Plant production and nutritive quality of savannah-like grasslands (dehesas) in semi-arid zones of the province of Salamanca

A study was made of two savannah-like (dehesa) pastures on which cattle (Morucha cows × Charolais bulls) are raised for beef. The soil of dehesa system A was developed over slate, while that of B was developed over granite. The type of vegetation, dry matter yield (DM), raw protein (RP) content, digestible organic matter (DOM) content and the metabolizable energy (ME) of each system was determined to provide the nutritive value of the two types of pasture. Dietary supplements were assessed in the same way. The DM yields for 1989, 1990 and 1991 were 751, 1900 and 1438 for system A, and 577, 1937 and 1361 kg ha for system B. The RP values for April and December were 17.6 ± 1.4%, falling to a minimum of 6.7 ± 0.4% in July-October. Intermediate values were seen for the rest of the year. The ME showed mean values of 9.4 ± 0.3 and 7.9 ± 0.2 MJ kg DM for systems A and B respectively, while mean metabolicity was 0.50 ± 0.2 and 0.44 ± 0.1 (P < 0.05) and the DOM was 67.5 ± 4.2% and 44.3 ± 3.9% respectively. These values were affected by the year and harvest date (P < 0.05) and were always greater (P < 0.05) on soils developed over slates. The greatest pasture ME was achieved in spring and represented some 77-95% of total pasture production, depending on the year. Supplements represented approximately 20% of the total available ME.

To maintain their live weight, adult ruminants require forage with a digestibility of 45%, if greater, enough energy is available for fattening or for work (Ward and De Gracia, 1986). The metabolizable energy (ME) is the fraction of the food that animals actually make use of (McDonald et al., 1986). Bellido et al. (1986) established the energetic balance for dehesas of southwestern Spain, taking into account the energy provided by pasture and supplementation, the nutritional needs of the animals, and the production of meat.
The aim of the present work was to characterise two dehesa systems representative of the Province of Salamanca (central-western Spain). This involved estimating the primary production of pasture and acorns and the botanical composition of the available vegetation over three years, as well as determining the coefficients of digestibility of the organic material, its raw protein content (RP), the ME of the pasture and supplements, and, finally, the balance between available energy and animal needs.

Edapho-climatic characteristics of the two dehesa systems
The experiment took place between the spring of 1989 and the autumn of 1991. Two representative dehesa systems were chosen, both developed over Tertiary beds, system A, an area with slate soils in the municipality of San Pedro de Rozados, and system B, an area with granite soils in the municipality of Villasdardo. These soil types occupy some 80% of the stock raising land in the Province of Salamanca. They are both shallow (30-40 cm), sometimes with rocky outcrops, and of low waterretaining capacity (except in low-lying areas with loam or sandy loam soils) (García Rodríguez, 1966).
Aerial photography was used to estimate the total surface area available for grazing in the two dehesas, i.e., the total surface area minus the unproductive area. For system A, the grazing area was 489 ha, distributed over nine enclosures with high, middle and low ground and with a slope of 14-74%. System B had a grazing area of 358 ha, distributed over six enclosures with high and low ground and a gradient of < 7%. The mean altitude of the greater study area was 800 m.
Surface horizon soil samples (0-25 cm) were taken for chemical and mechanical analyses. For each slope position (high, middle and low), 100 original samples were taken. These were then divided into 10 groups of 10 samples each, the contents of which were mixed for analysis.
The soils on which the dehesas are developed were acidic with high organic matter (OM) and low calcium contents. These contents increased from the high toward the low-lying ground, OM from 2.7 to 4.3% and Ca from 355.0 to 1060 mg kg -1 on the slate soil of system A, and from 2.8 to 3.5% and 40.0 to 356.0 mg kg -1 respectively on the granite soil of system B. P and K, however, were found in greater amounts on the higher ground. Mean P values were 5.0 mg kg -1 on high ground and 3.6 mg kg -1 on low ground for the slate soil of system A, and 7.6 mg kg -1 on the high ground and 4.1 mg kg -1 on the low ground for the granite soils of system B. Similarly, mean K values were 192.0 mg kg -1 on the high ground and 138.9 mg kg -1 on the low ground of system A, and 142.6 mg kg -1 on the high ground and 87.7 mg kg -1 on the low ground of system B. The P content of all these soils was therefore deficient, while K was sufficient in the high areas but deficient in the low areas. This agrees with the results of Martín Polo and Prat (1983).
Rainfall data were provided by the Centro Meteorológico Territorial de Castilla y León and collected at the meteorological stations closest to the two dehesa systems. Temperature data, collected at the Muñobela meteorological station (in the municipality of Barbadillo) were common to both systems.
During the study period, records were made of seasonal and annual rainfall and minimum temperatures. The mean rainfall for winter, spring, summer and autumn was 117.3, 104.8, 52.1 and 194.2 mm in system A, and 115.0, 102.1, 36.2 and 205.2 mm in system B. The mean rainfall of the last 30 years for system A was 160.3, 136.9, 61.5 and 167.7 mm for the same seasons, while that of B for the last 26 years was 151.6, 122.9, 51.0 and 163.5 mm respectively. The mean seasonal minimum temperatures (the same for both systems) were -0.2, 5.7, 12.0 and 2.9ºC respectively. The mean minimum temperatures of the last 17 years for these same four seasons were -0.67, 5.4, 10.41 and 2.44ºC. The study period was therefore warmer and drier than normal, except during the autumn.

