiltration, irrigation, interrill erosion and water quality) on different soils and slopes and different land uses (Sharply et al., 1999, de Lima et al., 2002) and can be used to collect data in a relatively short period, providing maximum control over plot conditions and rainfall characteristics (Wilcox et al., 1986). Performing experiments with a rainfall simulator make possible to compare runoff rates and soil detachment by raindrop impact between sites at which the same experimental procedure was used, thus providing a basis from which to understand spatial patterns of vulnerability to soil erosion over a broad area. However, splash detachment rates from very small plots can exceed soil erosion rates determined in large, conventional plots for comparable natural rainstorms because conventional measure require entrained particles to be transported to the lower edge of the plot (Harden, 2001).

The main disadvantages of using rainfall simulators are related to scale. First, it is cheap and simple to use a small simulator which rains onto a test plot of only a few square meters, but simulators to cover field plots are large, expensive and cumbersome, and secondly, measurements of runoff and erosion from simulator tests on small plots cannot be extrapolated to field conditions. They are best restricted to comparisons, such as which of three cropping treatments suffers least erosion under the specific conditions of the simulator test, or the comparison of relative values of erodibility of different soil types (Hudson, 1993).

With the development of USLE (Wischmeier and Smith, 1978) — the identification of the soil erodibility K factor became a central issue in erosion studies (Bryan et al., 1989). USLE continues being applied all over the world and provides a practical alternative to estimate K. Different studies show different results of applying USLE in the tropics (Vanelslande et al., 1984, Mati et al., 2000, Mati and Veihe, 2001, Baumann et al., 2002, Kim et al., 2005, Weill et al., 2006, Millington, 2006). Problems with the use of USLE in this environment appeared to be: (1) rainfall intensities are higher than those occurring in eastern USA, where it was developed; (2) different methods of soil aggregation that are found in tropical soils — particularly bonding by iron, aluminium and organic acids; (3) farming occurring on more ecologically and topographically marginal areas; and (4) cropping and management factor which are radically different (Millington, 2006).

Erosion can be divided into two components: rill and interrill erosion. Interrill erosion is caused by soil particles being detached by raindrops and transported by overland flow. Rill erosion, however, is the detachment and transport of soil particles by concentrated flow: it is a function of the shear of the water flowing in the rill (Lal and Elliot, 1994). Computer simulation models like the Water Erosion Prediction Project – WEPP (Nearing et al., 1989) – developed by the United States Department of Agriculture (USDA) require the input of two erodibility values for each soil type: interrill (Ki) and rill (Kr) erodibility. If the inputs are not available, WEPP includes two regression equations to calculate Ki and Kr, also based on soil properties like content of clay, silt, very fine sand, sand and organic matter (Flanagan and Nearing, 1995). The main objective of this study is to determine the interrill and rill erodibility values for a northern Andean highland watershed in Peru and to compare field measurements with existing models that describe erodibility.