Rainfall and land management effects on erosion and soil properties in traditional Brazilian tobacco plantations

ABSTRACT

No-tillage and inter-crops have been progressively introduced into traditional Brazilian tobacco plantations. However, there is a lack of information about their impact on soil erosion rates and soil properties. We studied 10 experimental plots in Paraná (Brazil) that rotated from no-tillage tobacco to two different inter-crop types (black beans and oats) and conventionally tilled tobacco to quantify erosion rates from September 2014 to February 2016. The results show that soil loss (18 Mg ha⁻¹) and runoff coefficient (8.3%) were higher under conventional tillage tobacco than under no-tillage tobacco (3.4 Mg ha⁻¹ and 0.6%). Bulk density was higher at the end of the cropping cycle than at the beginning. We concluded that conventional crops increased soil erosion, and the use of inter-crops and no-tillage is highly recommended for soil and water conservation. The findings should be valid for other regions that have similar cropping systems and environmental conditions.

KEYWORDS:

• traditional agriculture
• no-tillage
• runoff
• soil erosion
• rotated crops

Editor A. Castellarin; Associate editor A. Petroselli

1 Introduction

The past few decades have been revolutionary in terms of advances in agricultural production and the conservation of soil quality (Khaledian et al. 2017, Keshavarzi et al. 2018). These advances have been related to the introduction of new species (genetic engineering) and to the process known as “expansion of the agricultural frontier” (land conversion into croplands) (Girlanda et al. 2001); the second issue is particularly relevant in rapidly developing countries. These transformations are promoting constant changes in the traditional agrarian landscapes (Llausàs et al. 2009), making them increasingly more homogenous, such as cultivating the same species, using the same land-use management and enhancing the connectivity of land degradation processes (Fjellstad et al. 2001, Saleh et al. 2017).

The necessity of obtaining higher productivity from each piece of land has led to a process of intensification and, subsequently, degradation (Ferreira et al. 2015, Drabo 2017). Large expanses of land have been converted into cropland over the last 50 years, making it necessary to adapt land that is less suitable for agricultural purposes to cultivated areas (García-Ruiz et al. 2015, Panagos et al. 2017). Specifically, the introduction of new crops on unfit lands has provoked serious soil loss problems (Chartin et al. 2013). Moreover, high soil erosion rates have also been observed on lands that have been abandoned because profitable production was not achieved (Hatna and Bakker 2011). For example, in southern Spain, Lesschen et al. (2008) estimated soil erosion rates of about 87 Mg ha⁻¹ year⁻¹ in abandoned terrace fields due to gully generation. On the Loess Plateau, Liu et al. (2012) claimed that investigations published in the Chinese language quantified average soil erosion rates in watersheds with abandoned lands higher than 150 Mg ha⁻¹ year⁻¹. Cerdà et al. (2018a) observed that, in the year after abandonment, abandoned areas can register 2.9 times more erosion due to low vegetation recovery, resulting in 21 times less sediment yield after 9 years.

Lands characterized by fertile soils and low slopes that are most suitable for agricultural production have been progressively transformed into technified agro-industry complexes, as producers look for high economic profitability and high productivity (McMichael 1991). New research into techniques that minimize soil loss and land degradation is being conducted as well (Godfray et al. 2010). However, surveys of farmers’ perceptions show that they are expensive solutions that do not always have producer support (Cerdà et al. 2018b). So, this evolution from a traditional agriculture to the current systems has allowed improvements in production, improving sustainability in some cases, but its application is still a big challenge (Qiao et al. 2016).

In Brazil, mechanized agriculture coexists with traditional farms, where rudimentary practices and family labour are dominant features due to the difficulty of affording the mechanization process. In the rural areas of southeastern Paraná federal state (southern Brazil) this situation is common due to physical and economic mechanization restrictions. The clearing of Araucaria angustifolia (roça de toco in Portuguese) forests to develop agro-silvo-pastoral systems of faxinal (Thomaz and Antoneli 2015) and tobacco (Nicotiana tabacum) for leaf production (Antoneli and Thomaz 2014) are the most common traditional activities in these areas.

