Examining pasture-based dairy systems to optimize profitability environmental impact, animal health and milk quality

Primary Project at CEFS:
Group and breed differences in milk production levels and composition can be seen in Table 1. Typically, the Holsteins produced the most pounds of mature-equivalent milk, fat, and protein whereas Jerseys produced the least, and the crossbred cows were intermediate to both pure breeds. However, there was an exception in the 2005-2006 High stocking group. Holsteins had poor production that year, lower even than the Jerseys. In contrast, the crossbreds had unusually high milk production compared to other years, surpassing the other breeds. More thorough analyses of what happened in the third year needs to be done. Because the total numbers of animals included in year 3 was fewer than the desired 40, a few animals with disproportionately high or low milk yields may have skewed the results.

Across all 3 years of the study, the proportions of time spent on each of the 3 pasture types (annual ryegrass/sorghum-sudangrass; annual ryegrass/Bermudagrass; and fescue/Ladino clover), the stored-forage feeding area, and shade areas by lactating cows stocked at the Low stocking rate were 40.3%, 16.4%, 17.1%, 21.2%, and 5.1%, respectively. The corresponding proportions of time spent by cows stocked at the High stocking rate were 34.7%, 13.6%, 17.3%, 29.3%, and 5.1%. The major difference was 38.2 % more time (29.3 % vs. 21.2%) spent on the stored-forage feeding area for cows at the High stocking rate and less time on the annual ryegrass/sorghum-sudangrass pastures and the annual ryegrass/Bermudagrass pastures. These data do not consider hourly time spent milking and walking to and from the milking parlor, and consider the daily access to shade as 0.5 of a day (cows were moved to the wooded areas at about 9:30 am and were kept there until the afternoon milking time).

Differences between the 2 stocking rates were seen in that cows in the High stocking group produced significantly more milk, fat and protein than those in the Low stocking group each year and cumulatively. The advantage for the High stocking rate group in yield measures was about 7 to 11 percent per cow across the 3-year study with differences being greater for crossbreds and Jerseys than for Holsteins. However, based on land area allocated and actual stocking rates of 3.2 and 2.2 cows per ha, yield measures per ha would be about 10 to 16% higher for the herd at the higher stocking rate.

Reproductively, there were no significant differences between the two stocking rate groups. However, there were breed differences in measures of reproduction. Various reproductive measures for the different breeds averaged across stocking rates and all three years of the study are included in Table 2. Jerseys and Crossbreds had higher conception rates at 1st service and over all services as well as having higher pregnancy rate over the entire breeding season than Holsteins. However, the values for conception rates among the Jerseys were lower than expected in the first year of the study, likely because they were moved from a TMR-feeding system at another location in early lactation and had to adapt to the pasture-based system just before the breeding season. That year resulted in a lower average conception for the 3 years and likely does not reflect long-term reproductive success of Jersey cows (Washburn, et al; 2002).

There were no significant differences in linear SCC scores by either stocking rate or breed group across the 3-year study (Table 2). Other health issues included concerns about Staphylococcus aureus mammary infections in some cows, presumably spread by horn flies but not apparently related to differences in breed or stocking rate.

Grazing behavior at CEFS:
Preliminary observations on grazing behavior of purebred and crossbred cows under heat stress conditions were conducted at CEFS. However, there were difficulties in keeping the chewing monitors in place throughout 24-hour periods. For some cows, we were able to distinguish between patterns of jaw movements for grazing versus those for chewing cud. However, the preliminary observations were not adequate to ascertain breed differences in grazing behavior under differing environmental temperatures. However, the monitors are still functional and further, more comprehensive studies can be done.

Field Data from Commercial Farms- Clemson:
The content of total fatty acids decreased from spring to fall but the content of linoleic (C18:2) and linolenic acids (C18:3) (Figure 3) increased over that period. Freeman (2004) also reported a decline in total fatty acids over the life cycle of rye (linear) and ryegrass (quadratic). However, the life cycle for rye in that study was from November to March, and March to June for ryegrass and samples for the current study were collected between March and August. Freeman reported a linear decrease in C18:3 during the life cycle of both rye and ryegrass. The content of C18:2 remained constant through the life cycle of rye and exhibited a cubic increase for ryegrass.