Characterisation of the flora, primary production and main nutrients of the available pasture
The most important primary production of these systems was pasture and acorns.
An initial survey was undertaken to determine at macroscopic level the types of pasture communities present. Plant inventories were produced during the period of greatest development (end of May -beginning of June) by random sampling of the different communities. This was performed using a 0.5 × 0.5 m quadrat thrown at least 5 times to obtain samples approximately coinciding with ground level.
Exclusion cages (1 × 1.5 m) with four compartments were used to estimate the primary production of the pasture. Four readings were taken in spring (at the beginning of May, June and July) and two in autumn (the middle of October and December). Samples for the different tests were taken by harvesting quadrats (0.5 × 0.5 m, 3 cm from the ground) and measuring the growth of two successive harvests. The number of cages used depended on the size of the area examined: 58, 23 and 29 cages were placed on the high, middle and low ground areas respectively. In 1989, only two control values were made in spring since the experiment started in May.
Pasture samples were dried at 60ºC for 48 h before grinding and passing through a 1 mm sieve to provide material for further analysis. Dry matter was measured after keeping the material at 105ºC for 24 h. The OM content was determined by calcination at 460ºC overnight, raw protein (RP) was calculated from the Kjeldahl N content × 6.25, neutral detergent f ibre (NDF) was estimated according to Van Soest et al. (1978), and the digestibility of organic matter (DOM) by the neutral detergent-cellulase assay (Bassery and Campling, 1988). ME was estimated from the DOM using the equation ME = 5.406 + 0.058 DOM (Martín Polo and García Bellido, 1995).
Total ME (TME) obtained at each cutting time (harvest) was calculated by multiplying the available grazing area (in ha) by production per ha (in kg DM ha -1 ), and from the ME content of the dry matter (DM) (in MJ kg DM -1 1000 -1 ).

Acorn production
To calculate the primary production of acorns, the number of trees was estimated from aerial photographs. For an area of 489 ha and a density of 20 holm oaks per ha, Vázquez et al. (2001) recommend sampling 33 trees as well as checking 10-15 additional specimens to see whether they have acorns, and then determining the ratio of acorn producers/non-producers. In this work, acorn production was measured every year by sampling trees in each enclosure (at least one tree from the high, middle and low ground areas in system A, and one from the high and low areas in B). The numbers of producing and non-producing trees per enclosure were also estimated. Eighty one trees were sampled from system A and 54 from B over the study period. The influence of the trees on the pasture was not taken into account, nor was the nutrient supply from the litterfall.

Supplementation
Nutritional supplementation of the cattle was undertaken with alfalfa, meadow and vetch-oat hay, straw, maize, barley, oats, soy flour and cereal feed, but in very different quantities according to the year and system. The energy content of the hays was estimated using the methods described above. For the other components, the tables of McDonald et al. (1986) were used.
Daily samples of whatever was provided were taken and the weight of these products recorded every 15 days. The monthly and annual ME intake was then calculated. A 5% loss of feed, distributed in the field, was assumed (Gaillard, 1989).

