Geographic location and data of the study area

The study zone covered four municipalities in the center-west of the province of Málaga (Figure 1).

Figure 1 Location of the study area according to ^{JJAA (2005)}. 

The area is in the Intermediate Units of the Baetica systems, in the Flysch of the low Teba zone of the Guadalhorce River. According to ^{Gómez (1987)} this unit is composed by conglomerates, loams, clays, sands, limestones with Microcodium and silexite from the Early Miocene. For the study, the areas with greatest olive cultivation aptitude were selected, located on the clay-loam relief of the flysch, of soft morphology and medium slope (25 %) and with punctual maximums that can reach 40 %.

The climate is temperate Mediterranean with mean annual temperature of 18.4 °C and annual precipitation of 636 mm, with a water deficit period from April to September. The soils are classified as cambisols and calcaric regosols, rather homogeneous, clay or silty clay texture, content of organic C between 1 and 1.5 % in the superficial horizons, capacity of cationic exchange frequently lower than 25 cmol kg^-1, and saturation of calcium ions from its lithological nature (^{LUCDEME, 1994}, ¹⁹⁹⁵, ¹⁹⁹⁶).

The olive grove for table olives is the main crop in the study area, with 53.2 % (3768 ha) of the total surface farmed. The management is with conventional cultivation systems primarily, and ecological. As part of the study, soil management was analyzed as well as the biomass that grows under the olive trees in both systems.

In the ecological cultivation system, actions were carried out to control the growth of the biomass and to give superficial treatment to the soil (Figure 2): 1) mechanical weeding or by grazing through livestock introduction; 2) shallow tillage (< 15 cm), such as weeding, the tillage depends on the precipitations and density of the plant cover, generally one or two superficial tasks are carried out (< 15 cm), regulated with a tractor and using the cultivator of semi-suspended flexible arms; and 3) superficial tillage with dredges or bulldoze rollers to eliminate the retraction cracks, to adapt the terrain for the collection and to avoid the growth of new covers. In some cases, ecological farms treat plant cover with superficial tillage and weeding, or only with one of the two, which has allowed the management of the ecological productive system with tillage and without it.

Figure 2 Temporality in the management of ecological farms. 

Conventional olive grove cultivation controls the biomass aggressively with the depth of plowing. Adventitious plants are controlled through tillage, with the same tool than in the ecological system, without controlling depth, which is why the tillage is deeper (15 to 20 cm). Two or three tills are carried out after the collection and prolonged until June (Figure 3). The tillage with rollers has the same characteristics than in the ecological system.

Figure 3 Temporality in the management of conventional farms. 

The herbicides are used only in rainy years, when the higher growth of adventitious plants made tillage with machinery difficult, primarily in plots with greater slope. This management decreases the contribution of organic matter.

Study method of soil erosion by water

The method used (^{Blanco and Aguilar, 2015} and ²⁰¹⁶), quantified the erosion in the plot with a double analysis: 1) analysis of the erosion status of the soil (percentage of the surface affected by erosion), and 2) quantitative analysis of the soil losses (volume of the soil lost in m³ ha^-1). The method of soil erosion analysis in rills and gullies through ^{Hudson (1993)} volumetric measurements examines, through visual indicators, the types of erosion (splash, sheet, rills and gullies) and the processes that affected the soil (mechanical alteration of the soil through tools and from trampling, sediment depositions), which codifies through an index that indicates the type of process, and a sub-index, which provides complementary information (susceptibility to erosion, type of erosion, type of protective plant coverage, among others) (^{Blanco and Aguilar, 2015}).

Sampling and statistical analysis

The sampling was carried out in 36 plots of 15 farms, where the lithological and soil characteristics were kept homogeneous to reduce the variability of starting point environmental conditions. The influence of the following factors on erosion was analyzed: cultivation system and soil management (conventional tillage/no-tillage), slope gradient, plant coverage at tree height and superficial on the ground, and minimum height of the olive grove coverage. The sample was divided into three groups with 12 plots based on their slope (< 20, 20 to 30 and 30 to 40 %). Each slope group was subdivided into 3 plots based on the cultivation system (conventional, ecological with tillage, and ecological with no-tillage). The slope gradient was measured with a manual inclinometer and the distance from the soil to the tree cover was measured with a laser distance meter. The type of soil management in each farm was identified through an interview with the farmers.