The annual cropping cycle for tobacco in southern Brazil starts during September when seedlings are planted and ends during February with harvest. Until the 1990s the land remained resting with bare soil after harvest until the beginning of the new cycle and, consequently, exposed to erosion processes. In the last 20 years, in order to avoid soil erosion during the resting period, conservation techniques such as the incorporation of new inter-crops have been adopted. It is common to use black bean (Phaseolus vulgaris) in March or maize (Zea mays) in May for silage and feeding livestock in winter. When the secondary inter-crop is harvested, the soil is tilled for new plantings of black oats (Avena strigosa) that serve as dead cover for the next planting of tobacco, restarting the cycle.

Tobacco fields have been considered an important source of sediments due to soil erosion for several decades (Costa 1975). In fact, Panagos et al. (2015a) assigned the highest values for the RUSLE (Revised Universal Soil Loss Equation) C-factor to tobacco, cotton and fallow lands. The introduction of no-tillage cultivation of tobacco in order to reduce soil loss rates has been studied for over 30 years in the United States (Wood and Worsham 1986). Other environmental and social problems related to tobacco leaf production, such as deforestation, land degradation, chemical pollution or food insecurity, have also been studied worldwide (Lecours et al. 2012). In addition to burning wood to generate heat for drying tobacco leaves, it is estimated that each Brazilian property consumes 70 m³ of wood each year (information provided by farmers).

The combination of no-tillage and conventional crops in rotation is due to the belief of farmers that long-term no-tillage causes soil compaction problems and consequently a decrease in productivity. They consider it necessary to mix no-tillage and inter-crops such as black bean because they need to get extra income throughout the year. Moreover, they consider that this management approach is the best way to incorporate nutrients and increase organic matter because fertilizer prices are too expensive for small family agriculture. Therefore, the main goal of this research was to quantify soil and water losses and their impact on soil properties, such as organic matter (OM), bulk density (BD) and soil pH, which are some of the main indicators of soil quality (Brevik 2009), during the process of transforming from no-tillage to conventional tillage with two different inter-crops (black beans and oats) within the common rotation cycle of traditional Brazilian tobacco crops. The main novelty of this research is related to the special conditions of tobacco plantations under tropical climate conditions and a non-common rotation inter-cropping system, from which a quantitative assessment of soil erosion rates does not yet exist. We hypothesized that changes in crops (types and activities) would produce different land degradation responses.

2 Material and methods

2.1 Study area

The study was carried out on a 5.4 ha commercial farm belonging to a family-run business, of which 2.5 ha were cultivated to produce tobacco leaves. It is located in the municipality of Ivaí, in Paraná, Brazil (Fig. 1). This area belongs to the Ponta Grossa Plateau, which is formed of Paleozoic sedimentary rocks with a mean elevation of 750 m a.s.l., an average slope of 12% and slope lengths that reach 180 m. The tobacco plantation is situated on a terraced farm built by European pioneers (Ukrainian and Polish peasants of the 19th century) (Antoneli and Thomaz 2009).

Figure 1. Geographical location of the study area within the municipality of Ivaí (Paraná, Brazil).

Climate conditions are characterized by mean annual rainfall of 2054 mm, with an average maximum value of 270 mm in January and minimum value of about 105 mm in August. Mean annual temperature is 18°C, with average maximum values during November, December and January (>22°C), and average minimum values in July (~9.4°C). Mean annual air humidity is relatively high, reaching values of 83%.

The soils are dystrophic red nitosols (EMBRAPA 2013) with a nithic B-horizon rich in low-activity clay to a depth of 50 cm. They are clayey-textured (50% clay, 31% silt and 19% sand) with a mean bulk density of 1.15 ± 0.19 g cm⁻³.

Paraná is the third leading tobacco leaf production state in Brazil after Rio Grande do Sul and Santa Catarina federal states, all of them located in southern Brazil (Silveira et al. 2012). The size of individual properties is usually less than 10 ha. Despite being the most traditional crop of this region, some difficulties still exist. For example, the occurrence of frost during winter does not allow year-round cultivation. Moreover, the most common landscapes are dominated by steep slopes (15–20%) and soil erosion is high after extreme rainfall events, so the tobacco is cultivated on temporal terraces built by farmers. In the steepest areas, these terraces are built using animal labour (e.g. mules) instead of tractors (Antoneli and Thomaz 2014).

