Serpentinites are an uncommon lithology that occurs in very specific geodynamic environments, typically defining alignments, and exhibiting a typical greenish colour. Serpentine soils are found worldwide but have patchy distribution, covering less than 1% of the total land surface [1]. In Portugal, serpentinites are known to occur in the Bragança-Morais massif and Beja-Acebuches ophiolitic complex [2].

Soils that develop on this parent material are deficient in some macronutrients, show imbalances in nutrient elements (low Ca : Mg ratios) and display toxic concentrations of metals such as chromium (Cr) and nickel (Ni) [3]. Vegetation grown on these soils may accumulate high levels of metals, thereby causing potential health hazards for grazing animals and humans [4, 5].

Vegetation and serpentine soils have been studied extensively [6-8]. Outcomes have revealed high concentrations of potentially biologically toxic elements such as Cr and Ni, poor plant productivity and specific kinds of vegetation [9, 10]. Significant differences in plant metal uptake between plant species have been observed [11-14]. Also, metal accumulation in animal tissues of grazing animals raised on serpentine or other contaminated areas has been investigated [15, 16]. Cr, Ni and manganese (Mn) are essential elements for life and are involved in different metabolic processes [16, 17]; however, they may be toxic in excess.

Grazers are exposed to these metals via ingestion of contaminated vegetation and small amounts of soil, and in some cases also via drinking water [18]. López-Alonso et al. [19] pointed out that toxic metal accumulation in animal tissues is directly related to grazing activity, and toxic metal exposure through soil ingestion could be higher than the one through the diet. On the other hand, the choice of food plants may in some cases affect the metal exposure of herbivorous animals more than that reflected by the general contamination level in the area [20, 21].

In the north of Portugal, the small ruminant production system is based on the exploitation of rangeland resources and by-products of agriculture. In this region, small ruminants are produced by landless farmers or smallholders. In this way, the livestock system is supported by grazing itineraries over unfenced or uncultivated village territory [22]. Flocks are conducted by shepherds, crossing land with different uses [23]. Grazing circuits and resources consumed differ strongly throughout the year and within a territory. For this reason, diet composition, as well as animals’ exposure to metal, is variable.

No studies have been performed on how habitat use affects metal concentrations to which sheep and goats reared in serpentine soil areas are exposed. In order to get knowledge on this subject, in the present work, six herds (three of sheep and three of goats) were followed to find out their grazing circuits and diets. At the same time, soil and vegetation samples were collected, and the metal concentrations (Ca, Mg, Mn, Cr and Ni) in them determined in order to evaluate the animals’ metal exposure. In more detail, we want to find out which plant species and grazing circuits are more problematic to animals in terms of heavy metals, namely chromium and nickel.

Diet composition varied with the availability of plant species in the habitat (affected by the grazing circuit and season) and with animal species, explaining the different resources consumed by each flock. For example, the grazing circuits of both Flocks 1 (goat) and 6 (sheep) crossed the Morais ultramafic massif but their diet composition differed. It is known that sheep and goats consume different proportions of herbaceous and ligneous species. The former feed more often on herbaceous species than goats. This feeding strategy has been observed in many studies performed in different plant communities [23, 32, 33].

