The study was conducted during four years (2011-2014) at Cueva Palacios mountain research station, located in the natural park of Las Ubiñas-La Mesa, Quirós, south-central Asturias (43º03'N, 5º56'W), at 1600–1800 m above sea level. The experimental field comprised 36 ha composed of 70% herbaceous and 30% shrubland communities. The herbaceous pastures mostly consisted of mesophilous grasslands domina­ted by red fescue ( Festuca rubra L., a fine-leaved grass), with a lower presence of other characteristic grasses such as Agrostis capillaris L. and Nardus stricta L. There were small areas of hygrophilous fens dominated by Carex spp., and calcareous rocky pastures with low plant cover. The shrublands mostly consisted of dwarf heathlands dominated by heather ( Calluna vulgaris (L.) Hull), with bilberry ( Vaccinium myrtillus L.) as the most abundant companion species. Other shrublands included broom ( Genista florida L. subsp. polygalaephylla (Brot.) Cout.) formations on relatively deeper soils, in addition to small areas of furze ( Genista occidentalis (Rouy) H. J. Coste) and barberry ( Berberis vulgaris L. subsp. cantabrica Rivas Mart., T. E. Díaz, Fern. Prieto, Loidi & Penas) formations on calcareous soils.

A total of 42 yearling steers from the two Asturian local breeds were utilized (5, 6, 5 and 4 AV, and 6, 7, 5 and 4 AM in 2011, 2012, 2013 and 2014, respectively). All animals were born during winter-spring of the preceding year (from 2010 to 2013). As suckling calves, they spent the summer at the research station with their mothers, so they had previous experience on these summer pastures. Calves were weaned when returned to lowlands in October. When aged around one year, in January-February, they were castrated by surgical removal of the testicles. Previously, animals were sedated with 0.5 mL/100 kg of body weight (BW) of 2% xylazine and locally anesthetized with 60 mL/animal of lidocaine.

Experimental grazing season at summer pastures lasted from mid-June to late September or early October. Mean BW ( ± SE) at the start was 372 ± 7.9 kg for AV and 307 ± 11.2 kg for AM, with a mean age of 481 ± 7.7 days. At the experimental field, yearling steers grazed accompanied by other cattle in a herd of 19–24 cows (approximately three quarters of them suckling their calves) plus heifers. Applying the livestock unit (LU) coefficients of the European Union regulation (EC, 2009), managed stocking rate averaged over the four study years was 0.86 ± 0.054 LU/ha.

Before being moved to the mountain, animals were vaccinated against Clostridium spp. (Miloxan-Merial, Lyon, France), and drenched with deltametrine (Deltavex pour-on-S.P. Veterinaria, Tarragona, Spain). All procedures were carried out in accordance with European Union regulation on the protection of animals used for scientific purposes (EC, 2010).

Botanical composition of the main plant commu­nities was sampled in late June of 2011, 2013 and 2014 recording 1000 vertical hits at 200 random loci in grasslands and 500 hits at 100 random loci in heathlands using a point quadrat (Grant, 1981).

The availability of preferred pasture was assessed by monthly measuring the sward height at 300–500 random sites in the grassland areas with a sward stick (Barthram, 1986).

Samples of the main plant components ( i.e. herbage foliage from grassland, and current season's green shoots of heather and bilberry) were collected in July and September to assess their nutritive quality. Chemical determinations included organic matter (OM, no. 942.05) and nitrogen (N, no. 990.03) following the procedures of the Association of Official Analytical Chemists (AOAC, 2006). Crude protein (CP) was calculated as N×6.25. Neutral detergent fibre (NDF) was analysed according to the procedures described by Mertens (2002). Acid detergent fibre (ADF, no. 973.18) was analysed following the methods of AOAC (2006), and expressed inclusive of residual ash. Lignin (sa) was determined by solubilising cellulose with sulphuric acid (Robertson & Van Soest, 1981).

Grazing behaviour (grazing time and plant commu­nity selection) was monitored in mid-July and mid-September of 2013 and 2014 by visual observations of the steers every 15 minutes during daylight hours (15 in July, 13 in September) in two consecutive days (Gary et al., 1970). The grazing activity of each individual at the different plant communities (grasslands, fens, rocky pastures, heathlands, broomfields, furze shrublands and barberry formations) was recorded.

