Introduction

Ethiopia is endowed with a large number of ruminant population standing the first in Africa. Sheep production in Ethiopia has great potential in contributing to food security of smallholder farmers (Legesse 2008). The country has, however, benefited very little from this sub-sector owing to a multitude of problems, mainly attributable to feed in quantity and quality particularly during the dry periods. The consequent poor growth performance of ruminants hinders the country in penetrating to regional and international markets.

To meet the demand for food of the ever-growing human population in Ethiopia, natural grazing lands are alarmingly converted into cropping land (FAO 2011). Consequently, crop residues are becoming the vital source of feeds especially in crop-livestock mixed farming systems (Tolera 2007; Bogale et al. 2008; Tegegne and Assefa 2010). However, crop residues are poor in protein, which is the major limiting nutrient especially in the dryland areas of the tropics and sub-tropics. Therefore, looking for home-grown protein sources such as white lupin grain (WLG) is one possible option.

White lupin (Lupinus albus L.) is available in many parts of the world where it is used as a feed source (Watkins and Mirosh 1987; Brenes et al. 1993). White lupin is also produced by smallholder subsistent farmers mainly in the two regional states of Ethiopia; Amhara and Benishangulgumuz (Yeheyis et al. 2010). Until recently, it was highly neglected and considered as poor man’s food in Ethiopia, mainly used as live fence or sown on fallow land. L. albus is a valuable multipurpose plant contributing for improving soil fertility, food for humans and animal feed. Consequently, farmers have shown increasing interest in cultivating white lupin (Yeheyis et al. 2010). Unfortunately, because of the presence of high alkaloid content (16,752 mg/kg DM) (Yeheyis et al. 2012a), raw WLG is bitter, unpalatable and often toxic for humans and livestock. The presence of toxic and bitter alkaloids and glycosides in L. albus was previously reported (Watkins and Mirosh 1987; Brenes et al. 1993). According to Muzquiz et al. (2012), it has been possible to genetically eliminate the alkaloids in different Lupinus spp. In addition to genetic selection, there are some physical and chemical methods for eliminating these anti-nutritional factors. However, because of lack of access for chemicals and unaffordable cost, when available, traditional methods of processing such as roasting, soaking and dehulling are being exercised by smallholder farmers in Ethiopia. These practices are scientifically supported since most alkaloids of lupin are water-soluble and can be decreased to 0.04 % by soaking in running water (Pisarikova et al. 2008; Yeheyis et al. 2011). For WLG meant for human consumption, roasting, soaking and dehulling are practiced by smallholder farmers in Ethiopia.

The crude protein content of local WLG (Yeheyis et al. 2011) and cultivars elsewhere (Sujak et al. 2006; Bähr et al. 2014) is higher than most of the improved forage legumes promoted in Ethiopia (Melaku et al. 2004), indicating that it can be utilized as potential protein supplement. However, there is scanty information in Ethiopia on its supplement potential for ruminants. This experiment was, therefore, conducted to determine effects of supplementation of different forms of WLG on feed and nutrient intake, digestibility, growth performance and carcass characteristics.

Material and methods

Study site

The experiment was conducted at Wetet Abay town, northwest Ethiopia, located at 37° 8.97′ E longitude and 11° 24.62′ N latitude. The town has an altitude of 1950 m.a.s.l and about 1500 mm rainfall per annum. The mean minimum and maximum temperature of the area are 12 and 31 °C, respectively (WARDO 2009).

Feed preparation

Rhodes grass was harvested at late maturity stage, seed were separated manually and biomass was stored under shade. The hay was chopped to the size of about 3–5 cm—to minimize selectivity and increase intake. The hay was offered ad libitum at 30 % in excess of daily consumption determined based on the previous 5 days average intake.

Raw white lupin grain (WLG) was purchased from local markets and roasted using a clay plate until a black spot was seen at the centre of the grain. While roasting, grains were mixed thoroughly to minimize the difference in degree of roasting. After roasting, grains were air-cooled on a polyethylene plastic sheet. Soaking of roasted WLG was done by keeping it in sacks in running water (river) for 4 days as usually practiced by the farmers. Dehulling was performed manually. Both soaked and dehulled forms of the WLG were air-dried on a polyethylene plastic sheet. After preparation, the different forms of WLG were stored in a cool dry shed until used.