Regarding metal concentrations, a high degree of variability was observed between grazing circuits. Despite all soils coming from a serpentine area, soil composition varied widely between sites; it might depend upon the geological material, as suggested by Lázaro et al. [11]. The highest Mg, Cr and Ni content was found in soils collected from the circuits of the flocks that graze around the Morais ultramafic complex (Flocks 1 and 6). This type of soil is characterized by high concentrations of metals such as Mg, Mn, Cr and Ni [3], explaining the results obtained. Soils were characterized by low Ca/Mg ratios (medians < 1), Ca/Mg ratios less than 1 being characteristic of strongly serpentinized soils [9]. Moreover, the Ca and Mg concentrations determined in the present work were in line with the concentrations determined by Shallari et al. [34] in serpentine soils in Albania (2 to 24 mg/g of Ca and 15 to 25 mg/g of Mg). Regarding Mn, almost all concentrations determined in the present work were within the ranges of 5.5-670 and 660-800 µg/g reported for managed serpentine soils collected in the north-west of Spain [4] and in Iran [12], respectively. Concerning Cr occurring naturally in soil, this metal ranges from 10 to 50 µg/g, depending on the parental material [35]. In serpentine soils, Cr can reach up to 125 mg/g [35]. Nevertheless, the Cr concentrations found in our study were lower than this value and the range between 365 and 3865 µg/g reported by Shallari et al. [34] for serpentine soils of Albania. On the contrary, the Cr concentrations determined in Morais soils were higher than the range reported by Fernández et al. [4] (0.1-0.9 µg/g), and similar to that reported by Miranda et al. [16] (10.0 to 1162 µg/g) for serpentine soils in the north-west of Spain. Regarding Ni, the levels determined in soils from the circuits of Flocks 2 to 5 were within the range reported by Fernández et al. [4], 5-110 µg/g, for managed serpentine soils of north-west Spain. Nevertheless, much higher Ni concentrations were found in soils from the circuits of Flocks 1 and 6 because the shepherds took their animals to graze around the Morais massif (Fig. 1), an area where ultrabasic rocks predominate and high Ni concentrations are found. The Ni soil medians determined for these two flocks (830 and 934 µg/g) were similar to the range reported by Ghaderian et al. [12] (800-1240 µg/g), to the maximum of 940 µg/g indicated by Miranda et al. [16] and to the median of 1329 µg/g found by Kierczak et al. [36] determined in Iranian, Spanish and Polish serpentine soils, respectively. In addition, the soils collected on the Morais ultrabasic complex had Ni concentrations above the recommended threshold for agricultural soils (30 µg/g) [16], in line with the land use of this territory which is mainly covered by forest and shrub lands. Regarding pH, our pH_H2O values were within the range determined by Kataeva et al. [37], 5.45-7.46, in the Polar Urals. Concerning plants, Ca/Mg plant ratios were generally less than 1.0, as observed for soils. As reported by Reeves et al. [13], this proportion is uncommon. The Mn concentrations determined in the present work were within the ranges reported by Fernández et al. [4] (6-300 µg/g) and Lázaro et al. [11] (14.5-2200 µg/g) for plants collected in serpentine soils in the north-west of Spain and north-east of Portugal, respectively. Some plants collected from the Morais site had a Mn content above 300 µg/g, the threshold toxicity level in plants [4]. Almost all Cr plant concentrations determined in the present study were in close agreement with those of Fernández et al. [4] (0.8-11.5 µg/g), Miranda et al. [16] (0.39 to 7.11 µg/g) and Oze et al. [10] (0.003 to 5.76 µg/g) for vegetation collected in several serpentine soils. It is worth mentioning that our low Cr values found in the aerial parts of the plants could be mainly due to Cr accumulating more in roots than in leaves, as Cr is preferentially immobilized, suppressed and/or sequestered into the roots and/or woody tissue [10]. Cr is sequestered in the vacuoles of root cells, thus rendering it less toxic, which may be a natural toxicity response of the plant [35]. The soil Cr concentrations of Flock 5 were much lower than those of Flocks 1 and 6 (Table 2); however, the highest medians of plant Cr were found in Flock 5 (1.59 µg/g) and Flock 6 (1.50 µg/g), suggesting that the low pH values of Flock 5 soils could have accounted for a high soil-to-plant transfer of metals. Nevertheless, the highest Cr concentrations (27.8 and 18.6 µg/g) were determined in sorghum and meadow collected from the grazing areas of Flocks 1 and 6. Regarding Ni, the plants analysed in the present work presented content within the range reported by Ghaderian et al. [12] (55-90 µg/g), Fernández et al. [4] (3-35 µg/g), Miranda et al. [16] (11.1-39.3 µg/g), Reeves et al. [13] (5-275 µg/g) and Oze et al. [10] (0.13-24.4 µg/g) for plants growing in serpentine soils in Iran, Spain, Costa Rica and the USA, respectively. The highest medians were determined for Flocks 1 and 6 and were above the threshold level for toxicity in plants (10 µg/g) [4]. The plants with Ni concentrations higher than 10 µg/g (Table 5) were sorghum, holm oak, C. ladanifer and C. multiflorus for Flock 1, and meadow, holm oak, herbaceous sward, C. multiflorus and E. scoparia for Flock 6, all collected in the Morais ultrabasic complex. According to Kataeva et al. [37] and Baker et al. [38], Ni hyperaccumulation implies values higher than 1000 µg/g, so apparently no Ni hyperaccumulators were foraged by the animals. However, as meadows and herbaceous swards are composed of several species, a more refined technique for better understanding animal’s diet composition must be applied in the future in order to identify the existence or not of hyperaccumulators among this resource. On the other hand, some works have reported the existence of an endemic Ni hyperaccumulator, Alyssum pintodasilvae [8], but this plant was not among the three most consumed by the animals.

The high correlations between Ni and Cr, and Ni and Mg concentrations in soils were expected because it is known that ultrabasic soils are rich in these metals. The good correlations found between Mg and Cr, Mn and Cr, and Mn and Ni in soils are in line with those observed by Moraetis et al. [39] in serpentine soils in Greece where correlation coefficients of 0.68 and 0.53 were obtained for Cr-Ni and Mg-Cr, respectively. A lack of correlation between Cr concentrations in plants and soils was also reported by Fernández et al. [4] and Miranda et al. [16] for serpentine soils in the north-west of Spain.

The metal content of the grazing areas of Flocks 1 and 6, as well as of Flock 3, was quite well separated from the others. Three clusters were identified taking into account the metal concentrations of soils and/or plants of grazing areas. Flocks 1 and 6 form a homogeneous group related to high Ni, Mg and Cr soil concentrations associated with the Morais ultrabasic complex. Moreover, the highest metal concentrations in plants were also determined in those flocks. On the contrary, another cluster composed of Flocks 2, 4 and 5 had low Ni and Mg soil and plant concentrations, related to agricultural areas. The grazing area of Flock 3 formed another group because high Mn concentrations in soils and plants, as well as high Ca concentrations in plants, were determined occasionally. These results show that grazing circuits and the plants selected by the animals significantly influence the metal concentrations to which the animals are exposed.

The choice of grazing circuit by shepherds and the selection of the plants by animals influence the metal concentrations to which they are exposed. Those flocks that graze near the ultramafic complex (e.g. Flocks 1 and 6) are subjected to high Ni and Cr concentrations by soils and plants. In some areas, Ni concentrations in soils were above the recommended threshold for agricultural activity. Some plants grown in serpentine soils, such as sorghum, holm oak, C. multiflorus, C. ladanifer and E. scoparia, showed a Ni content above the threshold level for toxicity. Further studies are required to determine metal uptake by animals in order to establish policies for the management of flocks in such areas.