Stocking rate and management
The two systems are used to raise native Morucha cattle for beef (Morucha cows × Charolais bulls). The stocking rate was 0.55 and 0.43 cows ha -1 year -1 in systems A and B respectively; the mean live weights of adult cows were 360 kg in A and 440 kg in B. Calves were weaned at six months of age.
The different enclosures were grazed in rotation, the stock raiser deciding when the pasture was used up. The higher enclosures were grazed first (since these dry out earlier), followed by the middle and then low areas. In years in which a lot of grass is produced, some lower areas are reserved for feed harvesting, but this did not occur during the experimental period.

Statistical analysis
ANOVA was used to study the influence of year, harvest time, and high, middle and low slope position on the quantity and quality of primary production.
Differences between means were analysed using the Student t test. RP, NDF and digestibility were studied separately and together by year, harvest time and slo-pe, and the relationships between these variables assessed by simple and multiple linear regression. All calculations were made using Statgraphics Plus 2.1 software for Windows 95.

Results and Discussion
Botanical composition of the pasture The plant inventories were similar to those recorded by Rico et al. (1979) over a wider area of the region. The herbaceous vegetation, in agreement with the topography, was distributed following eutrophia-oligotrophia gradients from the damper to the drier areas. Humid pasture of Agrostis castellana was found in the low-lying areas of stream beds, while pasture of Agrostis castellana of different trophic nature was detected at mid slope, and pasture containing ephemeral species was seen in the highest areas (where soils are very thin and there are outcroppings of rock). Table 1 shows the species most frequently seen (and with the highest biomass) according to their position on the slope in both systems A and B (there is no middle zone in system B). Table 2 shows the mean pasture production for the three years according to harvest time and slope position. The standard deviation shows that this production was very different from one year to another, as is to be expected in an area with such changeable climate and where rainfall and temperature are limiting factors (Garmendia, 1965).

Primary production
Maximum production was reached in spring, with mean values of 2,549 and 2,363 kg DM ha -1 (Table 2) in May in systems A and B respectively. The autumn (December) production was 732 and 644 kg DM ha -1 respectively. With respect to slope position, the lowest areas were the most productive (in low areas on more fertile soils). The annual mean production of the harvest of 1989, 1990 and 1991 was 751ª, 1,900 c and 1,438 b kg DM ha -1 for system A, and 577 a , 1,937 c and 1,361 b kg DM ha -1 for B  (the letters a, b and c represent significant differences at P < 0.05). These results were obtained taking into account the surface area and production of each height zone at each harvest. The differences between harvests and between years were significant (P<0.01). Calvo et al. (1997) obtained a mean production value for the Extremadura (southwestern Spain) dehesa of 1,200 kg DM ha -1 . For the same area as the present work, Pérez et al. (1995), recorded a production of 1,260-3,040 kg ha -1 depending on slope position and the botanical composition of the area.
Annual acorn production was estimated at 14,800, 35,250 and 30,811 kg for system A, and 32,207, 4,436 and 18,499 kg for system B in 1989, 1990 and 1991 respectively. These large fluctuations in acorn production depend mainly on the weather conditions of each year. Spring frosts have a negative effect, as do high temperatures at the end of summer. With respect to location and orientation, those areas facing south are the most productive. Clearing and pruning favour production, as does landworking for cultivation. (No data are actually available for the region; all the above are personal observations.) Table 3 shows the mean chemical values of DM, MO, RP, NDF and DOM per harvest for the three years of the experiment. An increase was seen in DM from 22-86% between April and October due to the phenological development of the plants in both systems, as well as to slope position. RP fell from April (16.7%) to October (6.6%) but recovered in December (18.8%). Significant differences were seen between the harvests of each year (P < 0.05), but not between years for the same harvest, except for that of December 1990.

