Sustainable Intensification of the Montado Ecosystem: Evaluation of Sheep Stocking Methods and Dolomitic Limestone Application

Abstract

The objective of this study was to determine how application of dolomitic limestone and stocking methods (continuous stocking or deferred stocking) affect the soil compaction, sheep grazing location, height, and nutritional value of pastures when the pasture growth rate is at its maximum. A 4 ha field at Mitra farm—University of Évora—was divided into four plots: P1 and P2—without application of dolomitic limestone, continuous stocking (CS), and deferred stocking (DS), respectively—and P3 and P4—with application of dolomitic limestone, DS (2.3 AUE), and CS (1 AUE), respectively. In DS, animals were placed and removed from the plots depending on the height of the pasture (entry ≥ 10 cm; removal ≤ 5 cm). Throughout the pasture’s vegetative cycle, several measurements of pasture height and cut were carried out. From the beginning of March to the beginning of June, animal behavior was observed (animals’ activity grazing and location) by trained observers through binoculars on six dates. The results show the following: (i) the application of dolomitic limestone combined with CS provided higher values of pasture height; (ii) there were no significant differences in pasture quality between treatments; (iii) DS led to 50% more sheep grazing days that CS; (iv) there were no significant differences in soil compaction between CS and DS; and (v) the stocking methods and the application of dolomitic limestone did not seem to change the grazing pattern between treatments. This study constitutes a basis to support more informed decisions by agricultural managers and may also contribute to maintaining balance in the Montado ecosystem, as well as increasing the efficiency of livestock production systems based on rainfed pastures.

1. Introduction

Montado (or Dehesa in Spain) is an agro-silvo-pastoral ecosystem, characteristic of the Alentejo region in Portugal. This ecosystem is associated with various agricultural activities with environmental and social value, thus considered of high natural value [1]. Montado is considered a complex ecosystem due to the interrelations between its fundamental components—climate, soil, pasture, trees, and animals [2]. Montado is influenced by the Mediterranean climate, with rainy and mild winters and hot and dry summers with great variability and seasonality [3]. During the dry summer, temperatures can reach 40 °C, and minimum temperatures can drop below 0 °C during winter [3]. In this climate, severe droughts can often occur for long periods of time. The soils of this region are classified mainly as Cambisols, derived from granite [4]. These soils usually are degraded due to erosion and loss of nutrients, have fertility limitations, and are shallow, stony, acidic, poor in nutrients, and with micronutrient imbalances, namely the magnesium (Mg)/manganese (Mn) ratio, which cause Mn toxicity [5]. As a way of mitigating Mn toxicity and the effect of acidity, dolomitic limestone can be applied [6], together with phosphorus fertilization [7]. The typical pastures of the Montado, which tend to be of poor quality (low nutritive value and degraded floristic composition) and associated with low productivity [8], are greatly affected by the amount and the distribution of rain throughout the year and by the combination of temperatures and rain.

Grazing is a key issue for pasture management and nature conservation [9,10]. Plants, floristic composition, and biodiversity depend on the animal species, type, and grazing management (grazing intensity and frequency) [10,11]. In the Montado, some farmers practice the continuous stocking (CS) method on their farms, while others choose the deferred stocking (DS) method, with no common pattern. DS involves the use of pasture on a given plot, in longer or shorter periods of grazing, depending on the pasture biomass. In this type of grazing, the stocking rate is much higher than that of CS. In large plots, in the Montado, CS tends to prevail [1], with low stoking rates, allowing more selectivity and resulting in higher heterogeneity of pasture consumption, where over- and under-grazed areas can occur simultaneously [12]. In a study carried out by Carreira et al. [13], grazing days were compared between four study treatments (1—without application of dolomitic limestone and CS; 2—without application of dolomitic limestone and DS; 3—with application of dolomitic limestone and DS; 4—with application of dolomitic limestone and CS). In the plots subject to DS, there were approximately 80 fewer days of grazing (pasture recovery days) than in those subject to CS. In addition, the application of dolomitic limestone provided 7 more days of grazing than non-application in plots subject to DS.

Traditionally, pastures are generally managed with relatively low grazing pressures, in both CS and DS methods, allowing animals to choose their diet [14]. Selective grazing, stocking rate, and grazing seasons influence the range and communities of plant species [15]. Teague et al. [16] reported that DS provides satisfactory results in productive, ecological, and economic aspects. DS can contribute to an increase in the coverage percentage of legumes in the pasture, improving the floristic composition and reducing the number of unwanted species of low nutritive value [17]. Heavy grazing can lead to soil degradation (by exaggerated trampling) and loss of biodiversity, while underutilization can lead to a greater preponderance of species of weeds with lower nutritive value and loss of habitat, overlapping a shrub layer [18]. However, Barriga [19], working with cattle, notes that CS has various advantages over DS, such as the return of nutrients to the soil through urine and feces. In addition, a decrease in shrubs and increase in forage species with good nutritive value and an improvement in animal performance because they can select their diet are some advantages of CS compared to DS [20,21,22]. The knowledge of the nutritive value of pastures and their availability, throughout the grazing seasons, can lead to improvements in production and management systems for grazing ruminants [23]. In addition to the nutritive value, the height of the pasture also influences the intake, selectivity, and performances of grazing animals [19].

The improvement of feed efficiency of grazing animals is the major goal of livestock producers [24]. Ecosystems, such as the Montado, where animals have grazed for many centuries achieve their biological characteristics and balance through interactions between livestock and vegetation [25]. The behavior of ruminants, in grazing, is a function of the characteristics of the plant communities and grazing management decisions [26]. Therefore, the choice of grazing areas is also related to the physical and thermal characteristics of the plot since the animals prioritize their primary physiological needs—water consumption and thermal regulation [27]. Riedel et al. [25] state that the way in which grazing occurs, in each area, depends on the physical environment of that same area, which includes the productivity and quality of pastures, accessibility to certain areas, and the availability of water—very important in the Mediterranean regions mainly in late spring and summer. The season of the year affects the floristic composition of the pasture and its nutritive value, which in turn affect the behavior of grazing animals [15]. In small patches and when the stocking rate is high and there are plant species of high nutritive value, it can lead to animals always being closer to these areas [28].

Knowledge of animal preferences in pastures and their selectivity is crucial to a better understand of the relationship between animals, grazing, and pastures [29]. The use of pastures requires the animals to adapt to their diversity, thus being based on preferences for grazing locations, depending on variety and availability of vegetation [27]. The sheep do not graze continuously, but rather their grazing is interspersed with periods of rumination and rest [24]. The sheep could adjust their behavior on the pasture to maintain group cohesion if the space provided for each animal is greater than 200 m² [28]. The choice of grazing areas on a plot is related to the animal’s ability to select suitable diets, considering the height of the species, the phenological state of the plants, their nutritive value, and the floristic composition of the pasture [27]. Weather conditions also influence grazing behavior. High temperatures, such as those found in the summer in the Alentejo, tend to reduce the amount of daily time spent on grazing and rumination [30]. In sheep, ingestion periods tend to be shorter, and periods of rumination and leisure time are longer and occur during the hottest part of the day. In regions with a hot and dry summer, most grazing periods occur in the early morning, late afternoon, and night [27].

