Introduction

Agrosilvopastoral systems are highly complex due to the integration of different feed resources such as crops, trees, and grassland, as well as several ecosystem services (Pardini and Nori 2011). As a consequence, managing these systems means understanding and regulating the interactions among the components to increase positive outcomes (Jose et al. 2019). For the further development of agrosilvopastoral systems, enhancing sustainability and resilience is crucial because climate change is threatening their survival (Bernués et al. 2011).

In marginal and inland areas of the Mediterranean basin, agrosilvopastoral systems represent an opportunity for the development of rural communities. This is because they satisfy the increasing demand for high-quality animal products, maintain a high level of animal welfare, and are able to cope with climate change through both adaptation and mitigation strategies (Mele et al. 2019). Besides their environmental and economic importance, agrosilvopastoral systems can help preserve traditional landscapes (Paris et al. 2019).

In rainfed Mediterranean areas, due to the highly variable interannual precipitation patterns, the grazing season is short and concentrated during autumn and spring, while in the summer, these systems face severe pasture shortages which are the most important resource for animal diets. The situation is likely to be worsen due to the consequences of climate change as Mediterranean countries will experience increased droughts and extreme temperatures (European Commission 2009). Besides reducing herbage availability and quality, climate change decreases animal welfare (Lacetera 2019), with a consequent decrease in animal production and reduced income for farmers. Agrosilvopastoral systems have been recognized as a strategy to adapt livestock production to climate change, above all through the diversification of feed sources and improvement of animal welfare (Pulina et al. 2021).

Tailored management strategies that improve the environmental and socio-economic sustainability of agrosilvopastoral systems are essential to maintain farming activities. Improving knowledge regarding the relationship between the soil, forage resources and animals is essential to enhance grassland and animal productivity and sustainability (Stevens et al. 2021). In the last few decades, Mediterranean grasslands have received little attention from the scientific research as they are perceived as marginal land where grazing activity is the only viable solution. However, if managed and enhanced correctly, grasslands can become an important source of feed for livestock, thereby improving the system's overall productivity (Porqueddu et al. 2016).

The adoption of rotational grazing provides many benefits, including increased herbage production, reduced grazing pressure and greater animal productivity (Sanderman et al. 2015). In fact, rotational stocking enables cattle to graze on higher quality pasture by using forage at a relatively young growth stage (DeRamus et al. 2003). In addition, the adoption of correct grazing management can maintain the soil function in delivering ecosystem services (Teague and Kreuter 2020) and is considered a strategy for the sustainable intensification of grassland based livestock systems (Pretty 2018). Selecting the best grazing management, considering local agri-environmental conditions and animal use, are therefore essential to reducing the impacts on soil and to improve animal production (Avondo et al. 2013). The benefits from rotational grazing have been widely assessed, but further research is needed to exploit the full advantages of silvopastoral systems in Mediterranean areas (Aguilera et al. 2020). The pasture and grazing management, and the effect of trees on animal production and welfare have been recently investigated in tropical and subtropical silvopastoral systems, however, there is a lack of studies where the interactions of these elements are explored in temperate and Mediterranean silvopastoral systems (Vandermeulen et al. 2018).

In our study the aim was thus to compare two rotational grazing management systems, silvopastoral and pastoral, in an on-farm experiment during the main grazing season (spring). The effects were therefore evaluated on several indicators of sustainability and resilience, such as: (1) pasture quantity and quality, (2) herbage intake, (3) animal welfare and productivity, and (4) soil cover.

Materials and method

Study site and experimental design

The grazing trial was carried out on a real farm, and on 3.69 ha of temporary grassland located in southern Tuscany, Italy (42° 57′ 10.9″ N 11° 14′ 25.4″ E, 87 m.a.s.l.) from 25 March 2021 to 6 May 2021 (total = 42 days). The soil has a silty-loam texture (clay: 13%; loam: 50.5%; sand 36.5%), an acid pH (5.5), a low soil organic matter content (1.89%), and a balanced C/N ratio (10.3). The available phosphorus content is deficient (P2O5: 10.84 mg kg−1) while the exchangeable potassium is high (K2O: 190.34 mg kg−1). In 2017, an overseeding of Lolium multiflorum Lam., Dactylis glomerata L., Festuca arundinacea L., Trifolium pratense L., Lotus corniculatus L. was carried out on grassland surface with a broadcasting seeder. The farm management was certified according to EU organic rules, and no inorganic fertilisation had been performed on the grassland in the previous eight years. From 2018 to 2020 the grassland was grazed by cattle with continuous stocking during winter and rotational stocking during spring. In 2021, cattle were allowed to graze to standard sward height (from 11 January to 1 February). After this period, the pasture had not been used until the experimental trial.