Chemical composition of the biomass produced
NDF increased as the plants matured, with significant differences between months (P < 0.05), but not between systems nor with respect to slope position. For example, the mean NDF for the month of April was 43.7 and 50.9% in systems A and B respectively, while in October these f igures were 73.9 and 74.8%. In June and July of 1989, the NDF was greater (P < 0.05) than in other years.
Digestibility fell with the maturation of the plants (P < 0.05) from 73.0 and 62.5% in April to 38.8 and 40.6% in October in systems A and B respectively. In December it recovered again (60.5 and 60.2%). Neither year nor slope position affected this, although the harvest date did have an influence (P < 0.05). System A produced a better quality product than system B (P<0.05).
The DOM coincided with the mean annual value (55.5%) estimated by Bellido et al. (1986) for dehesa pasture.
In May and June in the present study area, Rico et al. (1986) obtained RP of 11.0 and 6.9%, NDF of 51.2 and 61.7%, and DMD of 66.8 and 62.7% respectively. Also for the same area, Pérez et al. (1998) obtained an RP, DOM and NDF in agreement with those of the present work, according to slope position and time of year. Table 4 shows the DM, ME, metabolicity (q) and TME of the available pasture at the different harvests of different years. ME varied between 7.6 and 10.0 MJ kg DM -1 , less than that indicated (10.8 MJ kg DM -1 ) for semi-arid pastureland by Corbett (1981), or for damp areas (11.2 MJ kg DM -1 ) by Givens et al. (1989). This difference is probably due to differences in the vegetation. Early et al. (2001) considered diets with 9.9, 10.3 and 11.4 MJ kg DM -1 to have low, medium and high ME respectively. The metabolicity measured (q) was between 0.43 and 0.52, in line with the value of 0.45 obtained by Susmel et al. (1987) for Mediterranean pastureland, and the 0.5 indicated by Guevara et al. (1991) for gramina pasture.
For systems A and B, the ME of early spring and late autumn pasture was medium (9.0-10.0 MJ kg DM -1 ), and therefore apt for nursing cows. From June to October however, when values are < 8.5 MJ kg DM -1 , this pasture is only good for non-lactating cows (McDonald et al., 1986). Though no hard data are available for the study area, experience shows that with the current management system the greatest number of births (80%) occurs in spring between March and June. According to the results, it could be necessary to supplement mother cows from July through to the end of October.
The energy supplied (GJ year -1 ) by the primary production of acorns is estimated by assigning according to Bellido et al. (1986) a dry matter content of 60% and an ME of 9.8 MJ kg DM -1 . The values obtained for systems A were 87, 207 and 181 for 1989, 1990 and 1991 respectively, while system B showed values of 189, 26 and 108.

Total production and supplementation of systems A and B
The products used to supplement the diet of the cattle were considered as inputs to the ecosystem.   Fig. 1 shows the energy these supplied to each system. There was a period of minimum supplementation from the middle of May until the end of June that was repeated every year of the study. Maximum supplementation occurred during the months of summer, part of autumn and the end of winter, falling slightly in November depending on how early the rains came. During the study period, supplementation was undertaken for approximately nine months, greater than the mean for the area (six months) the kind of supplementation required in years with poor weather conditions.  The total energy supply from the pasture, acorns, and supplementation was 9,407, 15,486 and 15,109 GJ year -1 for system A, and 4,928, 9,782 and 10,277 GJ year -1 for system B, for 1989, 1990 and 1991 respectively. The energy supplied by supplementation was 20% of the entire annual supply in 1990 and 1991.
In summary, significant differences (P < 0.05) were seen in pasture production between years, harvests and the low, middle and high slope position. In contrast, the chemical composition of the pasture was only influenced by the year and harvest date. In general, the RP content, DOM and ME of system A pasture were greater.
If the nutritive value of the pasture during the vegetative cycle is taken into account, the time when protein and energy supplementation is necessary is from the middle of June to the beginning of October.
Pasture production in spring provides the greatest energy input into the systems, corresponding to some 77-95% of the total production of pasture.
The metabolicity of the pasture varies between 0.43 and 0.53 values corresponding to a medium energy content at the beginning of spring and the end of autumn, and a low content over the rest of the year.
Supplementation made up 20% of the total usable energy. The greatest inputs in this respect were made between August and October and January and March, although these periods can be quite reduced depending on the arrival of the season and the criterion of the stock raiser. The parallels between these systems indicates coincidence in season, the way they are managed and climatic conditions. The results may therefore be extrapolated to the wider region.