Our research team has carried out several studies to monitor the effect of dolomitic limestone application on soil, trees, pastures, and sheep grazing interactions over time. In designing this study, the hypothesis was raised as to what the effect of dolomitic limestone and different types of grazing (translated into pasture height and its nutritional value) would be on grazing preferences in the spring.

Therefore, the study aims to determine the effect of applying dolomitic limestone and continuous stocking or deferred stocking on the height and nutritional value of the pasture, on the soil compaction, and on the grazing location and preference when the pasture presents the highest growth rate.

2. Materials and Methods

2.1. Study Area

This study is integrated in a long-term project to monitor the Montado ecosystem, which started in 2015 (Figure 1), being the culmination of all interventions in the study area since that date and their influence on the animals’ food preferences. Figure 1 shows the chronological scheme for the whole study, resulting in several publications [5,13,31,32,33,34,35].

The predominant soils of this region are classified as Cambisol, derived from granite [6]. The study area is in an area of Montado with holm oak (Quercus rotundifolia Lam.), and dryland pastures grazed by beef cattle and sheep.

The Alentejo is affected by the Mediterranean climate, characterized by hot and dry summers and wet and cold winters [3]. The irregular rain distribution and total year precipitation variation are also characteristic of the Mediterranean climate. Most of this precipitation occurs in autumn, winter, and spring. In summer, the precipitation will always have residual values [33].

The weather conditions of the 2020/2021 agricultural year are presented in Figure 2, which shows the temperature and rain graph for the Mitra meteorological station (Évora) between September 2020 and June 2021. The greatest amount of precipitation occurred in September and February (537 mm), while only 91 mm of precipitation was recorded during spring and early summer (when temperatures were more favorable for pasture growth and extension of its vegetative cycle).

The study began in November and ended in June, with two phases. In the first phase, from November to February, evaluations were carried out on pastures and measurements of grazing days in the different plots. The second phase began in March with the evaluation of the sheep’s grazing behavior, which lasted until late June. The results of height and nutritional value were obtained from November 2020 to June 2021.

This study took place in a 4 ha experimental area (38°32.2′ N; 8°1.1′ W), called ECO-SPAA, located at the Mitra experimental farm, University of Évora—Portugal. The 4 ha experimental area was divided into four plots with 1 ha each, corresponding to the following treatments: P1UC—without application of dolomitic limestone and continuous stocking (1 animal unit equivalent (AUE)/ha); P2UD—without application of dolomitic limestone and deferred stocking (2.3 AUE/ha); P3TD—with application of dolomitic limestone and deferred stocking (2.3 AUE/ha); P4TC—with application of dolomitic limestone and continuous stocking (1 AUE/ha) (Figure 3).

To represent the different existing plant communities, 48 sampling points were georeferenced, 12 in each of the four plots of the study (Figure 3). These 48 representative points were identified by a botanical specialist and permanently marked with a numbered flag and identified with a number (1 to 48). These points represent the plant communities identified previously, with species that vary in diversity and occurrence. The characterization of the floristic composition was carried out in January (winter), May (peak of spring), and June (summer). This characterization involved the identification of different plant species on each date in an area of 1 m².

The characterization of the surface layer of the soil (0–0.30 m depth), carried out in October 2015, revealed acidic pH (average value of 5.4 ± 0.3), so two applications with dolomitic limestone were carried out (2 tons/ha of dolomitic limestone, per application) in half the area (P3TD and P4TC) in November 2017 and June 2019. The application of dolomitic limestone was carried out with a centrifugal distributor attached to a tractor. In December 2018, the whole study area (P1UC, P2UD, P3TD, and P4TC) received 100 kg/ha of binary fertilizer (18-46-0). The experimental design was based on a factorial, with two plots subjected to application of dolomitic limestone and two others serving as the control (U treatments). Within each treatment with and without amendment with dolomitic limestone, two stocking methods were applied: CS with continuous stocking and moderate stocking rate and DS with deferred stocking and high stocking rate (2.3 times that applied in the CS). The pasture in the study area is natural, biodiverse, and dryland, consisting mainly of grasses, legumes, and composites [13], in the four plots under study.

2.2. Grazing Management, Pasture Measurements, and Sampling

The grazing experiment was carried out with non-pregnant and non-lactating adult white Merino and black Merino ewes. The ewes were always the same in each treatment throughout the experimental trial. All the ewes had similar body conditions at the beginning and the end of the trial. Every month, all the animals were evaluated in terms of their body condition score (BCS) to highlight possible weight loss or heterogeneities between the animals’ body conditions in the different plots. All animals had a mean BCS of 3.5, with a standard deviation less than 0.5. The scale used is from 1 to 5, where 1 is very thin and 5 is obese [36]. Changes in body conditions score are related to changes in animal weight [37]. The animals always had clean water and mineral supply at their disposal.

Different stocking rates were calculated for the treatments. The stocking density calculation was based on the animal unit for sheep, which was based on adult animals. Thus, the stocking rate was calculated based on the animal unit equivalent (AUE) for sheep (0.2 AUE). The AUE helps estimate the potential forage demand for different kinds of animals based on the standard animal unit and considers physiological differences. To estimate the stocking rate, the number of animals in the area was considered (AUE/ha), corresponding to 7, 16, 16, and 7 animals in the plots P1UC, P2UC, P3TD, and P4TC, respectively. In the case of deferred stocking, a correction factor was introduced corresponding to the percentage of days the animals remained in the respective fences during the grazing season period (AUE/ha*.days). These elements were then used for statistical calculations and corrections.

In CS, animals remained on the plots throughout the entire vegetative cycle of the pasture. In plots with the DS methods, the presence or absence of animals was linked to pasture height following the “put and take” method [38,39]. Grazing management criterion was a function of the average pasture height in each plot; the animals remained on the pasture until its height dropped to values close to 50 mm. Whenever this value was reached, the animals were removed and only returned when the average pasture height was at least 100 mm. Pasture height was measured with an electronic caliper. Pastures above 250 mm were measured using a ruler with a scale in mm, to which a horizontal plastic stick was attached to simplify measurements.

2.3. Pasture Height and Quality Estimations

Pasture heights were measured in the treatments of the study before and after each grazing period. Pasture samples were also collected to estimate the quality (crude protein, CP, and neutral detergent fiber, NDF).