The experiment was designed to compare two grazing systems: (1) pastoral (PA) and (2) silvopastoral (SP). On the start date of the trial (25 March 2021), 50 growing cattle of the Maremmana breed were allotted to two groups corresponding to a PA or SP grazing system. The PA group was only allowed to access the pasture, whereas the SP group was allowed to access the forest and the pasture. The forest surface available to the SP group consisted of about 3.31 ha of Turkey oak (Quercus cerris L.) high forest. On the 8th of April of 2021, a stand structure survey was performed. The results shown a net prevalence of Turkey Oak trees with diameter at breast high (DBH) higher than 12.5 cm over a scares presence of various tree species with DBH between 2.6 and 12.5 cm. Finally trees and shrubs with a DBH less than 2.5 cm were completely absent (Manetti et al. 2022). On the grassland, rotational grazing was performed by dividing the whole surface into six paddocks, three paddocks for each treatment (total n = 6; P1, P2, P3, P4, P5 and P6) as reported in Fig. 1. Animals were randomly allotted to the two experimental groups at the beginning of the trial. The characteristics of the two groups are shown in Table 1.

Fig. 1
figure 1

Aerial picture of the study site. Paddock P1, P2 and P3 were assigned to the pastoral (PA) system while paddocks P4, P5, and P6 were assigned to the silvopastoral (SP) system

Full size image
Table 1 Characteristics of the animals in each experimental group (value ± standard deviation)
Full size table

From the start date, cattle were allowed to graze in each paddock for one week, defined as the grazing period (GP), and to return on the same paddock after three weeks. Cattle grazed twice in the same area, and the trial finished on 6 May, after six GPs. The trial was planned to last another three GPs, however it ended earlier due to the absence of pasture regrowth due to drought conditions. Table 2 reports the surface and stocking rate of each paddock.

Table 2 Paddock characteristics in relation to treatment and grazing period. LU means livestock unit
Full size table

Both groups were supplemented with feed produced on-farm: a mixed flour of corn (Zea Mais L.) and triticale (x Triticosecale spp.) and oat (Avena sativa L.)—vetch (Vicia sativa L.) hay. Flour supplementation was calculated based on the average bodyweight of the entire group: nearly 1% of the average bodyweight. Consequently, at the beginning of the trial, flour supplementation was around 3 kg head−1 and was provided together with hay administrated ad libitum.

Measurements

The microclimatic conditions were monitored by an onsite meteorological station recording rainfall and air temperature, and a public meteorological station, located at a distance of one kilometer, (Regione Toscana 2021) was used to collect long-term data.

An exclusion cage technique was used to measure herbage growth and disappearance during each GP (Undi et al. 2008). Exclusion cages consisted of 1 m × 1 m metal fence with large-mesh wire that prevented the animals from grazing within the area. Five exclusion cages per paddock were placed following the gradient of the herbage mass variability. A total of thirty exclusion cages were used. Exclusion cages were moved once the paddock had been grazed.

Herbage mass was collected by cutting an area of 0.25 m2 at a stubble height of 0.03 m with the same person always performing the cuttings with a single-handed reaping hook. Before cuttings, a visual qualitative assessment of the plant species was performed; the percentage of grass, legume and weed present in 0.25 m2 was always estimated by the same person who performed the cuttings. The fresh weight of the above-ground biomass was measured immediately after harvesting the herbage.

Five replicates of pre-grazing herbage mass for each paddock were collected at the beginning of each grazing period. The sampling was performed in the morning. On the other hand, at the end of each GP, five representative post-grazing and five representative non-grazed herbage samplings (exclusion cages) for each paddock were collected in the afternoon. The time of the sampling was kept constant to be able to compare the nutritive value of samples during the experiment grazing periods with a comparable concentration of non-structural carbohydrates and proteins (Buxton 1996).

