Impact of topography and sedentary swidden cultivation on soils in the hilly uplands of North-East India

2.1 Study area

A hilly catchment (4.0 km²) was selected to investigate the topographical and land use impact on soil properties in the central part of the Meghalaya Plateau (Figure 1). The catchment bottom is located at about 1,750 masl, and hills topped by flattened and wide ridges rise to 1,850 masl. Slopes are straight and range from 100 to 150 m in length with a predominant gradient of 20°. Flat terrain (0°–2°) occupies 12% of the catchment, one third of which falls within the ridges, whereas the rest constitutes the valley bottoms. Steep slopes (>10°) represent 60% of the total. According to the Köppen system, the climate is classified as Cwb, subtropical with a dry winter. The average annual temperature is 14 °C, with some occurrence of minimum temperatures below freezing in winter. The summer monsoon between June and September provides 80% of the area's annual 2,400-mm average rainfall (Prokop & Walanus, 2015). The catchment is underlain by deeply weathered (up to 15–20 m) granites that form one of many intrusions in quartzites and gneisses of the plateau basement (Mazumder, 1986). The soils have been classified as Ultisols (Soil Survey Staff, 2014). Similar to Ultisols found in other geologically old landscape settings, they are acidic and strongly leached, with relatively low native fertility (Bhaskar et al., 2004).

2.2 LULC and sedentary swidden cultivation system

The LULC of the study catchment was prepared on the basis of Google Earth (DigitalGlobe) satellite image at spatial resolution of 0.5 m from 10/29/2015 digitised at scale of 1:5,000 and verified during field survey. LULC consists of cultivated fields (40%), grasses and bushes (40%), native deciduous forest (6%), pine forest (6%), and rural settlement (8%). The study catchment, like the upper part of the Meghalaya Plateau, was deforested due to charcoal production for iron smelting several hundred years ago (Prokop & Suliga, 2013). Thus, the natural vegetation in the catchment is restricted to subtropical deciduous forest preserved as a part of the religious beliefs of local inhabitants (Bor, 1942; Tiwari et al., 1998). This forest has a rich canopy of trees and a weakly developed undergrowth. Patches of secondary pine forest with dense grasses and weedy colonisers in the undergrowth grow in fallow land. Forests are surrounded by cultivated fields situated on the flattened ridges and steep slopes, and valley bottoms are occupied mainly by settlements and roads.

A shortage of agricultural land exists because of a high population density (~400 people km⁻²), and the increasing food demands of Shillong, the capital of Meghalaya State, have caused an intensive sedentary cropping (Government of India, 2011; Tiwari, 2003). The new swidden cultivation system, called bun, shows adaptation to limited biomass (forest) and land resources. The system still retains elements of traditional shifting cultivation; it encompasses a fallow period as well as slash and burn management, but only grasses collected on the fallow land and the lower branches of the nearby pines instead of entire trees. In the bun system, potato (Solanum tuberosum) is the dominant crop, cultivated in two seasons: February–June and July–December (Dubey & Sah, 2009; Tiwari, 2003). In December, the slashed vegetation is arranged in parallel rows for drying. In January, a 10–15-cm thick soil layer from the cultivated field is placed on top of the dried material with a hoe to form rows of flat-topped raised beds measuring around 1 m wide, 20–25 cm in height, and few metres long. The furrows to drain excess rain water between beds are about 40 cm wide. Burning the slashed organic material is conducted under soil cover (Figure 2). Sowing of usually two rows of potatoes in the wide beds is performed in February, and the crop is harvested by June. Field preparation for the second season includes manual turning of the soil and weeding. Sowing potatoes is completed by the end of July, and harvesting takes place in November–December. The land is cultivated for 1–3 years and then left fallow for the same period.