The following procedures and respective analyses were carried out to characterize the pastures in the four treatments. Pasture sampling (for nutritive value) was conducted on nine dates grouped into four periods. Period 1 corresponds to autumn (3 November and 11 December); period 2 corresponds to winter (10 February and 12 March); period 3 corresponds to the beginning and peak of spring (10 April, 24 April, and 8 May); and period 4 corresponds to late spring and early summer (22 May and 5 June). In periods 3 and 4, the samples were taken concomitantly with animal behavior observations. On each date, the pasture height was taken at each sample point (in 3 representative locations of the point, with 3 measurements for each location), followed by pasture cutting in a known area (frame of “0.40 m × 0.25 m”). All the 12 samples in each treatment were mixed, thus making a total of 4 composite samples (one for each treatment). During the animal behavior observation phase, these pasture procedures were carried out the day before each behavior observation. The pastures’ samples were sent to the Animal Nutrition and Metabolism Laboratory—MED Mediterranean Institute for Agriculture, Environment and Development, where they were placed in an oven at 65 °C for 72 h to be dehydrated. Next, the dehydrated samples were weighed and ground in a Perten instrument mill equipped with a 1 mm sieve, for subsequent determination values of CP and NDF. Both were expressed as percentages on a dry weight basis of the samples. For CP and NDF, conventional wet chemistry methods according to the AOAC were used [40]: (i) nitrogen content was analyzed with the Kjeldahl method, a colorimetric determination in a Bran + Luebbe autoanalyzer with a factor of conversion to CP of 6.25 (method no. G-188-97 Rev 2, Bran + Luebbe, Analyzer division, Norderstedt, Germany); (ii) the NDF content was analyzed according to the Goering and Van Soest (1970) method in a fibered digester (Foss Tecator AB, Box 70 SE-263 21 Höganäs, Sweden).

The elevation of the experimental field is represented on the altimetric map in Figure 4.

2.4. Cone Index Measurements

To evaluate soil compaction due to animal trampling in CS and DS, soil resistance to penetration (Cone Index, CI, in kPa) was measured with an electronic cone penetrometer “FieldScout SC 900” (Spectrum Technologies, Aurora, IL, USA) equipped with an ultrasonic depth sensor. This assessment was carried out in October 2021. Similar to Serrano et al. [35], for each sampling point (1 m² area), 5 measurements were taken (one in the center and one at each of the vertices of the quadrant) with the CI, at a depth of 0 to 30 cm. Afterwards, the mean value of the five measurements was calculated for each of the 48 sampling points at depths of 0–15 and 15–30 cm. To avoid variability in soil moisture (which could affect penetration resistance measurements), the same methodology described by Serrano et al. was used [35]: (i) All measurements at the 48 sampling points, with CI, were carried out on the same day. When measuring resistance to soil penetration, soil samples were taken at the central point of each sampling point, using a gouge auger and a hammer. (ii) Then the soil samples were weighed and placed in an oven at 70 °C for 48 h. (iii) After dehydration, the samples were weighed again to establish the soil moisture content (SMC).

2.5. Observation of Sheep’s Grazing Behavior and Spot Preferences

Throughout the period of observation of sheep grazing behavior (13 March to 7 June) to identify their favorite grazing spots, observations were made for 12 days. Considering that the stocking methods were different among treatments during autumn and winter, with potentially different levels of selectivity, it is essential to understand whether the disposition of the animals’ favorite spots in the pasture differs in the spring when the maximum growth rate occurs. Sheep behavior was observed approximately every 15 days. Each date observation corresponds to two repetitions on two consecutive days:

• Date 1—13 March (start—8 a.m.; end—7 p.m.).
• Date 2—28 March (start—7 a.m.; end—7 p.m.).
• Date 3—25 April (start—7 a.m.; end—8 p.m.).
• Date 4—9 May (start—7 a.m.; end—8 p.m.).
• Date 5—23 May (start—7 a.m.; end—8:30 p.m.).
• Date 6—6 June (start—7 a.m.; end—8:30 p.m.).

Four trained observers carried out behavioral observations simultaneously (one observer for each treatment). Observations using binoculars were carried out every 10 min, from sunrise to sunset, which represents about 12 h per day. The observers were placed far enough from the animals so as not to interfere with their natural behavior [15,30]. At each moment of observation, for each of the 4 plots (P1UC, P2UD, P3TD, P4TC), the location of the animals was referenced to the 12 sampling points. Thus, for each hour of observation (6 observation moments), a map like the one in Figure 3 was used for each plot, where the locations of the animals were recorded. Combining with the local site, individual grazing activity was registered.

2.6. Statistical Analysis of Data

2.6.1. Pasture Height and Quality

Considering that in P3TD, in December, the animals remained grazing for 7 more days than in P2UD, height measurement in P3TD was carried out once more than in the other plots. On the day the sheep left P3TD (December), height measurements were taken only in this plot, which meant that the number of measurements between P3TD and the remaining plots was different. For the statistical analysis of pasture height and nutritive value, two different approaches were used: one for pasture height and a different approach for CP and NDF.

(1) For pasture height, a “Generalized Mixed Additive Model” (GMAM) was used with a gamma response. A logarithmic link function was adjusted for pasture height. Treatments were entered into the model as fixed parametric factors, and their effect was controlled for mean temperature (as a smooth variable modeled with a thin plate regression spline), cumulative precipitation (as a smooth variable modeled with a thin plate regression spline), and the number of animals per grazing day (animals/(grazing days + 1))—as smooth variable modeled with a thin plate regression spline). Plot/spot pairs were entered as a random factor, as this is where repeated measurements occur.

This model was better than the corresponding Generalized Linear Mixed Model (GLMM), also with a gamma response and logarithmic link function, as it presented lower AIC and BIC values, a better fit to the data, and a better explanatory capacity (it can explain 65.1% of total deviance).

The assumptions of independence, homoscedasticity, and normality of the residuals were verified, as well as the normality of random effects. Multiple comparison tests adjusting for Tukey were performed.

(2) For CP and NDF, it was not possible to use a GLM model, as the residuals are correlated. The best approach was the use of a GLMM (Generalized Linear Mixed Model) since the random effect was the treatment itself, which made the main objective, of comparing the 4 treatment levels as a fixed effect, unfeasible. Therefore, a comparison was made between the levels of treatments 2 by 2, using the t-test for paired samples, adjusting the p values with the “Holm” correction. To evaluate the effect of the period, we considered mixed linear models (GLMM), adjusted by restricted maximum likelihood (REML) with the plot as a random effect, adjusting for the mean temperature (transformed by logarithmic), animals per grazing day (logarithmic transformation of the quotient between the number of animals and the number of grazing days plus 0.5, since there are cases with no animals), and the accumulated precipitation (also in the logarithmic transformation corrected plus 0.5, as there are cases in which there was no precipitation). All models fit the data well and satisfied the assumptions of homoscedasticity and normality of the residuals, as well as the normality of the random effects.

This statistical analysis was performed using R 4.4.0. software.

2.6.2. Cone Index

An analysis of variance (ANOVA) was carried out between the types of grazing (CS and DS) and between CI depths (0–15 and 15–30 cm). These analyses were performed using the IBM SPSS Statistics package for Windows (version 28.0, IBM Corp., Armonk, NY, USA). Tukey’s HSD test was also performed to compare the means.