The daily herbage growth (\(DHG\)) expressed as \({\text{g}}\,{\text{DM}}\,{\text{m}}^{ - 2} \,{\text{d}}^{ - 1}\) was calculated from the difference between the pre-grazing herbage mass and the non-grazed herbage mass over the grazing period duration using Eq. 1, as previously proposed by Mantino et al. (2021).

$$DHG \left[ {{\text{g}}\,{\text{DM}}\,{\text{m}}^{ - 2} {\text{d}}^{ - 1} } \right] = \frac{{ECHM\left[ {{\text{g}}\,{\text{DM}}\,{\text{m}}^{ - 2} } \right] - PreHM\left[ {{\text{g}}\,{\text{DM}}\,{\text{m}}^{ - 2} } \right]}}{{D\left[ {\text{d}} \right]}}$$
(1)

where ECHM is exclusion cage herbage mass, PreHM is pre-grazed herbage mass, while D is the length of the grazing period in days.

To calculate the herbage allowance (HeA), which is the relationship between the forage mass and animal live weight or number of animals over a determined time, the equation proposed by Sollenberger et al. (2005) was used (Eq. 2).

$$HeA \left[ {{\text{g}}\,{\text{DM}}\,{\text{kg}}\,{\text{BW}}^{ - 1} {\text{d}}^{ - 1} } \right] = \frac{{\left( {\frac{{PreHM\left[ {{\text{g}}\,{\text{DM}}\,{\text{m}}^{ - 2} } \right]}}{{D\left( {\text{d}} \right)}} + DHG \left[ {{\text{g}}\,{\text{DM}}\,{\text{m}}^{ - 2} {\text{d}}^{ - 1} } \right]} \right) \times A \left[ {{\text{m}}^{2} } \right]}}{{\overline{{{\text{BW}}}} \left[ {{\text{kg}}} \right]}}$$
(2)

where A is the paddock area, and BW is the average body weight of the whole treatment group.

Finally, the potential herbage intake (\(PHI\)) per animal and per kg of body weight (BW) were calculated with Eq. 3 which we developed and is reported in a previous paper Mantino et al. (2021). HeA and PHI were estimated for each treatment group.

$$PHI \left[ {{\text{g}}\,{\text{DM}}\,{\text{d}}^{ - 1} } \right] = \frac{{\left( {ECHM\left[ {{\text{g}}\,{\text{m}}^{ - 2} } \right] - PostHM\left[ {{\text{g}}\,{\text{m}}^{ - 2} } \right]} \right) \times A \left[ {{\text{m}}^{2} } \right]}}{{D\left[ {\text{d}} \right] \times \overline{{{\text{BW}}}} \left[ {{\text{kg}}} \right]}}$$
(3)

where ECHM is the exclusion cage herbage mass, PostHM is the post-grazed herbage mass, A is the paddock area, D is the length of the grazing period in days, and BW is the average body weight of the whole treatment group.

An unmanned aerial vehicle (model DJI PIII; DJI Technology Co., Shenzhen, China) equipped with a Parrot Sequoia+ (Parrot, Paris, France) multispectral sensor was used to collect the aerial photos. Aerial photos of the whole study-site were taken at the start and the end of the experiment. Aerial images were processed with Agisoft Metashape (1.6.2) software. NDVI (Eq. 4) index was used to retrieve vegetation soil cover at the beginning and at the end of the trial.

$$NDVI = \frac{NIR - Red}{{NIR + Red}}$$
(4)

where NIR and Red stand for the spectral reflectance in the near-infrared region and red region, respectively.