The day after the visual recordings, individual grab samples of faeces were collected to estimate steers' diet composition using n-alkanes as faecal markers (Dove & Mayes, 1991). After being stored at –20ºC, faecal and plant (the same also used for nutritive qua­lity analyses) samples were freeze-dried and milled to 1 mm before chemical analyses. n-Alkanes (from C ₂₁ to C ₃₆) were extracted using the method described by Dove & Mayes (2006), and their concentrations quantified by gas chromatography, using a Perkin Helmer Clarus 580 equipped with a flame ionisation detector and an auto-sampler. Identification was performed by comparison of retention times of the samples components with the previously injected mixture of known n-alkanes. The response factors for the individual n-alkanes were calculated from peak areas and the known concentrations. The n-alkane concentrations were quan­tified relative to known amounts of the internal standards C ₂₂ ( n-docosane) and C ₃₄ ( n-tetratriacontane) added at the beginning of the extraction process.

Faecal n-alkane concentrations were corrected for their incomplete recovery in faeces using data from validation studies with cows (Ferreira et al., 2007). Diet composition (proportions of herbage, heather and bilberry) was estimated using an iterative least-squares procedure, which minimizes the discrepancies between the observed faecal concentrations of each n-alkane (C ₂₅–C ₂₇ and C ₂₉–C ₃₃) and the estimated dietary proportions of plant components (Dove & Moore, 1995) using the Solver routine in Microsoft Excel. The other extracted alkanes were not used due to their low concentrations in both plant and faecal samples.

Steers were weighed at the beginning, middle (early August) and at the end of the grazing season. Daily BW changes were calculated for the different periods ( i.e. from June to August, and from August to September-October) and for the whole grazing season.

Data on plant chemical composition (OM, CP, NDF, ADF and lignin) were analysed by a General Linear Model (GLM) procedure to examine differences among the plant components, season (July and September) and their interaction. Steers' individual data on grazing behaviour and diet composition were subjected to a GLM for repeated measures, including the fixed effects of breed, year, season (two repeated measures) and the full interactions. For non-grazed communities in one of the two seasons, that season with no variance was removed from the model to test the effects of breed, year and the interaction breed × year. Daily BW changes during the different periods (the two halves and the whole grazing season) were subjected to a GLM to examine the effects of breed, year and their in­terac­tion. Tukey's test was used for multiple comparisons among means. Correlations between BW changes (averaged for each breed) and mean sward height at each grazing period (the first and second half of the four grazing seasons; n=8) were checked with Pearson's correlation coefficient. All analyses were performed using Statistica 8.0 (StatSoft Inc., Tulsa, OK, USA).

Grasslands presented a highly diverse botanical composition, with a total of 79 vascular species recor­ded. Although grasses predominated (54.6% cover on average, with red fescue as the most abundant), many other species were recorded, including other monocots (9.3% cover) such as Merendera montana (L.) Lan­ge and Carex caryophyllea Latourr., legumes (8.5% cover) such as Trifolium repens L., Lotus corniculatus L. and Vicia pyrenaica Pourr., and other dicots (25.4% cover) such as Plantago media L., Plantago alpina L. and Pilosella officinarum F. W. Schultz & Sch. Bip. to quote the most abundant ones. Heather dominated in heathlands with 44.8% mean cover, followed by bilberry (9.0%), with herbaceous species accounting for 31.6% cover ( Potentilla erecta (L.) Raeusch. and Succisa pratensis Moench were the most abundant). Other shrub species such as Chamaespartium sagittale (L.) P. E . Gibbs and Juniperus communis L. subsp. nana (Willd.) Syme were sparsely found (3.9% cover in total).

The grassland sward height (means ± SE of four years) decreased from 14.4 ± 0.74 cm at the start to 3.3 ± 0.06 cm at the end of the grazing season. Avera­ged over time, mean sward heights were 7.8 ± 0.22 cm and 4.0 ± 0.20 cm during the first and second half of the grazing season, respectively, averaging 6.1 ± 0.11 cm during the whole grazing season.