Experimental animals and their management

Twenty-five intact male yearling Washera sheep with initial body weights of 16.26 ± 1.41 kg (mean ± SD) were purchased from local markets in the area where white lupin is grown. The age of the sheep were determined by dentition and verbal report from owners. The experimental animals were quarantined for 2 weeks when they were treated against internal and external parasites using albendazole (300 mg/head) and diazinon (60 %), respectively. The sheep were also vaccinated against common sheep diseases in the area based on the recommendation of a veterinarian. The experimental animals were housed in a ventilated shed with individual pens equipped with watering and feeding troughs.

Experimental design and treatments

At the end of the quarantine period, the sheep were weighed and blocked into five blocks of five animals per block based on their initial body weight. The five treatment diets were randomly assigned to each animal in a block. The feeding trial lasted for 90 days. The treatment diets were Rhodes grass (Chloris gayana) hay (RGH) offered ad libitum (T1), RGH + 300 g raw white lupin grain (WLG) (T2), RGH + 300 g raw soaked dehulled WLG (T3), RGH + 300 g roasted WLG (T4), and RGH + 300 g raw soaked WLG (T5). Supplemented sheep were offered the supplement in two equal portions at 1000 and 1600 hours in a separate feeding trough. Common salt (NaCl) and water were available all the time. The amounts of feed offered and refused were recorded daily for each animal. Substitution rate was calculated using the following formula:

$$ \mathrm{S}\mathrm{u}\mathrm{b}.\;\mathrm{rate} = \frac{\mathrm{Basal}\;\mathrm{diet}\;\mathrm{D}\mathrm{M}\;\mathrm{intake}\;\mathrm{of}\;\mathrm{control}\ \hbox{-}\ \mathrm{Basal}\;\mathrm{diet}\;\mathrm{D}\mathrm{M}\;\mathrm{intake}\;\mathrm{of}\;\mathrm{supplement}\;\mathrm{group}\;}{\mathrm{Supplement}\;\mathrm{D}\mathrm{M}\;\mathrm{intake}} $$

Digestibility and feeding trials

Digestibility trial was conducted before the feeding trial. Experimental animals were acclimatizing to the dietary treatments for 15 days. Then, they were fitted with faecal collection bags for 3 days followed by 7 days of faeces collection. Total faeces voided over 24 h was weighed daily for each sheep, and after thoroughly mixing, 20 % representative samples were taken and kept frozen at −10 °C. Finally, samples were pooled for each animal, thoroughly mixed, 20 % of the composite samples taken and dried at 60 °C in a forced draft oven to a constant weight. The dried faecal samples were ground to pass through a 1-mm sieve and kept in airtight containers. The digestibility of the supplements was determined by difference. The apparent digestibility coefficient of dry matter (DM), organic matter (OM), crude protein (CP), neutral detergent fibre (NDF) and acid detergent fibre (ADF) were determined using the following formula:

$$ \mathrm{Apparent}\ \mathrm{D}\ \mathrm{coefficient}=\frac{\mathrm{Nutrient}\ \mathrm{in}\mathrm{take}\ \hbox{-}\ \mathrm{Nutrient}\ \mathrm{excreted}\ \mathrm{in}\ \mathrm{faeces}}{\mathrm{Nutrient}\ \mathrm{in}\mathrm{take}} $$

After the completion of digestibility trial, the experimental animals were weighed and re-blocked and then offered experimental diets and allowed to acclimatize for additional 15 days before the commencement of the feeding trial that lasted 90 days.

Body weight change and feed conversion efficiency

The initial and final body weights (BW) were obtained by weighing sheep on two consecutive days before feeding and watering using a spring balance with sensitivity of 100 g. Experimental sheep were subsequently weighed at 10-day intervals during the feeding trial. Feed conversion efficiency was calculated as average daily gain (ADG) divided by feed intake.