2.3 Soil sampling design

The study catchment and sample fields were selected based on topographical criteria, land use history, and cultivation practices (Bruun, Mertz, & Elberling, 2006) obtained from literature, topographical maps, field surveys, and farmer interviews. The catchment is representative of land use structure (Prokop, 2016) and landform contribution (Migoń & Prokop, 2013) for higher elevations of the Meghalaya Plateau and granite areas, respectively.

Data on spatial distribution of natural forest patches (Bor, 1942; Tiwari et al., 1998) combined with analysis of topographic maps for 1910, 1966, and 1980 at scales of 1:63,000, 1:50,000, and 1:25,000, respectively, indicated that the natural deciduous forest and cultivated land had the same land uses for at least 100 years. Field observations, interviews with farmers, and the literature indicated that intensification of potato cultivation (introduced to Meghalaya in the 1830s) was related to population pressure in recent decades (Mishra & Ramakrishnan, 1983a; Syiemlieh, 1989; Tiwari, 2003). Tree ring calculations of several fallen pine trees, together with field observations since 2008 and farmer interviews, estimate the age of the pine forest as 15 years old.

The soil sampling scheme assumed that long-term natural soil forming factors such as climate, parent material, and time are relatively uniform over the study catchment. In addition, the cultivated land also has the same bun cultivation, with potatoes as monocrop. Using this study design, we expected that differentiation of soil properties could be attributed to two contrasting landforms (flat ridges and steep slopes) and three land use types within each (natural deciduous forest, cultivated land, and secondary pine forest). The valley bottoms and footslopes were not considered because of settlement occupation and lack of pine forest. Studies at distances of 2–25 km north and south of the selected catchment showed that ridges are sufficiently wide and flat and their central parts are not affected by intensive soil erosion (Froehlich, 2004; Prokop & Poręba, 2012). Soil erosion increases where the ridge passes into slope and reaches the highest intensity in the most inclined part of the slope.

In November 2015, when the fields were levelled by farmers, soil samples were collected from two landforms: the central part of the flat ridges and steep slopes and the adjacent middle part of 120-m length straight slopes (~20°) under each of three LULC types (natural deciduous forest, cultivated land with potatoes, and 15-year-old pine forest; Gafur, Jensen, Borggaard, & Petersen, 2003; Mertz et al., 2008; Figure 3). Areas along the catena between ridge and midslope always had homogenous LULCs typical for catchment (i.e., only natural forest, pine forest, or potato cultivation). Within both landforms for each land use, 10 soil samples were collected using an 8-cm -diameter corer (i.e., 60 soil samples in total). Sampling was confined to the upper 20 cm of soils.

2.4 Data analysis

The samples were air-dried and passed through a 2-mm mesh sieve, and selected physico-chemical soil properties were analysed. The grain-size composition of each sample was determined using the combined sieving method and a Malvern laser particle sizer after pretreatment with H₂O. Soil bulk density was calculated by drying the soil ring samples (100 cm³) at 105 °C before weighing. Total C and N contents were determined by combustion in a CHNS vario EL III Element Analyzer. Total P and K contents were analysed by flame atomic absorption spectrophotometry and flow injection analysis, respectively (Ruzicka & Hansen, 1988). The exchangeable K, Ca, Mg, and Na in the solutions were measured using atomic absorption spectroscopy (Sparks, Page, Helmke, & Loeppert, 1996). The exchangeable Al was extracted with KCl and then determined by titration (Black, 1965). Cation exchange capacity (CEC) was determined by the ammonium acetate method (Aprile & Lorandi, 2012), and pH values in H₂O were measured electrometrically in a 1:2.5 soil/water mixture. The colour of the wet soil was described according to the Munsell system.

Data were examined for normal distribution using the Shapiro–Wilk test. One-way analysis of variance and parametric t tests were then performed. If the analysis of variance showed significant differences in soil properties between the land use types, means were compared with post hoc Tukey's honestly significant difference tests at p < .05.