Grassland herbage had greater CP (178 vs. 98 g/kg DM) and NDF contents (526 vs. 394 g/kg DM), and lower OM (930 vs. 962 g/kg DM) and lignin contents (34 vs. 141 g/kg DM) than shrubs, with ADF contents being greater in heather than in herbage and bilberry (319 vs. 239 g/kg DM; Table 1). Heather had greater OM and lignin contents, and lower CP contents than bilberry. Mean CP content in herbage increased from July to September, whereas it decreased in heather and bilberry ( p<0.01 for component × season interaction; Table 1).

Daily grazing time was similar for the two breeds and increased from July to September (from 488 to 557 min/day; SEM=11.2; p<0.001; Table 2). There were differences between years, with steers grazing for longer in 2013 than in 2014 (547 vs. 498 min/day; SEM=13.8; p<0.05). Grazing activity was concentra­ted from sunrise to midday (7:00–12:00 h) and again in the evening (18:00–21:00 h). Steers preferred to rest during the central hours of the day, especially from 13:00 to 15:00 h, although some differences between seasons were observed. The steers from both breeds showed more breaks in their grazing activity (both in the morning and in the afternoon) in July than in September (Fig. 1).

In general, AV steers grazed proportionally for longer times on herbaceous pastures (81.1% vs. 73.4%; SEM=1.98; p<0.05) and less on shrubby communities than AM steers (Table 2). The grazing time on herbaceous relative to shrub communities tended to decrease from July to September (from 80.1% to 74.4%; SEM=2.33; p=0.091). In addition, steers in 2014 tended to spend more proportional time grazing on herbaceous communities than those managed in 2013 ( p=0.054), especially in the case of AM breed ( p=0.085 for breed × year interaction). The dominant grasslands were more used by AV than AM steers (75.0% vs. 66.5%; SEM=2.02; p<0.05), whereas no differences between breeds were found in grazing time on other herbaceous communities with minor extension. In general, grazing time on both fens and rocky pastures decreased from July to September (from 3.5% and 5.8% to 1.2% and 2.4%, respectively; p<0.001), but with great differences between years ( p<0.01 for year × season interaction).

Regarding shrubby communities, AM steers grazed for longer on heathlands than AV steers (21.5% vs. 12.5%; SEM=1.80; p<0.01; Table 2), especially during the morning meals (Fig. 1). For AV steers, grazing time on heathlands increased from July to September, but not for AM steers ( p=0.055 for breed × season interaction). Heathlands were grazed proportionally for more time in 2014 than in 2013 (21.2% vs. 12.8%; SEM=1.90; p<0.01). There were no differences between breeds in the grazing time on broom scrublands, but it was longer in 2014 (6.6%) than in 2013 (3.6%; p<0.05). Noticeably, steers were not observed browsing the small-leaved shoots of brooms; instead, they were found only to graze below on the herbaceous or dwarf-shrub layer. Furze and barberry communities were only utilized in September with minimal grazing times (0.7% and 0.6%, respectively; Table 2).

Diet composition did not differ between breeds (Table 3). Steers selected greater percentages of herbage (80.0 ± 2.23%) than shrubs (15.5 ± 1.58% bilberry, 4.4 ± 0.89% heather). Herbage percentage in diet decreased from July to September (from 88.8% to 70.8%; SEM=2.84; p<0.001), whereas those of shrubs increased (bilberry from 9.7% to 21.5%; SEM=1.74; p<0.001; heather from 1.5% to 7.7%; SEM=1.46; p<0.01). There were no differences between years in dietary percentages of herbage or bilberry, while heather tended to account for greater percentages in 2014 than in 2013 (6.1% vs. 3.1%; SEM=1.11; p=0.068).