Every day, the amount of supplement residues left in the feed trough was measured to estimate the supplement intake of the experimental group. Herbage, hay and concentrate samples were dried to determine their dry matter (DM) content and their nutritive value. In particular, the following parameters were evaluated: (1) neutral detergent fibre (NDF), acid detergent fibre (ADF), and acid detergent lignin (ADL) content following Van Soest et al.'s (1991) methodology; (2) crude protein (CP) and etheric extract (EE) contents were assessed using the AOAC (1990) methodology; (3) Ash content by ashing at 550 °C for 5 h in a muffle furnace. The nutritive value of herbage sampling was calculated as net energy (NE) for growing using CNCPS v6.5/v6.55 model (Van Amburgh et al. 2015). Chemical composition and nutritive value are shown in Table 3.

Table 3 Chemical composition and nutritional value of supplements given to animals
Full size table

All 50 cattle were weighed three times: at the start (T0 = 25 March), in the middle (T1 = 15 April) and at the end (T2 = 6 May) of the trial in order to calculate their average daily gain (ADG). To monitor animal welfare levels as stress indicators, individual samples of blood were collected from a subsample of 27 animals, 13 in SP and 14 in PA. Blood samples were collected from the caudal vein on the same day as the weight recording in concomitance with routinary clinical monitoring carried out by the veterinary of the farm. Hair samples were collected from the rump region only at the start and at the end of the experiment, due to the absence of hair regrowth in T1.

Serum cortisol concentration was determined through CLIA technology (IMMULITE 2000 XPi Immunoassay System Siemens). The hair cortisol concentration was determined using an ELISA kit for saliva cortisol (DES6611, Demeditec Diagnostics GmbH) validated for hair samples according to Heimbürge et al. (2020). The cortisol was previously extracted following a procedure provided by Bacci et al. (2014).

The present study was approved by the Animal Care and Use Committee of the University of Pisa (prot. n. 6/2020).

Statistical analysis

Statistical analysis was performed using R software (R Core Team 2021). Pasture parameters, describing herbage production, use, and quality, were analysed in order to assess the differences between PA and SP grazing systems (factor S), and between grazing periods (factor GP). The effect of S and GP was determined using the lme() function for linear mixed-effect models of the ‘nlme’ R package (Pinheiro et al. 2020), with factors S (2 levels) and GP (6 levels) as fixed effects and paddocks (P) (6 levels) as random effect.

$${y}_{ijk}=\mu +{S}_{i}+G{P}_{j}+{\left(S\bullet GP\right)}_{ij}+{\mathrm{P}}_{k}+{\varepsilon }_{ijk}$$

In order to assess the differences in animal performance and welfare between PA and SP systems, body weight, hair and serum cortisol were analysed. The effects of grazing system (S) and time of sampling (T) were determined using the lme() function for linear mixed-effect models of the ‘nlme’ R package (Pinheiro et al. 2020), with factors S and T as fixed effects and animal (A) as random effect. Sampling times were: (T0) start date of the trial, (T1) after three weeks (except for hair cortisol) and (T2) after six weeks of trial (end trial).

$${y}_{ijk}=\mu +{S}_{i}+{T}_{j}+{\left(S\bullet T\right)}_{ij}+{A}_{k}+{\varepsilon }_{ijk}$$

Homogeneity of variance was checked through Bartlett’s test, while normal distribution of residuals was checked with the Shapiro–Wilk test. Tukey’s HSD post-hoc test was carried out by pairwise multiple comparisons using the ‘emmeans’ R package (Lenth et al. 2020) with the emmeans() function.

Results and discussion

Meteorological data

Figure 2 shows the microclimatic conditions recorded during the trial. The mean air temperature was about 12% lower (11.5 vs. 13.1 °C) than the long-term data (2003–2020). There was no significant precipitation during the trial; in fact, the cumulative rainfall was around 40 mm for the whole trial period. In addition, during the month preceding the trial there was only 7.2 mm of rainfall. Overall, from 1 January 2021 to the end of the trial, only 215 mm of rainfall was recorded. In contrast in the same period, the long-term (2003–2020) average cumulative rainfall was 284 mm (about 32% more than during the experiment). Given this low value of rainfall, we can assume that there was a low water content in soil at the beginning of the trial. The microclimatic variables recorded did not suggest a heat stress risk for cattle as the black globe humidity index measured was always below the critical value of 75 (Buffington et al. 1981).