2.5 Soil quality indices

The soil evaluation factor (SEF) and soil deterioration index (SDI) were calculated to quantify the degree of soil fertility and degradation. The SEF index was developed to assess various soils' fertility status under succession of secondary tropical forest in the Amazon region of Brazil (Lu, Moran, & Mausel, 2002). The suitability SEF to evaluate soil degradation was also verified in different regions of India (Pal, Panwar, & Bhardwaj, 2012; Panwar, Pal, Reza, & Sharma, 2011). The equation of SEF is expressed as (Lu et al., 2002):

The unit of SEF for exchangeable Ca, Mg, K, and Al is cmol_c/kg, and organic matter (OM) is a per cent. SEF indicates that high amounts of K, Ca, and Mg are facilitating vegetation growth, whereas high Al content restricts vegetation development. Accumulation of soil OM reflects nutrient availability, soil structure, and moisture. SEF values below 5 are considered the threshold of extremely poor soil fertility.

SDI evaluates the soil status under a new land use through a relative comparison with the native vegetation (Adejuwon & Ekanade, 1988; Islam & Weil, 2000; Lemenih, Karltun, & Olsson, 2005). Assuming that cultivated land and pine forest were transformed from a reference land cover (i.e., deciduous forest), the percentage of deviation of each soil property between reference and new land use type is calculated using the equation (Adejuwon & Ekanade, 1988):

where SDI (%) is the soil degradation index, x_i' is the value of the ith soil properties for the reference land use, x_i is the value of the ith soil properties for new land use, and n is the number of soil properties taken into account.

The texture, C:N ratio, and exchangeable Na are not included in SDI calculation to reduce the dimensionality of the variables and to eliminate their intercorrelations. A descending function was used for bulk density and exchangeable Al because their higher values often indicate soil deterioration.

3.1 Effect of land use on soil properties within flat ridges and steep slopes

Soils were dark brown (10YR4/4–7.5YR4/4), deep, and well-drained under all land use types (Table 1). Their grain-size composition was predominantly silt loam, except the sandy loam texture observed in deciduous forest on slopes. The silt and clay content was significantly higher on cultivated land and pine forest within both landforms and was the most visible difference in physical soil properties between land use affected by tillage and natural forest. The bulk density of soils varied in a narrow range between 0.8 and 1.0, with significantly higher values on cultivated land and deciduous forest on ridges and slopes, respectively.

The soils under all land use types were acidic, with pH between 4.4 and 4.9; however, significantly lower pH was found in soils in deciduous forest on ridges. Generally, total C and N contents of soils were significantly higher in pine forest, followed by cultivated land and deciduous forest within both landforms. The C/N ratios ranged from 10.79 to 13.16 but were not significantly different. High contents of total P and K were related to supply from weathering parent materials that contained alkali and plagioclase feldspars. Exchangeable Al values were highest in soils in pine forest and deciduous forest on ridges and slopes, respectively. Significant differences were observed only on the slopes. Generally, exchangeable K, Ca, and Mg values were highest on cultivated land on ridges and lowest on slopes, in contrast to soils in deciduous forest that had a relatively low nutrient content on ridges but higher on slopes. CEC and base saturation values were below 10% and 31%, respectively, under all land use types. Both measures were greatest on cultivated land on ridges and in deciduous forest on slopes.

3.2 Effect of landforms with the same land use on soil properties

An analysis of paired-sample t tests indicated a significant difference in most of the soil properties between ridges and slopes with the same land use (Figure 4). Soils on ridges contained less sand but more silt and clay than soils on slopes under all land use types. The greatest differences were observed in the case of soils under deciduous forest. Bulk density was not significantly influenced by landforms.