Conjugated linoleic acid content of milk also increased during from March to August Linoleic and linolenic acids, the two main fatty acids found in pasture, are the precursors for the synthesis of CLA in the rumen and mammary tissue. Kelly et al. (1998) reported that when peanut oil (high oleic acid), sunflower oil (high linoleic acid) and linseed oil (high linolenic acid) were added to diets of lactating dairy cows, milk CLA concentrations (mg/g of milk fat) were 13.3, 24.4 and 16.7, respectively. Dhiman et al. (2000) reported that diets rich in linoleic or linolenic acid can increase the CLA content of milk when the dietary oil is accessible to the rumen microorganisms. In their study, diets that contained 0.5, 1.0, 2.0, or 5.0% soybean oil or 1.0% linseed oil were fed to lactating dairy cows. Diets with soybean oil had a higher percentage of C18:2 fatty acid whereas diets containing linseed oil had a higher percentage of C18:3 fatty acid. Average CLA content in milk fat was 0.50% for the control diet (no added soybean or linseed oil) and 0.75, 0.76, 1.45, 2.08, and 0.73% of total fatty acids the diets with added oil, respectively.

Estimated pasture dry matter intake and milk CLA were highly correlated (R2=0.7829). Average CLA content in milk is typically between 0.3 and 0.6% of total fatty acids (Kelly et al., 1998); milk CLA varied between 0.49 and 1.15% in the current study. Increased consumption of pasture typically is associated with increased milk CLA. Jahreis et al. (1997) reported that cows fed silage had 0.34% CLA in milk compared to 0.80% for cows that grazed. They then concluded that the increase in CLA formation was due to the high level of unsaturated fatty acids in the grass. Stanton et al. (1997) fed 36 dairy cows three different amounts of fresh perennial ryegrass (16 kg, 20 kg, 24 kg DM) and found milk CLA content increased in the higher pasture allowances resulting in 3.94, 6.84, and 5.71 mg of CLA/g fat for the respective treatments.

Dhiman et al. (1999) studied the effect of feeding lactating Holsteins one-third, two-thirds, or their entire diet as pasture (bluegrass, quackgrass, bromegrass, and white clover). Cows fed only pasture had 500% more milk CLA than controls fed a total mixed ration and hay and a linear relationship was seen between pasture consumption and CLA. Concentrations of milk CLA concentrations were 8.9, 14.3, and 22.1 mg/g of fatty acids for the one-third, two-thirds, and all pasture groups, respectively. Kelly et al. (1998) also conducted a study that compared the CLA content of milk from cows that consumed all of their dry matter from pasture to cows that were fed a total mixed ration. Average milk CLA content for cows that grazed was 1.09 versus 0.46 % (of total fatty acids) for the cows fed a total mixed ration.

It was our initial intent to characterize the milk CLA response according to forage type. We were unable to do this, however, because of the wide array of forages grazed and because the farmers used a wide variety of forage combinations in their grazing scheme. However, we concluded that milk CLA content was highly correlated with estimated pasture intake over a wide variety of forages, breeds, herds, and management conditions.

Indicators of Immune Function – Va Tech:
System effects were consistently observed for measures of immune function (Table 1). Cows on Low stocking/supplementation rate systems displayed greater phagocytosis and higher levels of the antioxidant Superoxide Dismutase (SOD). However, Glutathion Peroxidase (GSH-Px) and Malondialdehyde (MDA), a product of lipid peroxidation resulting from cell membrane damage and a proxy measure for oxidative stress did not differ by system. No breed effects on phagocytosis or superoxide dismutase were observed.

Superoxide dismutase remained consistently higher in dairy cows maintained on Low stocking rate-Low supplement treatment compared to the High group, regardless of season (Fig. 1). This suggests dietary antioxidant supply was adequate from a proportionately high forage diet across stage of production. Conversely, the data may imply that compensating for high stocking rate and subsequent reduced forage mass per animal with concentrate, compromised antioxidant status (Fig.1). This notion can be supported by the lower phagocytic activity of cows in the high group compared to low treatment group cows (Table 3). A similar observation is apparent with WBC GSH-Px in high versus low group cows across all seasons. Both SOD and GSH-Px are integral to allowing the phagocytic immune cell to function adequately with in body. If they are inadequate, phagocytic cell function is severely compromised.