Fig. 2
figure 2

25 March 2021–6 May 2021 daily rainfall, mean air temperature, black globe temperature and humidity index (BGHI), and long-term 2003–2020 daily mean temperature in Paganico (Grosseto, Italy)

Full size image

Pasture characteristics

The statistical analysis performed on the pasture characteristics did not show any significant differences between the two systems (PA and SP) for the variables determined pre- and post-grazing: herbage mass, CP, NDF and NEg (Table 4). We can therefore assume that the herbage mass quantity and quality were homogenous for both groups at the beginning of each GP. In contrast, significant differences were found among GPs for all the variables, indicating that there were changes in pasture production and nutritive value during the trial.

Table 4 Results of statistical analysis of pasture characteristics
Full size table

The pre-grazing herbage mass did not differ between systems (averaging 114.6 g DM m−2) and was considerably lower than the values reported in the literature for grazing trials in both temperate and tropical areas. In the same environment and season, in a previous paper (Mantino et al. 2021) we demonstrated that with an adequate fertilization on more fertile soil, it is possible to reach higher pasture production in southern Tuscany (about 265 g DM m−2). In New Zealand, Realini et al. (1999) reported data on pre-grazing herbage mass in Brazil that varied from 119.3 to 379.3 g DM m−2. Piña et al. 2020, reported a pre-grazing mass of 266.28 to 337.42 g DM m−2, while Ruggieri et al. (2020) reported an amount higher than 500 g DM m−2, Simonetti et al. (2019) from 126.8 to 511.1 g DM m−2 and Difante et al., (2009) from 713 to 860 g DM m−2). Drought conditions observed before and during trials have been reported to accentuate the effect of poor soil fertility (as in our case, low pH and phosphorus content) on pasture productivity (Seligman and Van Keulen 1989). These conditions coupled with low-input organic management might explain the differences with the studies cited in which chemical fertilization was performed.

As expected, during the first GP, pre- and post-grazing biomass, and nutritive values were the highest. During GP4, GP5, and GP6, pre-grazing herbage mass was considerably lower than the first three GPs. Average pasture growth (3 g DM m−2 d−1) during the trial was not sufficient to compensate for the forage mass grazed by the animals. Considering that in the Mediterranean, spring is when the peak herbage growth is reached, the production we measured was lower than the potential peak growth of about 11 g DM m−2 d−1 reported for other pasturelands in the Mediterranean (Porqueddu et al. 2016). Scarce pasture regrowth caused by water stress conditions can negatively affected rotational grazing. In fact, the drought escape strategy of plants with earlier headings is needed to ensure that enough seed is produced before the summer season (Norton et al. 2016). These conditions might have led to the application of the wrong grazing management, as the most suitable time for grazing is after tillering and before flowering (Undersander et al. 2002). In our experiment, which was characterized by the high number of dry days, cattle grazing exacerbated the reduction in pasture growth potential.

Furthermore, in our study we observed no differences between systems regarding CP, NDF and NEg, which also means there was no influence of grazing management on pasture quality, since this management approach was only adopted for six weeks. Due to the short period of our experiment, we are not able to confirm that grazing management affects herbage quality as reported by Henkin et al. (2011),who highlighted how a high stacking rate and grazing intensity can increase herbage quality during the green season. In a mid-term (five year) experiment Teague et al. (2004) demonstrated that weather and grazing management have a crucial impact on plant species and quality of pasture. Although the results of the chemical analysis of pre-grazing herbage mass are comparable to values reported in the literature (Landau et al. 2018), pasture quality remained poor throughout the trial, with NEg values below 1 Mcal kg−1 DM.

Overall, pasture quality was higher at the beginning of the first GP and then decreased. The low CP content of the herbage was in line with the botanic composition of the grassland. In fact, as shown by the visual qualitative assessment performed during herbage sampling: a net prevalence of Poaceae (65%) and weed species (30%) and a very low percentage of Fabaceae species (5%). These findings confirm that herbage maturity is the most critical factor influencing forage quality (Buxton 1996). Generally, during spring growth, the forage quality declines due to the increased air temperature leading to rapidly advanced plant maturity (Van Soest 1994). Since we only evaluated herbage quality for six weeks and temperatures remained relatively low for the entire trial, we measured a significant decrease in herbage quality characterised by an increasing trend in fiber content as the season progressed, as expected.