The maps of soil variables (SMC and CI) and the altimetric map were carried out through geostatistical analyses with the “Geostatistical Analyst” extension of ArcGIS software (version 10.5, ESRI, Inc., Redlands, CA, USA). In all cases, omnidirectional experimental variograms were computed and spherical theoretical variograms were fitted to their points. The variogram ranges were between 73.8 m and 89.6 m, the sills were between 0.96 and 0.75, and the nugget effects were as low as possible, with values lower than 0.1.

2.6.3. Animal’s Location Preferences

The behavior analysis was carried out by observation date, based on animal presence or absence near each sampling point and in each plot. For this purpose, cross-tabulations of the animals’ permanence at the sampled points in each plot were created based on the observations every 10 min on each observed day. Statistical analyses of data on animal locations on the pasture were carried out using IBM SPSS 25.

With the aim of visualizing the animal distributions throughout the experimental field and during different dates, animal presence was estimated at any un-sampled location. As a continuous variable was necessary, an area of 1 m² was associated with any location, and, in consequence, animal density was mapped. Kriged maps showing the spatial distributions of animals in each date were generated.

Using geostatistical techniques, in this case ordinary kriging, estimated values were obtained for all unsampled locations based on the point measurements distributed throughout the experimental field. This allowed visualization of the spatial patterns of the variables considered, and finally, raster maps were obtained with a spatial resolution of 1 m².

That resolution was logically selected and introduced as an input parameter in ArcGIS when converting from vector (point) to raster format.

The ArcGIS software (version 10.5, ESRI Inc., Redlands, CA, USA) was used to model the spatial variation of grazing. Interpolation analyses were performed with the ordinary kriging algorithm using the Geostatistical Analyst extension in ArcGIS. In all cases, omnidirectional experimental variograms were computed, and spherical theoretical variograms were fitted to their points. The variogram ranges were between 73.8 m and 89.6 m, the sills were between 0.96 and 0.75, and the nugget effects were as low as possible, with values lower than 0.1.

3. Results

3.1. Grazing Days, Stocking Rate, and Pasture Height

The sheep grazing days were calculated for each treatment and are shown in Figure 5. Although the grazing days in the P2UD (151 days) and P3TD (158 days) plots were lower than in the remaining plots (236 days), when we multiply these days by the number of animals on each day and in each plot, we find that the sheep grazing days in the plots subject to DS are much higher (around 50%).

The distribution of pasture height values by date and plot, throughout the pasture’s vegetative cycle (autumn, winter, spring, and early summer), is presented in Figure 6. Generally, as expected, the treatments associated with CS had higher pasture heights; on some dates, pasture heights were higher. Within the CS treatments (P1UC and P4TC), the average heights were higher in the P4TC treatment on most dates. However, when we model the data with the correction for the number of grazing days and the number of animals, the differences between the treatments are not significant, mainly due to the dispersion of the pasture height within plots (Figure 6).

At the beginning of April, the pasture’s growth rate exceeded the sheep’s intake rate in all plots, which was reflected in a significant increase in pasture height in both treatments with a moderate animal load (P1UC and P4TC) as well as the treatments with a higher animal load (P2UD and P3TD).

All plots reached maximum pasture height in May and June, with some values exceeding 550 mm (Figure 6). It should be noted that in all plots, many pasture areas reached heights of over 500 mm, while other areas remained close to 70 mm, indicating a reduced growth rate or greater preference by the animals (Figure 6).

The multiple comparison tests adjusting for Tukey allow us to conclude that the P4TC treatment leads to a higher average height than P3TD and P2UD (p value < 0.05), with the average height obtained with the P1UC treatment also being significantly higher than that obtained with the P3TD and P2UD (p value < 0.05) treatments (

Table 1. Results for the height, in log scale of multiple comparison test, adjusting for Tukey.

	Estimate	SE	p Value
P4TC-P3TD	0.4230	0.120	0.0025
P4TC-P1UC	0.0882	0.117	0.8739
P4TC-P2UD	0.4437	0.120	0.0013
P3TD-P1UC	−0.3348	0.121	0.0300
P3TD-P2UD	0.0207	0.115	0.9980
P1UC-P2UD	0.3555	0.121	0.0182

SE—standard error.

).

3.2. Characterization of the Nutritive Value

The average percentage of CP of the pasture in each treatment throughout the animal observation period is presented in Figure 7a. At the end of winter (period 2) and spring (period 3), the highest percentage of CP occurred in the P3TD plot (20.1% and 13.4%, respectively). The highest percentage of CP occurred at the end of winter, with 15.9%, 18.8%, 20%, and 18.9%, for the plots P1UC, P2UD, P3TD, and P4TC, respectively. On the other hand, the lowest values were observed at the beginning of summer (Period 4), with values of 7.3%, 7.2%, 6.2%, and 6.1%, for the plots P1UC, P2UD, P3TD, and P4TC, respectively. The percentage of NDF in each treatment throughout the animal observation period is presented in Figure 7b. As can be seen, the beginning of summer had higher NDF values (62.2%, 63.4%, 66.0%, and 65.9% for P1UC, P2UD, P3TD, and P4TC, respectively), while the end of winter had the lowest values (46.7%, 47.4%, 42.4%, and 44.7% for P1UC, P2UD, P3TD, and P4TC, respectively). At the beginning of summer, the P1UC treatment showed the lowest value (62.2%).

The CP distribution of pastures for each treatment is presented in Figure 8a. The distribution of pasture NDF values for each treatment is presented in Figure 8b.

There were no significant differences between the treatments for CP and NDF (all p values greater than 0.05) (

Table 2. Means and standard deviations by treatment and adjusted p values of the paired-samples t-test for CP and NDF.

CP					NDF
	M ± SD	P2UD	P4TC	P3TD		M ± SD	P2UD	P4TC	P3TD
P1UC	13.8 ± 5.3	1.000	0.952	0.304	P1UC	53.7 ± 9.0	1.000	1.000	0.806
P2UD	14.4 ± 5.7		0.862	0.916	P2UD	55.7 ± 7.2		0.871	1.000
P4TC	13.8 ± 6.1			0.147	P4TC	54.4 ± 10.0			1.000
P3TD	14.9 ± 6.4				P3TD	53.3 ± 9.0

M—means; SD—standard deviation; CP—crude protein; NDF—neutral detergent fiber.

). Through Tukey’s multiple comparison test for the period, it was concluded that period 3 leads to mean CP and NDF values significantly higher than those of the remaining periods (

Table 3. Tukey’s multiple comparison test for period for CP and NDF.