Season by breed interactions were observed for both red and white blood cell glutathione peroxidase concentrations (Figure 2). For Holsteins and crossbreds, glutathione levels were greatest in Winter, least in Spring, and intermediate in Summer and Fall measurement periods. However, glutathione peroxidase levels for Jerseys were numerically greatest in Summer, and unlike the other breed groups, white blood cell glutathione peroxidase changed little from Winter to Spring.

Averaged across seasons, glutathione peroxidase in white blood cells tended to be greater for Jerseys and crossbred cattle than for Holsteins. In red blood cells, glutathione peroxidase levels were greater in Jerseys than for Holsteins and crossbreds. Some caution must be taken when interpreting the significance of these breed effects, however, because the number of Jersey cows available for breed comparison was limited.

Seasonal effects were observed for all immune function and oxidative stress measures (Table 2). Immune and antioxidant parameters were generally at their highest at the Winter measurement period (in January, about 3 months after calving). Interestingly, percent of phagocytic cells responding was lowest in Fall, just after calving, and in Summer, when heat stress would likely have been greatest, but nutritional demands would have been lower. Superoxide dismutase levels also were lowest in Summer but glutathione peroxidase levels were lowest in Spring and intermediate in Summer.

Treatment had little effect on concentrations of serum alpha tocopherol for all measurement periods except Winter, when serum alpha tocopherol was greater from cows on the low grain high forage system. Alpha tocopherol concentrations in serum were generally lower for Holsteins, intermediate for Jerseys, and greatest for crossbred cattle. Marked seasonal effects were observed, with low levels of alpha tocopherol observed near calving; concentrations in serum were double or triple these levels at the remaining measurement periods. This may reflect greater tocopherol intake attendant with lactation, but the pattern over time did not match a lactation curve, per se.

Serum gamma tocopherol concentrations, in contrast to alpha tocopherol, were much greater from cows in the high grain low forage system. This response was greatest in Winter, when gamma tocopherol levels were more than three times greater for cows fed high-grain diets. Moreover, Holstein cattle had about 33% greater gamma tocopherol concentrations (83 ppm) than did Jerseys (61 ppm), with slightly lower concentrations for cross bred (78 ppm). Seasonal effects for gamma tocopherol were similar to that of alpha tocopherol, but the large peak observed in winter appears to have been driven by the high level of grain consumption in the high grain low forage system.

Beta carotene, the vitamin A precursor, was typically greater in plasma from animals in the low grain high forage system, but this did not translate to measureable system differences for vitamin A. As was expected, both vitamin A and beta carotene were greatest in serum from Jerseys, least for Holsteins, and intermediate for the crosses. Both analytes followed the same, seasonal pattern as the tocopherols, with low levels at calving in fall followed by marked increases during the grazing season. The concentration of carotene peaked in spring, however, whereas vitamin A was greatest in winter (Figure 3). Concentrations of carotene in winter remained high for animals on the low grain high forage system, but much lower levels were observed for animals in the high grain system, suggesting that higher grain feeding (co-incident with peak lactation) negatively affected this carotenoid. Again, and surprisingly however, this did not translate into a similar interaction between season and treatment for vitamin A.

Carotene levels in forages typically were low in winter, but sorghum-sudan (sudex) sampled in July 2005 had quite high levels of beta carotene and alpha tocopherol (Figure 4). It is not clear whether lower levels of alpha tocopherol in forage from the low stocking rate pastures in July 2005 reflects greater stem material in the sample (or the stand in general). In contrast to beta carotene, alpha tocopherol levels were generally high in winter. Poor growing conditions may have precipitated the low levels of vitamins in forages sampled in May 2004.

It appears that Low stocking with minimal concentrate supplementation resulted in an enhanced antioxidant and immune cell capacity in the dairy cows grazing on this system. The effects appeared to be sustainable across season, which implies across production stages and changing forage types. Increased immunocompetence capacity would reasonably be linked to healthier cows although no remarkable differences in animal health were noted between the two stocking rates in the current study. It could have implications if stocking rates were to exceed those in the current work.