Overall, AM steers attained greater daily BW gains du­ring the whole summer than AV steers (252 vs. 133 g/day; SEM=36.0; p<0.05), with greater overall gains observed in 2011 compared to 2012 and 2013 (319 vs. 112 g/day; SEM=50.4; p<0.05; Table 4). During the first half of the grazing season (from June to August) AM steers gained significantly more BW than AV steers (487 vs. 360 g/day; SEM=31.9; p<0.01). From August to October between-breed differences persisted, and AM steers continued to gain BW, whereas AV steers lost BW on average (120 vs. –12 g/day; SEM=45.5; p<0.05). There were differences across years in BW gains in both periods. From June to August greater gains were observed in 2012 and 2013 compared to 2011 (505 vs. 292 g/day; SEM=44.7; p<0.01), but thereafter the gains were greater in 2011 than in the following years (347 vs. –43 g/day; SEM=71.3; p<0.001). Thus, BW gains were reduced from the first to the second half of the grazing season except in 2011 (Table 4). No significant interactions between breed and year were detected in any period.

Correlation analyses showed significant ( p<0.01) positive relationships between grassland mean sward height during a particular season and steers' mean BW gains for both breeds ( r=0.838 for AV, r=0.877 for AM).

No differences in total grazing time and diet selection were found between AV and AM steers in these summer pastures. In general, scarce differences in grazing behaviour have been observed between similar cattle breeds ( e.g. Funston et al., 1991; Dumont et al., 2007; et al., 2009b). Some studies comparing breeds differing in mature body size, origin or aptitude did find differences in grazing activity or selected diet ( e.g. Erlinger et al., 1990; Winder et al., 1996; Hessle et al., 2008). Much of the between-breed differences in foraging behaviour can be associated to differences in body size and related intake capacity, nutrient requirements and feeding efficiency for maintenance or growth (Rook et al., 2004; Dumont et al., 2007). In the current study AM steers grazed proportionally for longer time on heathlands and less on grasslands than AV steers. As heathlands generally occupied more rugged and steeper terrains than grasslands, the larger AV steers could be more constrained to move on and utilize those shrublands than the smaller AM steers, which are recognized to be more rustic and able to climb better than AV (Cañón et al., 1994; Osoro et al., 1999). Thus, AM steers could favour a more homogeneous forage use of mountain pasture resources (Bailey et al., 1998).

Nevertheless, such differences in plant community selection were not reflected in differences in diet composition. It could be due to the fact that, when grazing on heathlands, especially at early summer (July), AM steers selected mostly herbaceous species growing among the dominant shrubs. Particularly in open heathland areas, steers could be selecting the narrow grassland strips as we did not record the particular plant items eaten at each feeding site, but the location. Open heathlands are much more utilized by cattle than close heathlands (Mandaluniz et al., 2011).

Steers preferentially selected grassland herbage against woody species of lower nutritive quality. Cattle are well known to graze preferentially on grasslands over heathlands (Gordon, 1989; Mandaluniz et al., 2011; Ferreira et al., 2013). In September, how­ever, once grassland sward height was much shorter (3–4 cm), shrub percentage in diet increased, as previously observed in different studies with cows (Putman et al., 1987; Celaya et al., 2008). Bilberry was more selected than heather, similar to what was found by Grant et al. (1987) in Scottish heathlands. However, Sæther et al. (2006) observed low percentages of bilberry and similar to those of heather (around 2%) in cattle faeces by microhistological analyses, as most of the faecal fragments corresponded to herbaceous species (76% grasses, 8% sedges and 8% herbs).

On the other hand, the lack of observations of steers browsing on brooms agrees with previous findings on the cattle reluctance to consume this woody legume (Osoro et al., 2000). Broom formations have proliferated significantly over grassland and heathland areas since the small ruminants, mostly coming from transhumant herds, have disappeared for some decades in these mountains (González Díaz et al., 2019). Thus, the current replacement of sheep and goats by cattle may be ineffective to control broom encroachment, threatening the provision of quality pasture for livestock, the traditional pastoral landscape and its biodiversity. Nowadays, the removal of these shrublands has to be undertaken by costly mechanical clearings.