Soils on ridges were significantly more acidic only under deciduous forest. They also contained more total C (by 23.8%) and total N (by 16.7%) than soils on slopes. The differences in total C and N content at sites affected by agriculture between the two landforms were even greater. Soils on ridges contained significantly more total C (by 33.6%) and total N (by 22.7%) on cultivated land as well as more total C (by 21.8%) and total N (by 22.7%) in pine forest than in soils on slopes. A similar relationship was observed for exchangeable Al. Alternatively, soils in deciduous forest on ridges had lower values of total P, K, exchangeable K, Ca, Mg, CEC, and base saturation than on slopes; however, statistically significant differences were noted only for total K and base saturation.

In contrast to natural deciduous forest, soils affected by agriculture usually had higher nutrient content on ridges than soils on slopes and were particularly visible on cultivated land. In pine forest soils, the nutrient pattern was more complex. Higher total P, K, exchangeable K, Na, and CEC were noted on ridges, whereas exchangeable Ca, Mg, and consequently base saturation were lower on ridges than slopes.

3.3 Indices of soil fertility and deterioration

SEF values ranged between 7.3 and 13.3 (Figure 5). Only in natural deciduous forest were the SEF values similar within both landforms. Generally, sites affected by agriculture (i.e., cultivated land and pine forest) had higher SEF values on ridges than on slopes. On ridges, the SEF value was higher on cultivated land than in deciduous forest by 64%. The SEF value in pine forest was slightly lower than that on cultivated land but still higher than in deciduous forest by 28%. By contrast, on slopes, the SEF value was 16% lower on cultivated land than in soils in deciduous and pine forests.

SDI values ranged between 30% and −10% (Figure 6); values were positive on cultivated land on ridges but negative on slopes. In 15-year-old pine forest, soil quality improvement was visible on ridges, but soil deterioration was evident on slopes; however, soil deterioration on slopes was less severe for pine forest soils than for soil on cultivated land.

4.1 Impact of sedentary swidden cultivation system within different landforms on soils

A general pattern of soil response to sedentary cultivation system was decoupled by landform at higher elevations of the Meghalaya Plateau. The soils within flat ridges under sedentary swidden cultivation had the highest nutrient concentrations, followed by soils in pine forest, representing fallow land and natural deciduous forest. The high soil fertility found in the long-term cultivated flat terrain can be attributed to several causes, such as landform features, contribution of cultivation system, and natural properties of Ultisols (Gafur et al., 2003; Islam & Weil, 2000; Mertz et al., 2009).

Ridges were up to 20-m-wide flat landforms, where overland flow was small and unable to move large amount of soil particles to lower topographic positions (Froehlich, 2004; Prokop & Poręba, 2012). Lack of significant erosion allowed soil to accumulate and gradually release a substantial amount of organic matter and nutrients.

Burning slashed organic material under soil cover causes carbonisation and charcoal formation (Lehmann, 2007). Studies have demonstrated positive effects of charcoal on soil physical characteristics, including bulk density, water retention (Novotny, Maia, Carvalho, & Madari, 2015), and soil chemical properties, such as a long-term increase in the CEC (Glaser, Lehmann, & Zech, 2002). Burning organic material under soil cover also reduces loss of volatile nutrients and ash drift more efficiently than open burning (Mishra & Ramakrishnan, 1983a, 1983b). Regular soil gains organic matter-derived C, N, and nutrients from surrounding fallow land, offsetting the losses from deforestation and further crop removal.

Maintenance of monocultures homogenises Ultisols grain-size composition by incorporating silt and clay from the subsoil into topsoil, a process visible on cultivated land and in 15-year-old pine forest. The fine silt and clay particles facilitate formation of organo-mineral complexes and save C from microbial oxidation (Grigal & Berguson, 1998). In effect, SEF values on flat ridges under bun cultivation are twofold higher than those measured under monocropping on more fertile Entisols and Alfisols in India (Pal et al., 2012; Panwar et al., 2011) and similar to those in naturally growing forest Alfisols in Amazonia (Lu et al., 2002).