	CP				NDF
	Dif	LCI (95%)	UCI (95%)	p	Dif	LCI (95%)	UCI (95%)	p
P2UD-P1UC	7.9	2.1	13.7	0.003	−5.3	−11.6	0.9	0.128
P3TD-P1UC	−4.9	−7.6	−2.1	<0.001	6.0	0.3	11.7	0.034
P4TC-P1UC	−7.1	−11.6	−2.5	<0.001	14.0	7.8	20.3	<0.001
P3TD-P2UD	−12.8	−17.9	−7.8	<0.001	11.3	5.6	17.0	<0.001
P4TC-P2UD	−15.0	−20.1	−9.9	<0.001	19.4	13.1	25.6	<0.001
P4TC-P3TD	−2.2	−5.3	0.9	0.257	8.0	2.3	13.7	0.002

Dif—difference; LCI—lower confidence intervale; UCI—upper confidence interval; p—p value; CP—crude protein; NDF—neutral detergent fiber.

). Through this test, it was also concluded that (i) for CP, there are no significant differences between periods 3 and 4, and for NDF, there are no significant differences between periods 1 and 2 (Table 3); (ii) period 2 leads to mean values significantly higher than those of the remaining periods; (iii) period 4 leads to mean values significantly higher than those of the remaining periods.

3.3. Preferred Grazing Locations Versus Average Height of Pasture

The preferred grazing areas on the 12 days of observation, based on the 12 georeferenced points in each plot (according to Figure 3), are represented in a sequence of maps (Figure 9, Figure 10 and Figure 11). In these figures, it is possible to observe the evolution of height of pasture per point and preferred grazing areas throughout the observation period (end of winter, spring, and beginning of summer). The maps have up to four graduations, from lightest to darkest, which correspond to the following: no or low preference, medium preference, high preference, and very high preference, respectively. The numbers in the square in the upper right corner of each map of these figures represent the number of observations of animals grazing at each point on each observation date.

In the P1UC plot, sheep stayed mainly in the lower part of the field, where the pasture was not always the highest. The spots most preferred by sheep were 8 and 12, extending to spots 5, 7, 9, and 10. After the third observation date, the space became more evenly frequented, regardless of the height of the pasture, with spots 5 and 11 being the most grazed on the fourth observation date. Points 1 and 2 were the least sought after throughout the observation period.

In the P2UD plot, sheep grazing pattern was similar throughout the observation dates. The sheep preferred spots 17, 18, 19, and 20 throughout most observations. After the fourth observation date, the animals showed more dispersed grazing behavior throughout the available area, except for spot 23, which was less preferred on all observation dates. The animals did not choose spots according to the height of the pasture; other variables, possibly the species and its nutritive value, may have driven the different preferences for the various spots. Spots 13, 14 (except on the last observation date and in warmer weather), 15, 23, and 24 were the least grazed throughout all observations.

In the P3TD plot, a more restricted grazing area is visible. The most preferred spots during the first observation dates were 30, 31, and 32, with a lower frequency for spots 33 and 34. After the third and fourth observation dates, a more extensive range of grazed areas was observed. Sheep did not show a particular preference for areas where the height of the pasture was higher; in many cases, their preference was for lower pastures. Finally, on the last two observation dates, there was a return to the areas that were initially most grazed. The observations showed a clear tendency for spots 25, 26, and 35 to be grazed less.

In the P4TC plot, the sheep’s preferences were somewhat more heterogeneous. On the first observation dates, there was a clear preference for spots 40, 41, 45, 46, and 47. The pattern changed on the third and fourth observation dates, restricting preferences to spots 41 and 42 and adding spot 45 on date 5. Like what was observed in the other treatments, the motivations for choosing the spots were not just related to the height of the pasture. On observation date six, there was more dispersion across the available area, with spots 37, 38, 39, 47, and 48 remaining negligibly preferred.

3.4. Relationship Between Preferred Grazing Locations and Floristic Composition

With information from all observation dates, a map of accumulated grazing preferences was obtained (Figure 12). In P1UC, spots 1, 2, and 3 are very poorly grazed, unlike spot 8, which is the most preferred. In the other plots, it is important to highlight an extensive area with very little daily grazing time, especially spots 13, 23, 25, 26, 35, and 37. It should also be noted that the spots most consistently preferred by the animals were 30, 31, and 42 and, with some relevance, spots 41 and 45.

The botanical species identified in January in the sheep’s favorite grazing spots during the month of March are presented in

Table A1. Botanical species identified (winter) in preferred grazing spots on each plot in March.

Botanical Species	Spot 8	Spot 9	Spot 18	Spot 19	Spot 30	Spot 31	Spot 32	Spot 45	Spot 47
Chamaemelum mixtum	0	0	1	0	0	0	0	0	0
Erodium cicutarium subsp. bipinnatum	0	1	1	1	1	0	1	0	1
Calendula arvensis	0	1	0	0	0	0	0	0	0
Chamaemelum fuscatum	1	0	0	0	0	0	0	0	0
Diplotaxis catholica	1	1	1	0	1	0	1	0	1
Echium plantagineum	1	0	1	1	1	0	1	1	1
Geranium molle	0	1	0	0	0	0	0	0	0
Hypochaeris glabra	0	1	0	0	0	0	0	0	0
Iris xiphium	0	1	0	0	0	0	0	0	0
Leontodon taraxacoides	0	0	1	0	0	0	0	1	1
Ornithopus compressus	1	0	0	0	0	1	0	1	0
Plantago lanceolata	0	0	0	1	0	0	0	0	0
Plantago sp.	0	1	0	0	0	0	0	0	0
Ranunculus ollissiponensis subsp. ollissiponensis	0	0	0	0	0	0	0	1	0
Senecio jacobaea	0	0	0	0	1	0	0	0	0
Senecio vulgaris	1	1	1	1	1	0	1	0	0
Spergula arvensis	1	0	0	0	0	0	0	0	0
Stachys arvensis	0	0	0	0	1	0	1	0	0
Stellaria media	0	0	0	1	0	0	0	0	0
Urtica urens	0	0	0	1	0	0	0	0	0
Genus Trifolium	1	0	0	0	0	1	0	0	0
Several grass species	1	0	0	1	1	0	1	1	1

Presence: 1; absence: 0.

(Appendix A). Echium plantagineum L. and Diplotaxis catholica L. as well as several grass species were present in most of the preferred spots in March. The botanical species identified at the beginning of May in the sheep’s favorite grazing spots in the May observations are presented in

Table A2. Botanical species identified (spring) in preferred grazing spots on each plot in May.