In terms of milk production and the apparent enhanced immunocompetence, that evaluation has yet to be performed. This information in conjunction with the milk composition and quality data will help to better define the traditionally quantifiable parameters considered in dairy production management. These data should be considered a valuable starting point to promote dairy grazing systems. A closer look at specific forage type and inherent antioxidant nutrient availability can help clarify the data across seasons, but could likewise help direct forage feeding protocols across seasons and across breeds. Future studies in this area can utilize newer antioxidant-related assays, which were not available at the start of this study, to increase our ability to make a more field-applied interpretation of all data. Currently, the data may be most useful as a stepping stone (research tool) to plan future studies and as a classroom teaching aid.

Alternative Starch Sources for Grazing Cows - Clemson:
Treatments were grain supplements based on: (1) dry ground corn (CORN), (2) rolled barley and molasses (BM), or (3) citrus pulp and molasses (CM). Supplement intake was not different across treatments, but pasture intake tended (P < 0.10) to be lower for cows on BM (13.7 kg/d) than for cows on CORN (15.9 kg/d) or CM (16.1 kg/d) (Table 5). This resulted in a trend (P < 0.10) for a difference in total dry matter intake (DMI) among treatments with cows on BM (22.8 kg/d) tending to be lower than for cows on CORN or CM (25.0 or 25.2 kg/d). Intakes of CP and NDF were not different among treatments. Intake of ADF was lower (P < 0.05) for cows on BM than for cows on CM, because total intake tended to be lower for cows on BM, and because CM contained the highest ADF level. Body weights and body weight change did not differ among treatments.

Fiber-based concentrates may have advantages over feeding starch-based concentrates to grazing cows by increasing DMI. When early lactation cows grazed ryegrass pasture, pasture and total DMI were increased 0.7 (Meijs, 1986) and 0.8 kg/d (Sayers et al., 2003) when fiber-based concentrates replaced starch-based concentrates. Bargo et al. (2003) suggested that replacing starch-based concentrates with fiber-based concentrates would increase rumen pH, enhance pasture digestion, and result in higher DMI. Pasture and total DMI were similar with both types of concentrates for late-lactation cows grazing orchardgrass (Delahoy et al, 2003). Although there are a low number of studies, Bargo et al. (2003) reported that, overall, fiber-based concentrates slightly increased DMI 0.13 kg/d but there was a large variation among studies and ranged from -0.7 to 1.4 kg/d.

Treatments had no effect on yield of milk, 3.5% FCM, or ECM, or on milk fat percentage or yield (Table 5). Milk protein percentage was higher (P > 0.05) for cows on CORN compared to cows on CM (2.81 vs. 2.70%). Delahoy et al. (2003) also reported higher milk protein content in milk from grazing cows supplemented with ground corn compared to supplementation of non-forage fiber sources (beet pulp, soybean hulls, and wheat middlings), (3.23 vs. 3.19%). Khalili and Sairanen (2000) found no differences in milk protein percentage for grazing cows with no supplement or supplemented with barley or a mixture of concentrate sources that included non-forage fiber (wheat bran and molasses sugar beet pulp), 3.42 vs. 3.43 or 3.49%. However, protein yield was lower for cows on pasture only compared to cows fed barley, which was lower than that for cows fed non-forage fiber (0.61 vs. 0.67 and 0.73 kg/d) because of lower milk yield.

Meijs (1986) also reported increased milk production when fiber-based concentrates of beet pulp and soybean hulls replaced corn and cassava. Two other grazing studies, however, reported similar milk yields (Sayers, 2003; Delahoy et al., 2003) and others reported reduced milk yield (Valk et al., 1990). Bargo et al. (2003) reported that milk production was slightly reduced across published studies (-0.46 kg/d) when fiber-based concentrates replaced starch-based concentrates for grazing dairy cattle, but milk response ranged from -2.6 to 1.3 kg/d.

Sayers (2003) reported higher milk fat percentage with fiber-based concentrates compared to starch-based concentrates. Most studies, however did not report changes in milk fat percentage (Meijs, 1986; Valk et al., 1990; and Delahoy et al., 2003). In addition, Bargo et al. (2003) summarized that replacing starch-based concentrates with fiber-based concentrates reduced milk protein -.06 percentage units (range: -0.21 to 0.05 percentage units). In this study, partial replacement of corn with citrus pulp and molasses did not affect milk fat percentage or yield but did result in lower milk protein percentage (2.81% versus 2.70%); neither were different from BM (2.77%).