No significant interaction was found between the GP and S, and thus we can assume that the two factors are independent.

Because of the low pre-grazing herbage mass at the beginning of the grazing periods, the HeA was also low. No significant difference was found between systems, whereas significant differences were found among GPs (p < 0.0001). We noted a significant reduction in HeA in GP4, GP5, and GP6, as a direct consequence of pre-grazing herbage mass reduction. Again, the average HeA reported in the literature (Piña et al. 2020, 25 kg DM cow−1 d−1; Simonetti et al. 2019, 40–100 g DM kg BW−1) is higher than our case (17.55 g DM kg BW−1 d−1).

Daily intake

The estimated potential herbage intake did not present any significant difference between systems with an average of 8.34 g DM kg BW−1 d−1 (Table 4). In contrast, significant differences (P < 0.05) were found among GPs. A significant reduction in potential herbage intake was registered in GP4, and GP6 as a direct consequence of the reduction in pre-grazing herbage mass and forage allowance due to the poor regrowth of vegetation after the first grazing period. A limiting forage allowance reduces the herbage intake of cattle at pasture (Dougherty et al. 1989).

In our study, the potential herbage intake was lower than that reported in the literature (Mccaughey et al. 1997; Difante et al. 2009; da Frota et al. 2017). There are several factors affecting the level of herbage intake: (1) body size and condition, (2) physiological status, (3) amount and quality of concentrate supplementation, (4) forage preferences, (5) forage availability, and (6) grazing system (Allison 1985). In our experiment, forage allowance significantly affected herbage intake, as previously reported by Piña et al. (2020). In fact, herbage mass available on pasture was not sufficient to satisfy the energy and protein requirements of the grazing cattle.

As a consequence, feed integration was administered in order to reach a total daily intake that covered the animal nutrient requirements. The nutritive value of supplements was consistent with reference values for oat-vetch hay, corn and triticale flour grown in Mediterranean countries provided by INRA (2018). Adding supplements, the total estimated daily intake was 27.0 g DM kg BW−1 d−1 for the SP system and 23.6 g DM kg BW−1 d−1 for the PA system. Table 5 shows the average supplement consumption during the experiment for both groups. The supplement intake increased in GP4, GP5, and GP6 in line with the decline in herbage intake and nutritive value. The consumption of supplements was also particularly high in SP, probably due to the higher energy expenditure related to the larger surface area that the animals explored for browsing and walking. Our results are in line with Cañas et al. (2003) who reported an energy loss related to animals having to move around continually where there is a low herbage mass and quality and the consequent stress produced.

Table 5 Average daily supplement, herbage and total intake of grazing cattle in spring 2021 according to the investigated systems and grazing period
Full size table

Animal welfare and performance

As described previously, the climatic conditions did not influence the environmental parameters related to the risk of heat stress, as the BGHI values indicated a condition of thermoneutrality. The estimated total daily dry matter intake was around the reference value of 2–2.5% of BW, suggesting that the feed availability was adequate overall for the animals' requirements in both treatments.

The serum cortisol levels measured on the sampling day were higher than values reported in the literature (Table 6) (Bristow and Holmes 2007; Ghassemi Nejad et al. 2019, 2020). This suggests that handling during the blood samplings probably induced higher stress in the Maremmana cattle than in other cattle breeds reported in the literature. Maremmana cattle are usually managed in extensive systems where there is little interaction with humans. The stress induced by the handling during the sampling day was higher in Maremmana cattle compared to other breeds that are more used to the presence of humans. Statistical analyses did not show any significant differences between the system and times (Table 6), indicating that the cattle were always subjected to the same stress level on all sampling days.

Table 6 Average values and standard error of body weight, serum and hair cortisol, in the two grazing systems: Pastoral (PA) and silvopastoral (SP), and sampling times start (TO), middle (T1), and end (T2)
Full size table

As with the serum cortisol, there were no differences in hair cortisol levels between systems, whereas there was a significant decrease from the starting date to the end date (from 9.68 to 7.95 pg mg−1 in T0 and T2, respectively). The hair cortisol concentration is comparable to values reported in the literature (Ghassemi Nejad et al. 2019; Heimbürge et al. 2020). Since the hair cortisol is considered a good indicator of long-term stress, the environmental, nutritional and management factors of both treatments did not induce significant stress in the animals.