4.2 Soil fertility in the light of complex indices

Sedentary swidden cultivation improves fertility of naturally poor Ultisols. Regular supply of nutrients from organic matter burned under soil cover allows long-term cultivation of flat terrain. It is also evidenced by the SEF and SDI values on flat ridges in 15-year-old pine forest. The SEF and SDI were still higher in 15-year-old pine forest than in natural deciduous forest, despite the lack of burnt organic matter supply. The grain-size composition was similar to that on cultivated land, but total C and N contents were higher than in soils under all other land use types, indicating a relatively fast recovery of topsoil chemical properties in secondary pine forest with well-developed undergrowth.

Generally, soils on steep slopes under cultivation and under pine forest showed deterioration of physico-chemical properties compared with those on flat ridges. The greatest soil deterioration on the basis of SEF and SDI occurred on cultivated land. Its low values indicate that an effective sedentary swidden cultivation system on flat ridges loses most of its advantages on steep slopes, where overland flow and soil erosion are accelerated by removal of natural vegetation and further cultivation. Decrease of soil quality in the light of SDI, resulted from reductions in organic matter due to fire, deforestation, tillage, and accelerated erosion, was also observed in adjacent Bangladesh (Islam & Weil, 2000) and highlands of Ethiopia (Lemenih et al., 2005).

The role of forest protection against soil degradation on slopes has been clearly demonstrated, however. The SEF fertility values in natural deciduous forest and pine forest were higher than those in cultivated fields, and the SDI negative values in pine forest indicated soil deterioration but was lower by half than on cultivated land.

Conditions in 15-year-old secondary pine forest with developed undergrowth were even more suitable for C and N accumulation than in deciduous forest. Grasses and shrubs usually decrease the impact of erosion and facilitate deposition of organic matter (Fu, Liu, Chen, Lü, & Qiu, 2004). The low negative SDI value indicated that secondary pine forest was able to reduce soil erosion and leaching; therefore, the inclusion of agroforestry elements with pine trees in the bun system can mitigate the rate of soil degradation on steep slopes (Grogan et al., 2012; Tripathi, Pandey, & Tripathi, 2009). A similar solution is already in practice in the Nagaland State in the North-East India, where farmers have adopted alder trees that contribute to soil nutrient enrichment during shifting cultivation (Ramakrishnan, 1992).

The contrast in soil fertility between flat ridges and steep slopes indicates that spatial susceptibility to soil degradation depends on the proportion of both landforms in the study catchment. Areas with slopes steeper than 10° are defined as steeplands (Lal, 1990), which when cultivated in tropical conditions are prone to degradation. A high contribution of steep slopes (60%) and a small share of flat areas (12%) create pressure to use slopes for cultivation. Considering the 90-km² granite batholith as a whole, the percentage of land surface belonging to a steep slope gradient class is similar (Migoń & Prokop, 2013); thus, high soil erosion rates can be expected under sedentary swidden cultivation in the region.

4.3 Soil erosion on deep weathered granites in the context of sustained swidden cultivation

Potato production accelerates overland flow and soil erosion (Ruysschaert, Poesen, Verstraeten, & Govers, 2004). The studied sedentary swidden cultivation system employs a downsloping tillage direction and a harvest and sowing time that coincides with the peak of summer monsoon rainfall. The resulting bare soil exposed to high rainfall facilitates overland flow and erosion. The dominant role of slope wash in land degradation related to cultivation is also supported by reduction of soil thickness compared with that of reference sites on slopes and thick colluvia at the footslopes (Migoń & Prokop, 2013; Prokop & Bhattacharyya, 2011; Figure 7).