Botanical Species	Spot 3	Spot 4	Spot 5	Spot 8	Spot 10	Spot 11	Spot 12	Spot 17	Spot 20	Spot 30	Spot 31	Spot 41	Spot 47
Agrostis castellana Boiss. & Reut.	0	0	0	0	0	0	0	0	0	0	0	0	1
Agrostis pourretii Willd.	0	1	0	0	0	0	1	0	0	0	0	0	0
Anagallis arvensis L.	0	0	1	0	0	0	0	0	0	0	1	0	0
Anthriscus caucalis M.Bieb.	0	0	1	0	0	0	0	0	0	0	0	0	0
Arum italicum subsp. italicum	0	0	1	0	0	0	0	0	0	0	0	0	0
Avena barbata subsp. lusitanica (Tab. Morais) Romero Zarco	0	0	0	0	0	1	0	0	0	0	0	0	0
Bromus hordeaceus L.	0	0	0	1	1	1	0	0	0	1	1	1	1
Bromus sterilis L.	1	0	1	0	1	0	0	0	0	0	0	0	0
Callitriche stagnalis	0	0	0	0	0	0	1	1	0	0	0	0	0
Carduus tenuiflorus Curtis	1	0	1	0	0	0	0	0	1	0	0	0	0
Cerastium glomeratum Thuill.	1	0	1	0	0	0	0	1	0	1	1	0	0
Chamaemelum fuscatum (Brot.) Vasc.	0	1	0	0	0	0	1	1	0	0	0	0	0
Chamaemelum mixtum	0	1	0	1	1	1	0	1	1	1	1	1	1
Crepis capillaris (L.) Wallr.	0	0	0	1	1	0	0	0	0	0	1	0	0
Crepis vesicaria subsp. taraxacifolia (Thuill.) Thell.	1	1	0	0	1	1	0	0	0	0	1	0	0
Cynodon dactylon (L.) Pers.	0	0	0	0	0	0	1	1	0	1	0	0	0
Diplotaxis catholica (L.) DC.	1	1	0	1	0	1	0	0	1	0	1	1	1
Echium plantagineum L.	0	1	0	1	0	1	0	1	1	1	1	1	0
Erodium cicutarium subsp. bipinnatum (Cav.) Tourlet	0	0	1	0	0	0	0	0	1	0	0	0	1
Geranium molle L.	1	0	1	0	1	1	1	0	0	1	0	0	0
Geranium purpureum Vill.	0	0	0	0	0	0	0	0	0	1	0	0	0
Hedypnois cretica (L.) Dum.-Courset	1	0	0	0	1	1	0	0	1	0	0	0	0
Hordeum murinum subsp. leporinum (Link) Arcang.	1	0	1	1	0	1	0	0	0	0	0	1	1
Hypochaeris glabra L.	0	0	0	0	1	0	0	0	1	0	0	0	0
Hypochaeris radicata L.	0	0	0	0	1	0	0	1	0	1	1	0	0
Juncus bufonius L.	0	0	0	0	0	0	1	1	0	1	1	0	0
Lathyrus angulatus L.	0	0	0	0	0	0	0	0	0	1	1	0	0
Leontodon taraxacoides (Vill.) Mérat	0	0	0	0	1	1	0	1	0	1	1	0	0
Lythrum borysthenicum (Schrank) Litv.	0	1	0	0	0	0	1	1	0	0	0	0	0
Medicago polymorpha L.	1	0	1	0	0	1	0	0	0	0	0	0	0
Mentha pulegium L.	0	1	0	0	0	0	0	0	0	0	0	0	0
Ornithopus compressus L.	0	0	0	0	1	0	0	0	0	1	0	0	0
Plantago coronopus L.	0	1	0	0	0	1	1	1	0	0	1	0	0
Plantago lagopus L.	0	0	0	1	1	1	0	0	1	1	1	0	0
Plantago lanceolata L.	0	0	0	0	0	0	0	0	0	0	1	0	0
Poa annua L.	0	1	0	0	0	0	1	1	0	0	0	0	0
Polygonum aviculare L.	0	0	0	0	0	0	1	0	0	0	0	0	0
Polypogon maritimus Willd.	0	0	0	0	0	0	0	1	0	0	0	0	0
Ranunculus parviflorus L.	1	0	1	0	0	0	0	0	0	0	1	0	1
Rumex bucephalophorus L.	0	0	0	1	1	1	1	0	0	0	1	0	0
Rumex pulcher subsp. woodsii (De Not.) Arcang.	1	1	1	0	0	0	0	1	0	0	0	0	1
Senecio jacobaea L.	0	1	0	0	1	0	1	1	0	1	0	0	0
Sherardia arvensis L.	1	0	0	0	0	0	0	0	0	0	0	0	0
Silene gallica L.	0	0	0	1	0	1	0	0	0	0	1	0	0
Sisymbrium officinale (L.) Scop.	0	0	1	0	0	0	0	0	0	0	0	0	0
Stachys arvensis (L.) L.	0	0	0	0	0	1	0	0	0	1	1	0	0
Tolpis umbellata Bertol.	0	0	0	0	0	0	0	0	1	1	0	0	0
Trifolium campestre Schreb.	0	0	0	0	0	0	0	0	1	0	0	1	0
Trifolium glomeratum L.	0	0	0	1	0	0	1	1	1	1	1	0	0
Trifolium medium subsp. medium	0	0	0	0	0	0	0	0	0	1	0	0	0
Trifolium repens L.	0	0	0	0	1	0	0	1	1	1	1	0	0
Trifolium resupinatum L.	0	0	0	0	0	0	1	1	0	0	0	0	0
Trifolium scabrum L.	0	1	0	0	0	0	0	0	0	0	0	0	0
Trifolium subterraneum L.	0	0	0	0	0	0	1	1	1	0	0	0	0
veronica sp	0	0	1	0	0	0	0	0	0	0	0	0	0
Vulpia bromoides (L.) S.F.Gray	0	0	0	0	0	0	0	0	0	1	0	0	0
Vulpia geniculata	1	1	1	1	1	1	1	1	1	1	1	1	1

Presence: 1; absence: 0.

(Appendix A). As we can see in these tables, floristic diversity was high in all treatments, and in May, a greater number of botanical species were identified. In this case, the main botanical species identified in the preferred points were Bromus hordeaceus L., Chamaemelum mixtum L., Diplotaxis catholica L., Echium plantagineum L., Geranium molle L., Plantago lanceolata L., Rumex bucephalophorus L., Trifolium glomeratum L., Trifolium repens L., and Vulpia geniculata L.

3.5. Soil Compaction Measured by Cone Index

Soil compaction (cone index) was measured in the two types of grazing and at the two depths (0 to 15 and 15 to 30 cm) (Figure 13). Although the stocking rate doubled in the DS relative to the CS and the CI values were higher in the DS, there were no statistically significant differences (p > 0.05). The average SMC values were 14.7% ± 2.9% for the CS area and 15.2% ± 2.3% in the DS area (depth 0–30 cm). Soil compaction maps were made from 0 to 15 (Figure 14a) and 15 to 30 cm (Figure 14b).

4. Discussion

The long-term evolution and maintenance of certain species in natural pastures composed of multiple plant families are influenced by how animals use these species throughout their phenological cycle. On the other hand, interactions between the soil type, topography, different species and floristic combination, height, and nutritional value can influence the preferred animal grazing areas. Pastures are very heterogeneous biological systems due to environmental factors, soil characteristics, variations in floristic composition, differences in the phenological cycles of species, and changes caused by grazing [41]. The floristic composition of the pastures is affected by the selectivity and the stocking rate [15], as well as the animals’ preferred areas for grazing.