2.2 Sampling design and soil erosion assessment procedures

This research was focused mainly on the estimation and quantification of soil loss rates (Mg ha⁻¹) and runoff (%). As several authors have affirmed, soil erosion has major impacts on soil properties (Dai et al. 2015, Martínez-Casasnovas et al. 2010, Panagos et al. 2015b). This study investigated OM (g kg⁻¹), BD (g cm⁻³) and pH, as they are some of the most important soil quality indicators with important influences on soil fertility, biology and aggregate stability (Bienes et al. 2016, Zeraatpisheh et al. 2017, Pulido et al. 2018). Moreover, driving factors of soil erosion, such as exposed soil (%), total daily rainfall (mm) and farming activities, were recorded either daily or weekly, depending on the intensity of the activities, from September 2014 (planting of no-tillage tobacco) to February 2016 (harvesting of conventional tobacco).

Soil losses were quantified using Gerlach troughs (Gerlach 1967) in a closed plot system, which allowed quantification of soil losses and surface flow after each rainfall event. They were installed in 10 individual plots of 1.30 m (width, distance between furrows) × 2.0 m (length), with a known-contribution area of 2.6 m², separated by a wooden structure to avoid external water and sediment fluxes. Water fluxes that originated within the erosion plots after each rainfall event were collected in the lowest part of the Gerlach trough and forwarded to and stored in 25-litre carafes (Fig. 2).

Figure 2. Technical characteristics of the installation of the Gerlach troughs: (a) planting, (b) harvesting, and (c) manual raingauge.

The material collected in the carafes was immediately transported to the laboratory after each rainfall event where it was air-dried and weighed to estimate soil loss rates (g m⁻² and Mg ha⁻¹) and runoff coefficients (%) for each trough as follows:

(1)

The percentage of bare soil was estimated in the centre of each erosion plot (n = 10) using a 1-m² wooden quadrat that had a grid of 100 mini-quadrats of 100 cm² each. Every 30 days, the same person made a visual estimation of bare soil by counting the number of mini-quadrats where the soil was uncovered. The results were expressed in percentage of bare soil (number of mini-quadrats that were uncovered). These values were then averaged to get a mean value for the 10 erosion plots. Rainfall was recorded daily using a manual raingauge of stainless steel (Ville de Paris type) with a reception orifice of 400 cm², located near the erosion plots at a height of 1.5 m.

We collected 20 soil cores (10 each for 0–5 cm and 5–10 cm soil depths), sampling 10 at the beginning of each crop plantation and the rest at the end of each harvesting period. The median values for each representative area (n = 5) were used for the statistical analysis to avoid sensitivity to outliers. Organic matter was measured by the wet combustion method of Walkley and Black (1934). Bulk density samples were collected in five representative areas around the erosion parcels using 98.17 cm³ soil cores (Coile 1936). A PH500 Benchtop® pH-meter was used to measure pH with a 1:2.5 soil/water ratio. The agricultural calendar was obtained from the owner of the farm and he was often interviewed during the monitoring period. Furthermore, the farmer was particularly careful to avoid any disturbance in the experiment areas (e.g. Gerlach troughs, pluviometer, etc.). So, all the information provided here is considered to be of high reliability.

3 Data analysis

Data were analysed all together and classified into four groups corresponding to the main crops: no-tillage tobacco (NT), black bean (B), oats (O) and conventional tobacco (CT). First, soil loss rates and runoff coefficient were depicted in box plots with bar chart graphics with the total daily rainfall. After that, soil losses for the different crops were represented in scatter plots and also analysed using linear regression. Statistically homogenous groups were identified using Fisher’s least significant difference post-hoc comparisons. Finally, soil properties at two different depths (0–5 and 5–10 cm) were compared for each group using an ANOVA test. If data did not pass the Shapiro-Wilks normality test, the Holm-Sidak test was used. Differences were considered statistically significant at α < 0.05. The SigmaPlot 13.0 version (Systat. Inc.) software package was used to analyse the data.