The soil erosion and erodibility factors were analyzed through bivariate correlation (Pearson correlation coefficient, r) (expression 1), multiple regression analysis (stepwise method) (expression 2) and ANDEVA through Mann-Whitney (expression 3) and Kruskal-Wallis (expression 4). The cultivation system is a qualitative and categorical variable of three classes, and was transformed into quantitative through Dummy variables. Thus, it was substituted with two indicating variables, with dichotomous response: presence (1) or absence of tillage (0) and presence of superficial plant coverage (1) or absence (0). The statistical analysis was carried out with the SPSS software, version 22.0.

where: S _xy is the covariance between variables x, y; S _x is the standard deviation of variable x; and S _y is the standard deviation of variable y.

where: Y is the dependent variable (erosion); X ₁ , X ₂ …X _k are the independent or explicative variables of erosion; b ₁ , b ₂ … b _k are the magnitude of the effect of independent variables and b₀ is the constant value of the model.

where: the means of two independent samples (sample X of size n ₁ and sample Y of size n ₂ ) are compared. Since there are n ₁ ·n ₂ pairs (x _i , y _j ), the number of pairs (x,y) such that x < y will be n ₁ ·n ₂ /2. The Mann-Whitney U statistic is the number of pairs with this property. A significant deviation of U with regard to n ₁ ·n ₂ /2 indicates that the homogeneity of the variances is rejected.

where: the means of k independent samples of sizes n ₁ , n ₂ …n _k are compared, the n observations are ordered and intervals are assigned from 1 to n to the set; Ri is the sum of intervals assigned to the ni observations of sample k.

If the variances of the samples are homogenous, it is expected that the average range is approximately equal for k samples, but when these averages are different the homogeneity of variances is rejected.

Ground cover

The sampling plots presented a dependent variable of the cultivation system. The coverage was lower in the plots with conventional cultivation system and was higher than 20 % in the ones cultivated with ecological system, and in the ecological one with tillage and no-tillage (Table 1). The quality of the table olive depends on the caliber of the fruit, which depends on the pruning and fertilization. Pruning in the conventional system is aggressive and allows a larger size of the fruit and lower coverage of the olive grove. Inorganic fertilization with products of fast assimilation tends to ensure the adequate caliber of the fruit. Pruning in the ecological system is accompanied primarily by organic fertilization, of slower assimilation, which decreases production. The lower caliber and productivity of ecological olive groves are compensated by European subventions and higher value of ecological products.

The height of the olive canopy showed a maximum of 1.39 ± 0.74 m in the ecological plots with tillage, without any apparent relationship to the cultivation system. The ground cover, which included weeds and litter, was noticeably variable. The lowest was seen in the conventional system, the ecological with tillage doubled it, and the ecological with no-tillage presented the highest coverage and lowest data dispersion. The different techniques from each cultivation system allow explaining these differences. The conventional system uses deep tillage that mixes the cover with the soil and considerably reduces the remainders on the surface. On the contrary, the ecological system bases the control of weeds on the combination of shallow tillage and mechanical weeding (ecological system with tillage), or exclusively on this last action (ecological system with no-tillage).

Erosion status of the soil and the factors that influence erosion

Olive groves with conventional and ecological productive system with tillage presented the highest erosion rates. On the contrary, erosion in the ecological system with no-tillage was around a fourth of these (Table 2). The soil showed mainly sheet erosion and secondarily by splash. The erosion favored by the mechanical alteration of the soil with tools (plow) (Hi and Hl) was higher than the conventional and ecological systems with tillage, and the ecological with no-tillage was virtually not affected by the use of tools, and only slightly by the transit of people.

The erosion in rills and gullies was not important, was absent in plots with ecological system with no-tillage and was minimal, with great dispersal, in the conventional and ecological cultivation system with tillage (Table 3).

Erosion, in percentage of surface affected, presented a significant relationship with the cultivation system, which was analyzed through the dummy variables: tillage (r = 0.91; p ≤ 0.01) and coverage (r = -0.46; p ≤ 0.01), surface of soil exposed (r = 0.62; p ≤ 0.01) and superficial plant coverage (r = -0.54; p ≤ 0.01). The correlation coefficients increased in all the cases with surface not affected by erosion: dummy tillage (r = -0.92; p ≤ 0.01), dummy coverage (r = 0.62; p ≤ 0.01), surface of soil exposed (r = -0.73; p ≤ 0.01) and ground cover (r = 0.70; p ≤ 0.01); and if the rock cover on the soil is incorporated (r = -0.40; p ≤ 0.05). Erosion, measured in soil volume lost, did not present any significant relationship to the variables introduced in the analysis.

The fact that the slope gradient did not present statistical relationship with the erosion stood out, which could be due to the fact that effects from other factors, such as soil coverage and management with plowing, stand out in erosion. Thus, the presence of plant coverage on the soil reduces erosion and seems to minimize the effect of the slope. Similarly, the action from plowing eliminated the live plant cover and plant remainders, and altered and homogenized the soil structure, which is why its effect on erosion was not statistically significant. These results are similar to those found by ^{Blanco and Aguilar (2015} and ²⁰¹⁶⁾ with other crops.