It can be assumed that all cattle were subjected to the same conditions given that (1) SP and PA grazing systems did not have a significantly different herbage allowance and consequently no significant potential herbage intake; (2) the quality of pasture and concentrate did not present any significant differences; and (3) in neither group was the potential sources of environmental stress (management, nutrition, and climate) significantly affected the hair cortisol. The only difference was in the amount of surface available for grazing and browsing under the canopy of the turkey oak trees.

Regarding animal performance, the statistical analysis showed a significant difference in time (P < 0.0001, and a significant first level interaction (P = 0.0011) between time and grazing system. In fact, the post hoc comparison showed that at the start of the trial, cattle body weight was not significantly different between the two systems (P = 0.1), but in the middle and at the end of the grazing trial, significant differences were recorded: P = 0.08 in T1, and P = 0.04 in T2 (Table 6). Moreover, the slightly higher percentage of steers presents in the PA respect to SP (45% vs. 38%, respectively) did not influence the faster growth of the PA. Indeed, when sex factor was added in the statical model was not significant and it was not used for the analysis of data. This result is consistent with outcomes reported by Blanco et al. (2020) for steers and heifers of Pirenaica breed.

These results highlighted that throughout the entire experiment, the SP group showed a lower daily gain, around 20% less than the PA group (1.02 kg d−1 vs. 1.20 kg d−1). The values found in our study are in line with the literature. Conte et al. (2019) reported similar ADG values (1.05 kg d−1) for young Maremmana bulls in Tuscany during the finishing period, while Foggi et al. (2021) reported a lower value of 0.78 kg d−1 ADG for growing Maremmana and Aubrac steers in the same environment. Lower ADG values were recorded by other authors in Italy. In an organic farm located in northern Italy, Cozzi et al. (2010) showed that Limousine heifers under rotational grazing management had a ADG of 0.74 kg d−1 in summer. Grass-fed local breeds in Italian Alpine showed lower values of ADG during the finishing period ranging from 0.30 to 0.37 kg d−1 (Kreuzer et al. 2021). Difante et al. (2009) found an ADG of 0.66–0.80 kg d−1 for Nellore bulls reared on Panicum maximum grassland in Brazil. McCaughey et al. (1997) found an ADG of 1.26–1.29 kg d−1 for steers in rotational grazing systems; da Frota et al. (2017) found an ADG of 0.14–0.19 kg d−1 in a dry period, while an ADG of 0.96–1.15 kg d−1 in a wet period.

Since pre-grazing herbage mass, herbage allowance, potential herbage intake and pasture quality were not statistically different between treatments, differences in animal performances cannot be explained by these factors. The differences in animal performance might be explained by the different grazing management. In fact, the SP group had more than three hectares of forest to freely frequent and during the experiment the SP group was daily observe laying, standing, and walking under the canopy of trees. Therefore, we assume that the SP group had a higher energy consumption for walking. Thus, reducing the net energy for growth. The higher energy requirements incurred by the cattle having to move through the paddock more often might explain the differences in total dry matter intake.

Given the stand structure survey performed and the absence of herbaceous biomass present during the experiment period; we assumed negligible the percentage of herbaceous vegetation, trees and shrubs on the total diet. However, further investigation on the role of the forest component on the animal diet are necessary. Indeed, in this specific case we could assume as irrelevant the role of the understory vegetation in cattle diet because the area of forest available for the SP group was subjected to a high-grazing pressure in the previous months. For management reasons, the forest area has been available for cattle since December 2020. In the following months, the forest understorey available increased a little but not enough to be sampled at a height of 0.03 m, (the same height at which herbage samplings in the open pasture were performed). In March, two exclusion cages were positioned to estimate the herbage mass growth in the absence of grazing animals before the beginning of the grazing experiment. On 10 June, after the end of the grazing experiment, eights herbage samplings were performed inside the exclusion cages. An average above-ground biomass of 37 g DM m−2 was calculated.