4 Results

4.1 Agricultural calendar

Agricultural activities were monitored based on the calendar supplied by the farmers (Table 1). During the research period, tobacco was planted with no-tillage from September 2014 to February 2015. Between the tobacco plantations, black bean was seeded from March 2015 to May 2015 and black oats were added in July 2015 as winter cover. In September 2015 tobacco was planted (conventional crop) after removing the layer of dead oats and building temporal terraces using mules. The harvesting of this conventional crop was done in February 2016. The no-tillage tobacco plantation marked the completion of a crop cycle. During this cycle, soils were not tilled and weeding was accomplished with herbicides. Black bean was seeded after that because the farmers needed some extra income. The seeding of oats was done to avoid leaving an exposed soil surface and to fix nutrients between cash crops. During this period, furrows were also made to prepare for the next tobacco plantation. The collection of tobacco leaves was made at least once a week from December to January using mules to transport the leaves.

Table 1. Agricultural calendar for the farming activities. NT: no-tillage; CT: conventional tillage. Month in the form mm/yy.

Month	Crop	Agricultural activity
09/14	NT tobacco	Desiccation of oats (dead layer). Tobacco was planted under oat straw (planting activities).
10/14		Weeding using herbicides. Soil was not stirred (covered soil).
11/14
12/14		Weekly collection of leaves and transport by mules to the farm warehouse.
01/15
02/15
03/15	Black bean	Soil grading with tobacco crop residues. Black beans were planted.
04/15		Weeding using herbicides. Soil was not stirred.
05/15
06/15		Black beans harvesting. Soil was turned using mules.
07/15	Black oats	Soil was turned using mules. Oats were planted (exposed soil).
08/15		Weeding using herbicides. Soil was not stirred.
09/15	CT tobacco	Oats was removed. Furrows were built. Tobacco was planted (exposed soil).
10/15		Soil was turned in the furrows using mules.
11/15		Use of a hoe for weeding in the inter-row areas.
12/15		Beginning of weekly collection of leaves and transport by mules to the farm warehouse.
01/16
02/16	CT tobacco	End of weekly collection of leaves and transport by mules to the farm warehouse.

4.2 Soil loss rates and runoff coefficient monitoring

A total of 45 rainfall events with an accumulated precipitation above 10 mm were recorded during the study period (September 2014–February 2016). Rainfall events with values less than 10 mm were not considered during this study because they did not produce surface runoff. Twelve rainfall events were recorded in the no-tillage tobacco crop, seven during the black bean crop, and seven in the oat crop. Finally, 19 rainfall events generated surface flow in the conventional tobacco crop.

Runoff coefficients are represented in box plots (Fig. 3). The lowest mean runoff coefficient (0.6%) was recorded in the NT tobacco plantation. For the black beans, the mean runoff coefficient was 10.8%. In the oat plantations, a lower runoff coefficient (4.0%) was registered than in the other inter-crop. Finally, a mean runoff coefficient of 8.3% was found for the conventional tobacco plantation.

Figure 3. Box plots of runoff coefficient per crop type during the studied period. The graphs above each box plot show the amount of rain that fell during each rainfall event.

Box plots are also used to depict soil losses (Fig. 4). In the CT tobacco plot, a total loss of 1795 g m⁻² (17.9 Mg ha⁻¹) was collected during the total period. Lower total soil erosion rates (507.7 g m⁻² or 5.1 Mg ha⁻¹) were recorded in the no-tillage tobacco plantation. A total of 434.4 g m⁻² (4.3 Mg ha⁻¹) and 400.6 g m⁻² (4.0 Mg ha⁻¹) were collected in black beans and oats, respectively. The maximum single event value of 251 g m⁻² (2.5 Mg ha⁻¹) was also recorded in the CT tobacco plot after a rainfall event of 84 mm d⁻¹.

Figure 4. Box plots of soil loss rates per crop type during the studied period. The graphs above each box plot show the amount of rain that fell during each rainfall event.

4.3. Relationships between soil loss rates, daily rainfall events and bare soil

Linear correlations and regression equations and fits of rainfall to soil loss rates are shown by crop in Figure 5. The highest correlation coefficients (moderate–high) with statistical significance between rainfall amounts and soil loss were found in black bean (R² = 0.56) and oat (R² = 0.62) crops. These values were lower in the tobacco cultivations, where rainfall amount did not seem to be the most influential factor on soil erosion. In the conventional tobacco tillage, rainfall events of 20–40 mm generated greater soil losses than in the other crops.