The olive canopy and its minimum height also did not present a statistically significant relationship with erosion. In this regard, ^{Nanko et al. (2008)} showed that the tree canopy does not reduce erosion and may even favor it. The (indirect) rain runoff through leaves and stems of the tree canopy may have more kinetic energy than direct rainfall, because the canopy can generate larger drops. The impact of indirect rainfall on the soil without plant cover can break its superficial structure and cause a reduction in the infiltration rate and an increase in the superficial runoff and erosion. The regression coefficients confirmed the influence of these factors on soil erosion.

The regression analysis with the surface affected by erosion as dependent variable, included only the tillage dummy variable in the model as predicting, with a coefficient of determination (R²) of 0.82; thus, 82 % of the variability of erosion is explained by the variable indicated (equation 1). The regression analysis with the surface not affected by erosion as a dependent variable generated two models. The first was simple, also with the tillage dummy variable as predicting, with a R² of 0.85 (equation 2). The second model was multiple and included, in addition to the prior variable, the soil surface exposed, with a R² of 0.90 (equation 3).

Equations 2 and 3 improve in three and eight points, respectively, the coefficient of determination of the first, which is why the latter predicted erosion more accurately.

The influence of soil management on water erosion

The Mann-Whitney variance test between the surface not affected by erosion and the management of soil (tillage/no-tillage) confirmed that there are significant differences in the confidence interval of 99 % (U = 1.00; p ≤ 0.01). These results, expressed as cultivation systems, indicate that the soils with conventional and ecological cultivation with tillage presented the lowest surface not affected by erosion (7.15 ± 8.68 %), compared to the soils with ecological cultivation system with no-tillage (52.64 ± 9.82 %) (Figure 4).

Figure 4 Soil not affected by erosion in olive crops with different soil management (mean ± standard deviation). 

Soil management tends to be one of the most important factors to explain erosive processes. Tillage breaks the original structure of the soil which when disaggregating is more vulnerable to erosion, because it is less resistant to the impact of rain and dragging from runoff. This is one of the most important causes that explain the higher rates of erosion in cultivation systems with conventional management with plowing, as compared to management with no-tillage (^{Evans, 2006}). In this regard, ^{Zhang et al. (2007)} highlighted the importance of conserving the soil structure to improve the structural stability and resistance to erosion. ^{Arshad et al. (1999)} explained the lower erosion in the agricultural systems where tillage was not practiced, from the conservation of the structure and the influence that it exerts on the hydrologic properties of the soil.

Influence of the soil surface exposed on water erosion

The Kruskall-Wallis variance test between the surface not affected by erosion and that with soil exposed, grouped in the intervals 0 to 10, 10 to 40, and 40 to 70 %, confirmed the significant differences (Chi-squared = 21.07; p ≤ 0.01). The soils with more than 40 % surface exposed showed a lower surface not affected by erosion, followed with close values by soils with 10 to 40 % of surface exposed, and the surface not eroded rose 40 % in soils with surface exposed lower than 10 % (Figure 5).

Figure 5 Soil not affected by erosion in olive crops with different intervals of surface with soil exposed (mean ± standard deviation). 

The importance of the ground cover to reduce erosion was confirmed, since it protected against the impact of rainfall and superficial runoff. ^{Hudson and Jackson (1959)} and ^{Zanchi (1983)} pointed out that the stability of the aggregates in face of an episode of rains is lower when the cover is scarce, because the electrical forces of calcium and magnesium are weakened with the contact of the soil with water. On the contrary, in the soil with herbaceous cover the consistency of the aggregates is higher when they are saturated. The factors that depend on the surface vegetation, the roots of the plants, the activity of organisms linked to it (earthworms, fungi and others), and the decomposition of plant remainders improve the cohesion of soil particles between each other, the aggregates will be stable in water and will be maintained with humidity processes (^{Sullivan, 2008}).

Ground cover reduces the risk of deterioration of the superficial structure, favors infiltration, increases the capacity for water retention, and reduces runoff and soil erosion. The results from our study agree with those obtained by ^{Leys et al. (2010)}, who indicated that the ground cover of the soil was the most important variable to explain runoff and erosion in the farms.

Assessment of the study method

The analysis method of soil water erosion applied in this study was considered appropriate to quantify the erosion status of the soil and to establish the factors of erodibility in olive crops. The results confirmed the possibility of using these factors to establish the GAEC, which require the rules of cross-compliance of the CAP to the particular conditions of the application areas.

The environmental and management differences between farms, in addition to the temporal and spatial variation of erosion in a zone, impede extrapolating the results from isolated studies to identify the environmental or human causes of erosion. The low cost and ease of application are important qualities of the method that we used to analyze 36 plots with three types of management and can favor the monitoring that the CAP regulations require.

The results of our study in Mediterranean mountains are added to those of tropical environments obtained by ^{Blanco and Aguilar (2015}, ²⁰¹⁶⁾.