In this study, cows consuming BM tended (P < 0.10) to have a greater efficiency of energy-corrected milk yield than cows consuming CORN (1.40 vs. 1.29 kg milk/kg DMI), while there were no differences between CORN and CM (1.30 kg milk/kg DMI) or BM and CM (P > 0.11). There were no differences among treatments for efficiency of milk or fat-corrected milk yield.

Selected milk fatty acids are also shown in Table 5. Milk from cows on CORN and CM was higher (P < 0.05) in trans-11 C18:1 than for cows on BM. Trans-11 C18:1 is an intermediate in the biohydrogenation of C18:2 and C18:3 (Bauman and Griinari, 2003). It can also be converted to cis-9, trans-11 C18:2, commonly known as CLA, by the action of stearoyl-CoA desaturase in the mammary gland. C18:3 was higher (P < 0.05) in milk from cows on CM compared to milk from cows on CORN and BM. There were no differences among treatments for C4:0, C6:0, C8:0, C14:0, C14:1, C15:0, C16:0, C16:1, C18:0, C18:1, C18:2, cis-9, trans-11 CLA, trans-10, cis-12 CLA, or other fatty acids. Kelly et al. (1998) reported cows consuming mostly ryegrass pasture produced milk with 1.09 g/100g fatty acid CLA while cows fed a TMR in confinement produced 0.49 g/100g fatty acid CLA. CLA content in this experiment averaged 0.62 g/100g fatty acids and was slightly lower than levels reported by White et al. (2001) for Holstein cows grazing crabgrass and supplemented with a concentrate (0.72 g/100g fatty acids). Bargo et al. (2006) found that supplementation with a corn-based feed lowered milk CLA concentration as compared to no supplementation (1.18 vs. 1.36 g/100g fatty acid), but the values reported were higher than those found in this study. Bargo et al. (2006) also found a tendency (P = 0.07) for cows supplemented with cracked corn to produce higher levels of CLA than those consuming steam-flaked corn (2.73 vs. 2.26 g/100g fatty acid) but found no differences when cows consumed ground corn or a non-forage fiber supplement (average: 2.85 g/100g fatty acids).

BUN and MUN can be used as indicators of rumen N capture as these values are positively associated with rumen ammonia concentrations (DePeters and Ferguson, 1992). Data for BUN is shown in Figure 5. Average BUN did not differ among treatments (average: 10.60 mg/dL). As expected, there was an overall effect of time on BUN (P < 0.05). BUN was lower for cows on CM than for cows on CORN at 0400 h.

Blood urea N values for this study were lower than expected for cows on pasture supplemented with a carbohydrate source. Bargo et al. (2002b) reported BUN values of cows grazing high quality pasture while being supplemented with a corn-based concentrate to be 17.2 mg/dL, while Kolver et al. (1998) reported BUN averaged 22.05 mg/dL for supplemented cows on pasture. Delahoy et al. (2003) found average BUN values of 13.1 mg/dL for cows supplemented with corn.

MUN was higher (P < 0.05) for cows on BM compared to cows on CORN and CM (Table 1). Similar to BUN values, MUN values are also lower in this study than expected. Other research reported MUN value of supplemented cows on pasture to average 19 mg/dL (range 14.8 to 37.6 mg/dL), (Bargo et al., 2002b; Delahoy et al., 2003; Khalili and Sairanen, 2000) with MUN values for cows on pasture only reported as high as 40 mg/dL (Khalili and Sairanen, 2000).

While there were differences in MUN among treatments, no significant differences were found for BUN values. However, numeric differences among treatments for BUN follow a similar pattern as those seen for MUN values, with cows consuming CM having the lowest BUN and MUN values (10.19 and 9.85 mg/dL), followed by CORN (10.62 and 10.05 mg/dL) and BM (10.99 and 11.43 mg/dL). The reason for the conflicting findings is unknown; however, the small differences in N capture were not reflected in changes in milk production.

One of the strategies to improved efficiency of grazing cows is to match the rate of degradation of the pasture N with the rate of carbohydrate degradation from the supplement. Kolver et al. (1998) reported peak ruminal ammonia concentrations were reduced 33% when grazing cows were fed concentrate synchronously with pasture rather than 4 h after pasture was fed.