Effects on soil cover

Kairis et al. (2015) demonstrated that a high-stocking rate reduces the plant cover due to overgrazing, while adopting good grazing management with a lower stocking rate can preserve vegetation cover. In our experiment, aerial photos processed with NDVI confirmed this trend. Overall, there was a decrease in NDVI values at the end of the grazing trial compared to the beginning. This trend is in line with expectations: given that grazing reduced herbage mass, post-grazing herbage mass was lower in each GP compared to pre-grazing (Table 4). Aerial photos and density charts reported in Fig. 2, show that at the end of our experiment, there were more NDVI values lower than zero (bare soil) in PA paddocks compared to SP paddocks.

As shown in Fig. 3: the presence of pixels with a yellow to red colour (NDVI value lower than 0) is higher in P1, P2 and P3 (PA system) at the end of the experiment compared to P4, P5, P6 (SP system). This is also confirmed by the deciles analysis, which shows that the distribution of pixels with a value lower than zero is higher in PA than SP at the end of the experiment. Prohibiting access to the forest and the higher stocking rate on the pasture led to a higher grazing pressure on the paddocks which caused: (1) pasture depletion, (2) a reduction in vegetation cover, and (3) an increase in bare soil. Maintaining the ground cover is important to reduce soil erosion and to maintain soil fertility due to the high presence of organic carbon in the topsoil (Sanderman et al. 2015). In addition, in the long-term high-stocking rate management could become detrimental, causing a reduction in herbage production, and exacerbating the effects of droughts (Fynn and O’Connor 2000).

Fig. 3
figure 3

NDVI aerial photos and pixel density distribution divided into deciles. P1 = paddock 1, P2 = paddock 2, P3 = paddock 3 are managed by an open pasture system, while P4 = paddock 4, P5 = paddock 5, P6 = paddock are managed by a silvopastoral system

Full size image

Conclusions

Optimizing animal productivity without compromising animal welfare and the environment is complex. Our study highlights the role of rainfall scarcity in limiting system productivity. Drought had a detrimental effect on pasture production and quality. This confirms the current trends caused by climate change in the Mediterranean.

The low level of pre-grazing herbage mass highlights the poor productivity, underlining the need to improve forage production by enhancing soil fertility and selecting the best management practices. High-stocking rotational grazing led to a better animal performance. However, to benefit from rotational stocking, the grazing timing and the synchronization with the phenological stage of vegetation and climate conditions are essential.

In our study, pasture regrowth was limited by water scarcity, and the animals did not exploit the full potential of herbage production due to the faster progression of the phenological cycle. As a consequence, the pasture availability and grazing period throughout the year were shorter. Knowledge of the potential herbage intake enabled us to understand how system resources (pasture, forest understorey and supplements) are used by cattle. In our experiment, the forest understorey was not available in spring, while herbage intake was limited by the low level of forage allowance.

Consequently, supplements were essential to satisfy the energy and protein requirements, considering also that the pasture quality was scarce. Supplement consumption increased with the reduction in forage allowance and potential herbage intake, together with the increase of animal body weight. A possible dependency on supplements was noticed in the SP grazing systems because of the higher energy requirements of the animals probably caused by the larger surface available for walking and browsing. Excluding animals from the forest and using rotational grazing with s high stocking rate during the spring season can lead to a better animal performance, and at the same time it reduces the consumption of supplements without an increase in animal stress.

The hair cortisol concentration is a more valuable indicator of animal welfare compared to the serum cortisol level. We detected a reduction in hair cortisol during the trial and therefore can affirm that grazing under high stock density did not lead to stress for the cattle in spring. Although rotational grazing in the open pasture system led to a better animal performance than the SP system, a high stoking-rate could be detrimental to soil characteristics and pasture productivity, as demonstrated from the aerial pictures.

In conclusion, in agrosilvopastoral systems when microclimatic conditions do not create a stressful environment for grazing cattle, and where pasture and fodder availability are not a limiting factor, a high-stocking rate and rotational grazing in pastureland can improve animal performance and energy conversion without compromising animal welfare. Further studies are necessary to understand the role of forests and grazing management in other seasons than spring in extensive agrosilvopastoral Mediterranean systems.