4.1. Relationship Among the Use of Pasture, Height and Quality

During autumn, the combination of temperature and rainfall was sufficient to allow the pasture to germinate and grow satisfactorily. During winter, especially in January, the low temperatures limited pasture growth, and this continued until February, perpetuating itself for a few days in March and April. This caused the number of grazing days in the plots subject to DS to be lower than in those subject to CS (around 80 days). However, the pasture utilization rate was much higher, since on each day of grazing in DS, the number of animals was much higher. The absence of animals for nearly two months allowed some pastures to recover despite the slow growth rate, showing agricultural managers the importance of making grazing period decisions based on pasture height. Pasture resting periods are essential after each grazing period to allow the plants to restore their reserves and produce new leaves [17]. According to the same authors, recovery time depends on the plants’ response capacity, soil humidity, and temperature.

Measuring pasture height as an indirect indicator of photosynthetically active leaf area is crucial for managing the pasture itself, as well as grazing, showing high correlations with pasture biomass production [42,43]. Furthermore, Bell et al. [23] state that pastures with an average height of less than 70 mm tend to have low nutritive value. In another study carried out by Santos et al. [27], in which the percentage of living leaves (indirectly, greater nutritive value) was compared at four pasture heights (15 cm, 25 cm, 35 cm, and 45 cm), it was concluded that this percentage is greater when the average height of the pasture is 15 cm and 25 cm. Iason et al. [44] found that daily pasture intake by sheep in CS was limited when the pasture was 30 mm height, which did not occur above 55 mm.

The application of dolomitic lime and the type of grazing did not influence the mean values of CP and NDF (p value > 0.05). These values contrast with those found by Xiao et al. [15]; the authors found that the higher the pasture height, the lower the CP percentage and the higher the NDF value. According to this study, with pasture height values around 15 cm, CP and NDF values are 3% and 75%, respectively. On the other hand, when the average pasture height was around 12 cm, the CP and NDF values were 7% and 65%, respectively. In our study, the pasture quality (CP and NDF) was based on a composite sample. This is a limitation of our study, which makes it impossible to relate the nutritional value to each sampling point, penalizing the eventual association between preferred areas and nutritional value. In a study in China comparing the sheep pasture selectivity in four grazing systems at different periods of the year, the authors found that the grazing systems did not affect biomass production, density, height, or pasture quality [29]. Voisin and Lecomte [17] note that DS can contribute to improved pasture floristic composition and consequently its nutritional value, which was not observed by our study in statistical terms.

On the other hand, the application of dolomitic lime appears to have a positive influence on pasture height, resulting in greater heights in the plots P3TD and P4TC when compared to plots P2UD and P1UC, respectively. In this sense, it seems to us that the application of dolomitic limestone is a great advantage, leading to greater biomass production in the pasture, thus allowing a greater number of animals per hectare. In this way, the efficiency of extensive production systems can be improved without apparently harming the Montado ecosystem. Between P2UD and P3TD treatments, the variations in height amplitudes were quite similar. However, considering that both treatments have the same grazing intensity, given the additional decrease in pasture height in the P2UD, a lower pasture growth rate can be inferred, which could be attributed to the application of dolomitic lime.

According to Carvalho [45], the application of dolomitic lime in some soils can increase the production of dryland pastures up to five times. In previous work carried out by our team in this test field, there was a positive effect of liming in the soil, though very slow, on pH and the Mg/Mn ratio and increased pasture productivity. In some of these studies, it was also shown that pasture productivity increased with the application of dolomitic limestone. Therefore, the continuation of these studies in the long term is justified.

4.2. Relationship Between Preferred Grazing Locations, Type of Grazing, Stocking Rate, Soil Compaction, Floristic Composition, and Pasture Height During the Observation Period

This study intended to understand the development of the preferred grazing areas in the four treatments from the end of winter to the beginning of summer (without biomass restrictions). In addition, we also aspired to understand if the animals had the same geographic grazing pattern in the four treatments and if some significant factors could explain those differences.

The stocking rate used in CS was higher than in the usual production systems (two to three sheep per hectare) [46]. Even so, biomass availability in all plots meant that sheep could choose the areas with the most preferred species (CS and DS). In this sense, stocking methods with high stocking rates are often identified as responsible for the degradation of soil, pasture, and trees in the Montado ecosystem. Animal trampling due to high stocking rates is correlated with negative effects on soil properties [16]. However, in our case, there were no significant differences between the two stocking methods tested (CS and DS), which encourages producers in the Montado ecosystem to intensify sheep production, with a greater number of animals per hectare, in DS (grazing periods depending on pasture height). Although the CI values in the plots subject to DS were higher than those recorded in CS, the fact that there were no significant differences takes us to the concept of sustainable intensification. Higher CI values were recorded at depths of 15 to 30 cm. However, several studies report that soil compaction due to animal trampling occurs mainly in the surface layers of the soil [47,48,49].

The greater amplitudes in pasture heights in the CS treatments indicate a tendency towards more selectivity, with potential negative effects on pastures. In the DS, this amplitude is smaller, although there are areas where the pasture has been consumed more than others. In general, the spots with the lowest grazing preference were those with the highest pasture heights. Furthermore, according to Di Grigoli et al. [50], even in the most preferred areas, where biomass availability decreases, sheep end up ingesting growing plants (due to a delay in phenological cycles) with high CP content and low NDF content. The distribution of sheep grazing across the pasture depends on external factors, such as topography and climate [28].

On the six observation dates (Figure 12), in all treatments, there was a tendency for the lowlands to be preferred grazing areas, where legume plants prevailed [13]. In heavily grazed areas, all plants can have access to light, with benefits for the Fabaceae family [20]. This shows agricultural managers the importance of homogeneous (non-selective) grazing across plots, allowing for the ingestion of different botanical species in an equal manner to allow the emergence and development of legumes (greater nutritional value [17]). On the other hand, on the six observation dates and treatments, the areas less grazed by sheep were those near the road (Figure 12), where many plants of the genus Urtica and Carduus were identified. This explanation is only partially valid for P3TD and P4TC. In fact, the animals at P1UC and P2UD tended to avoid areas closer to the road; however, when the favored species began to congregate at other points, the animals increased their frequency in those areas. The fact that sheep graze for less time near the road also can be a result of frightening factors related to passing cars and trucks changing their behavior. Clair and Forrest [51] state that vehicle traffic alters the natural behavior of elk, reducing their normal behavioral patterns and increasing levels of vigilance. However, punctual observations were made that allow us to infer that these points were grazed on during the night or at the end of the day and in the early hours of the morning, when the passage of vehicles was almost non-existent.

On the first dates of observation, the frequency of drinking was low, but as spring progressed (from April), the higher fiber content of the pasture and the higher temperature led to a greater frequency of visits to the watering trough. As a result, the animals tended to graze much closer to this area. In the dryland pastures, the choice of a grazing area often reflects the effort to reduce energy expenditure when walking rather than a true feed preference [24].