Although soil erosion was not measured directly on studied granite hills, the erosion measurements on Inceptisols developed over quartzites surrounding granite batholith under the same cultivation system provide information on possible erosion rates. Mishra and Ramakrishnan (1983a) found soil losses up to 56 Mg·ha⁻¹·year⁻¹ on 40° slopes near Shillong. The estimated medium-term (using the ¹³⁷Cs tracing method) annual soil loss ranged from 32 to 79 Mg·ha⁻¹·year⁻¹ on slopes from 2° to 27° located only 2 km north of the studied catchment (Prokop & Poręba, 2012). These soil losses are of similar magnitude to those reported for manual tillage of steep slopes in intensive shifting cultivation in South-East Asia and Latin America (Forsyth, 1994; Nagle, Lassoie, Fahey, & McIntyre, 2000) and agree with an observed increase in soil erosion with increasing slope gradient (Dupin, de Rouwb, Phantahvong, & Valentin, 2009) and cultivation of root crops downslope (Zhang, Lobb, Li, & Liu, 2004). Soil erosion rates indicate removal of a 0.5-cm soil layer per year on bun cultivated slopes; therefore, some parts of steep slopes, where soil depth is only 20 cm, can be eroded to the quartzite bedrock within 40 years (Prokop & Poręba, 2012).

The effect of soil erosion on cultivated slopes on Ultisols developed over granites is notably different. As the surface horizon is partly eroded, deeply weathered subsurface horizons are incorporated into the topsoil during tillage (cf. Figure 7). An abundance of deeply weathered (up to 20 m) parent material results in a lack of exposed granite bedrock at the surface, despite the soil erosion. (Migoń & Prokop, 2013). The naturally lower soil fertility in the subsurface horizon is improved by adding and burning slashed organic matter, allowing longer soil cultivation than in shallow soils developed over neighbouring quartzites.

4.4 The role of sedentary swidden cultivation with potato monoculture in soil degradation of the Meghalaya Plateau

Potatoes are shallow-rooted plants with roots concentrated mostly in the tillage layer (Fageria, Baligar, & Jones, 2010). The crop prefers sandy well-drained and moderately acidic conditions with pH >5.5; pH below 4.8 impairs growth, and lower pH values can cause aluminium ion toxicity.

Nutrient uptake by potato plants and its removal with tubers is closely related to yield. The annual average productivity of potato in the Meghalaya State is only 0.9 kg/m² far below the India average of about 1.8 kg/m² (Saxena & Mathur, 2013).

This low yield means that potato plants absorb only about 0.007 and 0.008 kg/m² of N and K, respectively, throughout the growing season (Dean, 1994). P and the minor elements are required in relatively smaller quantities, less than 0.002 kg/m². The annual removal of nutrients with tuber harvest in the studied cultivation system can reach about 0.0017 and 0.046 kg/m² of N and K, respectively (Rosen, 1991); removal of other nutrients varies between 0.0005 kg/m² for P and 0.0001 kg/m² for Ca and Mg. Low pH, not nutrient stock, is likely a major cause of low potato yields. High concentration of Al in soils where pH is less than 5.5 can reduce Ca and Mg uptake and decrease root growth (De Wit, Eldhuset, & Mulder, 2010).

Low potato yields combined with the growing food demand forced farmers to expand the cultivation area to maintain per capita production and income (Tiwari, 2003). Areas cultivated for potato crops grew in Meghalaya State from 15,000 ha in 1971 to 18,500 ha in 2006 (Directorate of Economics and Statistics, 2007), or about 8.6% of the total cultivated area of the State, which is the highest among all potato-producing states in India (Dubey & Sah, 2009). Most of the new cultivation areas are steep slopes, the dominant landform in the Meghalaya Plateau. In effect, potato cultivation increases the risk of soil degradation in the region.

Intensifying cultivation also affects food security, both positively and negatively, at various spatial and temporal scales (Behera et al., 2016). The short-term economic efficiency as well as sustainable cultivation on flat areas contrasts with soil degradation risk on steep slopes where the new system may become less productive in the long run (Prokop & Poręba, 2012; Tiwari, 2003). General vulnerability of monocrops to pests and diseases and reduction of the dietary diversity in local population are already evident (Behera et al., 2016; Department of Agriculture, 2006).