Because the starch in barley degrades significantly faster than the starch in corn (24.5 vs. 4%/h), a partial replacement of corn with a barley and molasses mix should result in starch degradation that more closely matches the N degradation of pasture. García et al. (2000) reported that ruminal ammonia concentration was significantly reduced when heifers fed fresh forage were supplemented with barley compared to corn (19.4 vs. 26.9 mg/dL). Khalili and Sairanen (2000) found that barley supplementation did not reduce rumen ammonia levels in cows grazing pasture that was 20.9% CP compared to corn supplementation, however, it was reduced by feeding a combination of barley, oats and beet pulp (28.7, 32.1 and 21.8 mg/dL for corn, barley, and barley/oats/beet pulp mix, respectively.) There were no differences in MUN between the concentrate mixture and barley (37.6 and 36.3 mg/dL), but both were significantly lower than corn (40.0 mg/dL). The grain mixture also increased yield of milk protein over corn or barley (0.73, 0.67, and 0.61 kg/d, respectively for mix, barley and corn) as well as milk yield (21.0, 19.7, and 18.4 kg/d, respectively for mix, barley and corn).

Because the neutral detergent soluble fiber in citrus pulp is thought to degrade at similar rates as ryegrass pasture N, 13%/h (Hall et al., 1998), a partial replacement of corn for citrus pulp and molasses should offer an advantage. Miron et al. (2002) reported that partial replacement of corn by citrus pulp in TMR fed to high-producing dairy cows resulted in improved feed efficiency because the digestibility of neutral detergent soluble carbohydrates was higher for the diet with citrus pulp versus the diet with corn. Fermentation of pectin is different from starch in that, although it is extensive, it produces little or no lactate and results in a higher acetate to propionate ratio than starch (Hall et al., 1998). Although other sources of non-forage fiber, including beet pulp, soybean hulls, and wheat middlings have been evaluated for grazing cattle (Delahoy et al., 2003), there is a lack of grazing studies that have evaluated citrus pulp as a supplement for grazing cows.

Few studies that considered replacement of starch-based concentrates with forage-based concentrates reported ruminal ammonia, BUN, or MUN. Delahoy et al. (2003) included non-forage fiber sources (beet pulp, soybean hulls, and wheat middlings) in addition to ground corn in a supplement for late-lactation grazing dairy cows and reported that the cows fed ground corn had lower MUN than cows fed the non-forage fiber concentrate (14.9 vs. 15.4 mg/dL). Plasma urea N, however, was not different between treatments.

In this study, partially replacing corn with barley and molasses did not improve the capture of ruminal N and in fact, resulted in higher MUN. Blood urea N, however, was not different across treatments. Cows on BM, however, did result in improved efficiency of EMC because pasture intake was lower but milk yield was not different. Partially replacing corn with CM did not improve milk yield or overall capture of pasture N, but BUN was reduced during one collection period compared to CORN. Milk protein content was lower for cows on CM than for cows on CORN but milk protein yield was not different. One of the reasons that treatments effects were minimized may have been due to low CP content of the ryegrass pasture utilized in this experiment which averaged 16.5%.

Bargo et al. (2002b) found BUN and MUN levels for cows consuming TMR with 16.9% CP content to average 13.8 and 10.6 mg/dl, respectively, while BUN and MUN levels for cows consuming pasture averaging 26.3% CP and a corn supplement were found to be 17.2 and 14.9 mg/dl, respectively. BUN and MUN levels for cows on this trial were similar to cows on TMR than cows on pasture. If BM or CM improved nitrogen capture, the CP content of the pasture may not have been high enough to allow for detection of differences. Another explanation could be that the degradation of the corn in starch was more rapid than expected. Oba and Allen (2003) reported a degradation rate of 14%/h for the starch in corn, which is considerably higher than the 4%/h previously reported by Tamminga et al. (1990).

Partial replacement of corn with BM or CM did not offer advantages to cows grazing ryegrass pasture as measured by milk yield. Cows fed BM had higher MUN. However, cows on CM did have lower BUN during one collection period and may have shown more advantage if the pasture CP content was higher. For this reason, partial replacement of corn with citrus pulp for grazing cows should be further studied using pasture with higher CP content. In addition, if the price of barley or citrus pulp is favorable compared to corn, their inclusion in rations should be considered since milk yield did not decline and in fact, efficiency of ECM yield was improved with barley. Milk protein yield declined for cows fed C, so graziers that are paid for milk protein should limit the amount of citrus pulp that replaces corn.