The diurnal grazing pattern shown by steers, with two main grazing periods in the morning and in the evening separated by a more or less marked break at noon, has been described in several studies with cattle foraging on mountain pastures (Funston et al., 1991; Aldezabal et al., 1999; Ferreira et al., 2013). Nevertheless, cattle show flexible circadian grazing patterns depending on environmental and pasture conditions, and may have between three and five grazing periods during a day depending on season (Erlinger et al., 1990; Linnane et al., 2001; Celaya et al., 2008). In the current study, the length of the meals increased from July to September, which resulted in longer daily grazing times late in the season, even though steers' activity was recorded for two hours less than in July (Fig. 1) according to the shortening of sunlight time. Similar results were obtained in heifers of traditional and commercial breeds grazing in Swedish semi-natural grasslands (Hessle et al., 2008). Regardless of some possible grazing activity at night (Linnane et al., 2001), the observed increase in daily grazing time from July to September would be mainly related to the reduced availability of preferred grassland herbage, thus forcing the steers to graze for longer in an attempt to compensate the reduced intake rate (Hodgson, 1990; Ferrer Cazcarra et al., 1995).

In spite of the lower growth potential, AM steers achieved greater BW gains than AV ones grazing on these summer pastures. Differences were observed in the two periods considered, i.e. regardless of the preferred herbage availability (mean grassland sward heights of 7.2–8.3 cm from June to August, and 3.7–4.6 cm thereafter). On the contrary, comparing the same animals when suckling their dams during the preceding year on the same summer pastures, AV calves showed greater BW gains than their AM counterparts (Román-Trufero et al., 2015). Thus, as yearlings, AV steers could not fully express their growth potential even with relatively high availabilities of palatable herbage of acceptable nutritive quality. Other factors like rugged terrain and increased energy costs could be affecting more AV steers compared to AM breed (Morris & Wilton, 1976).

With the exception of the first study year (2011), BW gains were reduced from the first to the second half of the grazing season. The nutritive quality of the main forages hardly changed, so the lowered performances from August onwards were primarily due to the reduced herbage availability, which limited intake. This is supported by the positive correlations found between grassland sward height and steers' BW gain for both breeds. The positive relationship between grassland sward height and cattle performance has been previously observed ( e.g. Wright & Whyte, 1989; Morris et al., 1993; Realini et al., 1999). Even the exception of 2011 could be explained by the lowest sward height observed from June to August in that year (7.2 cm), reverted to be the highest from August to late September (4.6 cm) among the four study years, thereby implying a less severe restriction in herbage intake during the second half of the grazing season compared to the following years.

The lower performance of AV steers during the second half of the grazing season (with BW losses in three of the four years) indicates their worse suitability to graze short pastures than AM steers. This supports the hypothesis that smaller animal species, breeds or individuals can thrive better under conditions of scarce food availability because of their higher ability to graze on short swards and lower total energy demands (Morris & Wilton, 1976; Illius & Gordon, 1987; Wright et al., 1994). Previous studies in the same and nearby summer pastures as in the present study observed that AM cows had more favourable BW changes than AV cows when grazing on short grasslands with mean sward heights below 4 cm (Osoro et al., 1999). However, other studies in mountain pastures did not find significant differences between cattle breeds in animal performance, whereas pasture type had a great influence (Casasús et al., 2002; Fraser et al., 2009a). These contrasting results depend on the particular breeds (namely the extent of between-breed differences in adult body size) and the type of pasture studied.

Regardless of between-breed differences, steer productive performance in these summer pastures was relatively low compared to what can be achieved in lowland grasslands with milder climate (unpublished data). The great interannual variability observed reflects the dependence of this type of animal production on climatic factors that affect animal welfare and performance directly or indirectly through available pasture characteristics. However, it is noteworthy that the steers from the current study showed a compensatory growth once they returned to lowlands, achieving greater BW gains and similar final yields at slaughter compared to steers managed at lowlands (Román-Trufero et al., 2015).

In conclusion, summer pastures could be utilized to produce steers in a sustainable way, provided that their BW gains are not too much reduced late in the season due to the usual shortage of quality herbage. Yearling steers could be returned to lowlands a few weeks before ( e.g. towards the beginning of September depending on weather and pasture conditions) to alleviate the slowdown of their growth. Yearling steers from AM breed showed a better ability to graze heterogeneous Cantabrian summer pastures than their AV counterparts as a result of their smaller body size and greater rusticity. Besides the higher performance, AM steers could favour a more homogeneous utilization of available pasture resources.