Figure 5. Scatter plots of rainfall (x-axis) and soil loss rates (y-axis) for different crops.

The influence of soil exposure was also analysed by correlating the percentage of bare ground surface to soil loss rates after each of the 45 rainfall events recorded (Fig. 6). The variables were generally less correlated as the biannual cycle progressed, i.e. the construction of furrows and soil disturbance led to greater soil exposure and changed the relationship between soil exposure and soil losses. The highest coefficients of correlation (statistically significant) were observed in the black bean (R² = 0.84; high) and no-tillage tobacco crops (R² = 0.61; moderate), respectively. The highest data variability was found in the conventional tillage tobacco, where rainfall events that occurred with a similar percentage of bare ground surfaces (~50% and ~80%) as in NT produced differences that were 0.2 Mg ha⁻¹ greater in soil loss rates. Except in the no-tillage tobacco crop, soil losses were only recorded when the bare soil surface was above 40%. Under no-tillage conditions, the soil surface was only partially uncovered (never reached more than 60%) but soil losses were observed even with a low degree of soil exposure (Fig. 6).

Figure 6. Scatter plots of bare soil percentage (x-axis) and soil loss rates (y-axis) for different crops.

Considering the data for all crops together, soil exposure was the variable that was best correlated with soil loss (r = 0.460, p < 0.001), while rainfall was poorly correlated with soil loss (r = 0.151, p < 0.01). The highest soil loss rates (~1 Mg ha⁻¹) were in October and November 2015 when soil was being disturbed for weeding before harvesting of the conventional tobacco plantation. During the harvest, soil loss rates around 0.70 Mg ha⁻¹ were recorded even though harvest was the activity that was supposed to have low soil losses in the conventional crops.

4.4 The effect of land management and soil erosion on OM, BD and pH

The mean values, standard deviations, and maximum and minimum values of OM, BD and pH at the beginning and end of each crop cycle and at different soil depths are given in Table 2. In addition, statistical differences were compared by an ANOVA and Holm-Sidak tests (Table 3). In NT tobacco, the highest OM values (>34 g kg⁻¹) were recorded at the end of the crop cycle, reaching values higher than 39 g kg⁻¹ in some cases. BD also increased with progression through the cropping cycle and at different soil depths, from 1.07 to 1.21 g cm⁻³ (0–5 cm) and 1.10 to 1.21 g cm⁻³ (5–10 cm). The pH became more acidic at the end of the cycle, descending 0.6 and 0.7 units at the 0–5 cm and 5–10 cm depths, respectively. Statistical tests demonstrated that these differences were significant.

Table 2. Evolution of some soil properties during the study. OM: organic matter; BD: bulk density; NT: no-tillage; CT: conventional tillage.

		NT				Bean				Oats				CT
	Soil depth (cm)	0–5^1	0–5^2	5–10^1	5–10^2	0–5^1	0–5^2	5–10^1	5–10^2	0–5^1	0–5^2	5–10^1	5–10^2	0–5^1	0–5^2	5–10^1	5–10^2
OM (g kg^−1)		25.2	34.3	21.7	35.4	26.9	28.3	28.5	28.2	29.0	32.2	28.2	32.0	19.7	20.9	23.2	31.5
		±2.7	±4.5	±3.2	±2.9	±3.6	±2.8	±2.3	±1.8	±2.3	±2.1	±3.0	±2.2	±1.3	±2.2	±2.8	±2.7
	Max	29.0	39.4	25.4	39.0	31.0	30.1	32.1	30.4	31.5	34.8	31.4	34.8	21.8	23.1	26.9	34.7
	Min	23.1	27.8	17.9	32.4	21.7	23.4	25.7	26.1	26.4	29.7	23.7	29.3	18.7	17.9	20.7	26.9
BD (g cm^−3)		1.07	1.21	1.08	1.18	1.02	1.09	1.05	1.16	1.00	1.05	1.01	1.16	1.01	1.22	1.02	1.20
		±0.05	±0.02	±0.02	±0.03	±0.03	±0.03	±0.05	±0.04	±0.06	±0.02	±0.05	0.04	±0.03	±0.02	±0.06	±0.04
	Max	1.12	1.24	1.10	1.21	1.06	1.14	1.09	1.21	1.07	1.07	1.08	1.21	1.03	1.24	1.08	1.24
	Min	0.99	1.18	1.06	1.13	0.98	1.05	0.97	1.11	0.94	1.01	0.94	1.11	0.97	1.19	0.96	1.14
pH		6.2	5.5	6.4	5.8	6.5	6.3	6.2	6.3	6.3	6.2	6.3	6.0	6.0	6.1	6.0	5.7
		±0.2	±0.3	±0.2	±0.1	±0.3	±0.3	±0.3	±0.2	±0.2	±0.2	±0.4	±0.3	±0.3	±0.3	±0.2	±0.3
	Max	6.3	5.9	6.7	5.9	6.9	6.6	6.7	6.7	6.7	6.4	6.8	6.4	6.4	6.4	6.3	6.1
	Min	5.9	5.2	6.2	5.7	6.2	5.9	5.8	6.1	6.1	5.9	5.7	5.5	5.7	5.7	5.8	5.3