In P3TD, there was an area highly preferred by sheep for grazing (points 30 and 31). In this same plot, the points in the high and middle zone were the least preferred. Sheep prefer to graze in areas with an average pasture height of 60 mm rather than 300 mm [52]. In P4TC, the preferred area for grazing on all observation dates was in the lowland area, where more leguminous species were identified. Only the area near points 38 and 39 was less grazed (characterized by high pasture close to the road). In this sense, livestock producers should be aware that animals prefer pastures with medium heights, avoiding those that are too high, which contributes to selectivity. Therefore, the stocking rate should be established according to the availability of biomass.

In winter (14 January) and at the peak of spring (4 May), the botanical species were identified at each sampling point. In this work, we include only the inventory of botanical species identified in the places preferred by sheep for grazing. Of the plants identified at the sampling points that are most preferred for grazing, some, such as Senecio vulgaris L., Senecio jacobaea L., Echium plantagineum L., Iris xiphium L., and Ranunculus ollissiponensis subsp ollissiponensis Pers. are toxic to ruminants in certain phenological states, and sheep avoided them [53]. On the other hand, other botanical species belonging to the Fabaceae family (clovers and Ornithopus compressus L.) are not very palatable in the first phenological stages and are only consumed by sheep in the middle and late spring and summer as dry feed.

Although our study plots were only 1 ha in size, from April onwards, due to an increase in temperature and fiber contents in plants, the drinking frequency increased, which led to an increase in grazing in areas near water troughs. The importance of proximity to the drinking trough increases as summer approaches [54].

In April, preferred grazing areas expanded to almost all the plots. This fact may have occurred due to the decrease in the amount of water in the soil and the wind that was felt at the end of March and beginning of April, which harmed the development of the pasture, leading to the early maturation of some species. In areas with a high availability of biomass, sheep will tend to prefer places where the species have a higher nutritive value, trying to select plants that have a higher CP and a lower fiber content [55]. Different preferred grazing locations throughout the season may reflect different pasture characteristics in terms of quality and quantity [29]. In addition, the floristic composition of the pasture is affected by grazing, namely through selectivity, stocking rate, and grazing season [15].

In points 3, 4, 5, 10, and 11, several botanical species characteristics of nitrophilous zones with low palatability, low nutritive value, or even toxicity were also identified. The Poa genus shows significant initial growth but tends to have a short cycle, reducing its palatability early on. Sheep appreciate the Cynodon genus at certain stages of its phenological state. Still, they soon show high levels of fiber and significant reductions in CP content, causing sheep to reduce their preference for these plants. The plants of the genera Rumex and Arum are toxic and normally avoided by animals, as well as the Urtica genus [53]. These areas have a high density of Urtica spp., which sheep tend to avoid. Similar situations were observed in P2UC at points 23 and 24 and in P3TD at points 25 and 36. On the other hand, other parts of these locations had shade from the tree canopy, which may protect the pasture from wind and sun, thus preserving a higher soil moisture level. These plants display more green leaves, corresponding to an early stage in the phenological cycle. The higher percentage of organic matter (due to the tree leaves and branches) and the shade provided greater soil moisture in spring and a consequent delay in the plants’ phenological cycles. This occurred at the end of May in the upper part of the plot P4TC, when grazing animals were observed near the watering trough, and under the tree shade. In addition, animals prefer plants in their initial phenological stages when they contain less fiber and more protein [55].

In June, the vegetation was dry, and sheep grazed at a greater variety of locations, suggesting that they had no natural preference for a particular area, confirming Santos’s [27] description that grazing selectivity decreases as pasture quality decreases. According to Carreira et al. [13], there were no significant differences in the probability of occurrence of most of the identified plants in the four treatments. However, the DS tends to favor the appreciation of plants with greater palatability and nutritive value (i.e., Trifolium genus, especially Trifolium repens L.), also contributing to the reduction of botanical species of lower feed value (i.e., Rumex genus, Diplotaxis catholica L., Senecio jacobaea L.) for sheep.

To better understand the relationships between pasture quality and preferred grazing location, we suggest that a similar study be carried out characterizing the quality of the pasture at each sampling point. In future works, it also would be interesting to carry out observations on the behavior of grazing animals throughout the pasture’s vegetative cycle (autumn, winter, spring, and summer).

5. Conclusions

In the treatments associated with DS, the lower number of grazing days (around 80) was compensated for by the greater number of sheep per hectare in these plots, compared to CS. In this sense, we found that in plots subject to DS, pasture consumption was 50% higher. The application of dolomitic limestone allowed 7 more days of grazing in the P3TD treatment than in the P2UD, which shows greater biomass production, encouraging livestock producers to apply it.

In statistical terms (with 95% confidence), pasture quality was not affected by the stocking method and the application of dolomitic lime.

Throughout the period of sheep grazing observations, a similar pattern of preferred grazing areas was observed among the four treatments, with the lowland areas presenting more grazing density.

At the beginning of summer (June), the pasture was almost dry, and sheep grazed more evenly across the plots, with no evident areas of preferential grazing. Higher stocking rates (P2UD and P3TD) did not provide a more homogeneous distribution of grazing areas across the fences. Even with 16 sheep per hectare, highly preferred areas were observed, especially in P3TD, which means that if there are no biomass limitations, there will always be areas of the pasture and species that the sheep prefer to graze on first. The floristic composition does not seem to have been decisive for the choice of grazing locations. However, agricultural managers should consider that the stocking rate should be established according to the availability of biomass (height), leading to homogeneous grazing of the plots, avoiding selectivity and, consequently, improving pasture quality.

On the other hand, the fact that sheep do not ingest, or only ingest at certain times, certain botanical species (for example, the genera Rumex, Urtica, and Arum) due to their lower palatability or because they are toxic allows the maintenance of these plants that are an integral part of the ecosystem, thus contributing to the conservation of biodiversity.

In addition, periods for pastures are also essential for them and should be carried out based on their height.

The fact that in DS, soil compaction is not statistically different from that in CS shows that the sustainable intensification of sheep production in the Montado could be possible without degrading this ecosystem due to excessive trampling by grazing animals.

The results indicate that higher stocking rates, wisely used to maintain adequate recovery periods, tend to favor a more uniform biomass growth, revealing greater species homogeneity and variability. However, given climate variability and the trend towards higher levels of aridity, studies will be needed over several years to analyze the evolution of soil organic matter and compaction and the monitoring of species and their relative preponderance to preserve biodiversity.

This work contributes to understanding the relationships between different types of grazing in dryland pastures, with and without application of dolomitic limestone, and preferred grazing locations for sheep.

This study can be the source for more informed decision-making by agricultural managers to promote the sustainability of the Montado ecosystem as well as the efficiency of ruminant production systems, aiming for animal welfare.