¹Soil properties at the beginning of the crop season; ²soil properties at the end of the crop season.

Table 3. Analysis of the statistical differences between soil properties. Bold indicates statistical differences were found. NT: no-tillage; CT: conventional tillage; OM: organic matter; BD: bulk density.

	NT		Bean		Oats		CT
	0–5	5–10	0–5	5–10	0–5	5–10	0–5	5–10
OM	p < 0.005*	p < 0.001*	p < 0.512	p < 0.782	p < 0.053	p < 0.029*	p < 0.349	p < 0.001*
BD	p < 0.001*	p < 0.001*	p < 0.012*	p < 0.003*	p < 0.135	p < 0.001*	p < 0.001*	p < 0.001
pH	p < 0.002*	p < 0.001*	p < 0.296	p < 0.596	p < 0.578	p < 0.288	p < 0.84	p < 0.082

*Normality test failed and Holm-Sidak test was conducted.

In black beans an increase in OM was recorded at the 0–5 cm depth (+1.4 g kg⁻¹), with a decrease at the 5–10 cm depth (−0.3 g kg⁻¹). No differences in pH values were noted. On the other hand, statistical differences were noted for BD values at different soil depths (p < 0.012 at 0–5 cm; p < 0.003 at 5–10 cm). In oats, significant differences were found only in the 5–10 cm layer for OM and BD, where a drastic OM decrease from 32.0 to 19.7 g kg⁻¹ and a BD increase from 1.01 to 1.16 g cm⁻³ were noted, respectively. Finally, significant changes (p < 0.001) in OM at the 5–10 cm soil depth were detected in CT tobacco, with values that changed from 23.2 to 35.1 g kg⁻¹ from the beginning to the end of the cropping cycle. BD also showed significant differences (p < 0.001) in CT tobacco, changing from 1.01 to 1.22 g cm⁻³ in the 0–5 cm interval and from 1.02 to 1.20 g cm⁻³ in the 5–10 cm interval from the beginning to the end of the cropping cycle. Finally, soil pH demonstrated similar values in CT tobacco from the beginning to the end of the cropping cycle and did not show significant differences with soil depth.

5 Discussion

Most studies of soil erosion in Brazilian tobacco production have been made in Paraná state, highlighting that the number of farmers cultivating tobacco has increased in the past few decades (Kraiczek and Antoneli 2012). These studies have generally been published in Portuguese in Brazilian journals. Several soil parameters have been analysed, such as infiltration (Zaluski and Antoneli 2014) and penetration resistance (Antoneli et al. 2017) in grazing areas and conventional tillage farms (Antoneli and Bednarz 2010, 2016, Antoneli and Thomaz 2014). However, an evaluation of the temporal evolution of soil erosion rates in a complete traditional tobacco plantation cycle has not previously been published.

This research demonstrated that there is a positive effect of no-tillage on the reduction of soil losses compared with conventional tillage management. No-tilled tobacco plots recorded an average soil loss of 5.1 Mg ha⁻¹ (runoff coefficient: 0.6%) as compared to the 17.9 Mg ha⁻¹ average loss in conventional tillage tobacco. These results are similar to those in several other works that compared conventional tillage and no-tillage crops in different land systems throughout the world (e.g. Blavet et al. 2009, Labrière et al. 2015, Bartimote et al. 2017), including Brazil (Trabaquini et al. 2015, Anache et al. 2017, Rocha Junior et al. 2017). We observed that no-tillage was as effective as mulching in the prevention of erosion. However, there are authors who have demonstrated that no-tillage can enhance soil erosion and runoff and reduce infiltration because of the generation of soil crusts and heavy soil structures and aggregates, as shown in olive orchards and abandoned lands (Taguas et al. 2015, Kou et al. 2016).

The introduction of conventional tillage tobacco reduced the soil organic matter content over the short term, even when it was preceded by an oat crop that reached OM values recorded at the end of the no-tillage tobacco crop. The OM content of the conventional crop was quite low compared to the other crops. This may have been a consequence of the removal of grass cover during harvest, i.e. black oat residues were removed before planting and weeding operations in the CT tobacco. Other researchers have also found higher soil OM in NT systems as compared to CT (Varvel and Wilhelm 2011, Carr et al. 2015). Finally, soil losses mean soil organic matter accumulated in the top 5 cm can be transported (Berhe and Kleber 2013). So, no-tillage can be considered a good practice to maintain some parameters related to soil fertility, such as OM or pH.

The soil was always at least partially covered (>50%) by grasses during the no-tillage crop cycle. This prevented any rainfall event from producing soil loss rates above 1 Mg ha⁻¹, values that are within the tolerable limits of soil loss proposed by Verheijen et al. (2009) in Europe. In the conventional crop, percentages of bare soil above 60% were often recorded. At these percentages it is common to find soil loss rates >1 Mg ha⁻¹, even in areas that are not very humid and when the rainfall intensity is not overly high (Schnabel et al. 2009). Therefore, the importance of bare soil in allowing high levels of soil erosion, as has been found in other studies (Bagarello et al. 2013, Biddoccu et al. 2016), was confirmed in this study. Additional studies on splash effects related to rainfall intensity could be helpful to provide a better understanding of soil erosion processes in Brazilian tobacco production.

The significant increase of soil erosion rates after new plantings due to soil detachment, among other processes, has recently been highlighted in other types of crops such as vineyards (Rodrigo-Comino et al. 2018). In this multi-crop system, where tobacco is combined with other plants, soil detachment is difficult to avoid, although direct planting into oat residues (no-tillage) could be a good option to reduce these negative effects.

Our results suggest some soil properties can be improved, and farmers state they can get extra income by introducing new crops such as black beans and oats to the cropping rotation. Finally, although soil OM (Conforti et al. 2013), BD (Pulido et al. 2017) and pH (Zornoza et al. 2015) are important indicators of soil quality, further research using additional properties (Brevik 2009) would be useful in forming a more complete understanding of the impact of tobacco rotations on soil quality in Brazil.

6 Conclusions

The impact of alternating different crops and agricultural practices on soil erosion and quality were studied at an intra-annual scale in a tobacco-based agricultural system in southern Brazil. The commonly known positive effects of no-tillage cropping on the reduction of soil losses and on improving selected soil properties were confirmed. In addition, inter-cropping with black bean and oats showed benefits in terms of reducing soil erosion and improving soil properties as well as improving profitability for the farmers. The percentage of soil exposure was the main factor driving soil erosion processes. Finally, no evidence was found to indicate that soil compaction is reduced when conventional tillage is included with no-tillage management, which is a belief of many farmers. A wider study covering a longer time period and involving more soil properties is still needed in order to draw definitive conclusions. We acknowledge that our results could be adapted to other tropical regions with tobacco plantations and inter-cropping changes at an intra-annual scale.

Acknowledgements

We thank the farmers from Ivaí community for their collaboration in this research. We would like to thank Ilan Stavi for his useful comments on the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The study was financially supported by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq-Brazil; Grant no. 457595/2014-0 MCTI/CNPQ/Universal 14/2014 - Faixa A).