Carcass characteristics

At the end of the feeding trial, all experimental sheep used in the study from each feeding treatment were fasted overnight, weighed and slaughtered to assess the effect of treatments on carcass characteristics following the slaughtering procedures developed by Clotter (1985) for small ruminants in developing countries. The empty BW was calculated as slaughter live weight less alimentary tract contents. The hot carcass weight (HCW) was estimated after removing the weight of the head, thoracic, abdominal and pelvic contents as well as legs below the hock and knee joints. Dressing percentage (DP) was computed as proportions of HCW to slaughter weight and empty BW. The cross-section of the rib-eye muscle area (REMA) was traced on transparency paper between the 11th and 12th of the left half side of the carcass. The REMA was taken as the mean of the two sides of the ribs. The offal was categorized into edible and inedible components. Offal components are categorized into edible and inedible based on testing, tradition, culture and taboos of the people in different parts of the country. Total edible offal component (TEOC) was taken as the sum of the empty gut, blood, liver, kidney, heart, tongue, kidney fat, abdominal fat, reticulo-rumen, omasum, abomasum, small and large intestines (SLI), testicles and tail whereas total inedible offal component (TIOC) was computed as the sum of the spleen, skin and feet, penis, lung, trachea and oesophagus (LTO), gut content, and head.

Chemical analysis

Feed samples were dried in an oven at 60 °C for 72 h and ground to pass a 1-mm sieve screen and kept in airtight plastic bags pending analysis. Dry matter (DM), organic matter (OM) and nitrogen (N) were determined according to the procedure of AOAC (1990). The CP content was calculated as N*6.25. Neutral detergent fibre (NDF), acid detergent fibre (ADF) and acid detergent lignin (ADL) were analyzed according to the procedure of Van Soest and Robertson (1985).

Statistical analysis

The data obtained on feed and nutrient intake, digestibility, body weight change and carcass characteristics were subjected to analysis of variance (ANOVA) using the general linear model of SAS (2000). Differences between treatment means were separated using Duncan’s multiple range tests when P values of the means were significant (P < 0.05). The model to be used for the analysis of data were

$$ {Y}_{ij}=\mu +{t}_i+{b}_j{+}_{eij} $$

where:

Y ij :

Response variable

μ :

Overall mean

t i :

ith treatment effects

b j :

jth block effects

e ij :

Random error

Results

Chemical composition of experimental feeds

The chemical composition of feed items constituting treatment diets is presented in Table 1. Though hay was made from mature Rhodes grass, its CP content was moderately high. As expected, the CP content of WLG was high. Soaking (T5) resulted in 13 % increase in CP content; soaking and dehulling combined (T3) improved by 51%, while roasting (T4) resulted in decline of CP by 6 %.

Table 1 Chemical composition of experimental feeds (g/kg DM)
Full size table

Feed and nutrient intake

Feed and nutrient intakes are presented in Table 2. Though supplementation level was set at 300 g DM/day/sheep, WLG refusals on average were 167, 141, 108, 148 g for T2, T3, T4, and T5, respectively. A highly significant (P < 0.001) difference was recorded in WLG DM intake within the supplemented groups with the highest consumption recorded in sheep assigned for T4. Rams receiving WLG supplementation had significantly (P < 0.05) lower DM intake of RGH. Supplementation of different forms of WLG resulted in significantly higher (P < 0.001) total DM, OM and CP intake. Among supplemented groups, the highest CP intake was recorded for T3 and the lowest for T2.

Table 2 Daily feed and nutrient intake of Washera sheep fed on Rhodes grass hay supplemented with different forms of white lupin grains
Full size table

Feed and nutrient digestibility

The DM, OM, CP, NDF and ADF digestibility coefficients of treatment diets are given in Table 3. Significantly higher digestibility of DM (P < 0.05), OM (P < 0.01) and CP (P < 0.001) were observed in supplemented compared to the non-supplemented groups. The digestibility coefficients of different forms of white lupin grain are given in Table 4. There were no significant differences (P > 0.05) among the treatments.

Table 3 Apparent digestibility coefficients of nutrients in Washera sheep fed on Rhodes grass hay supplemented with different forms of white lupin grains
Full size table
Table 4 Digestibility of different forms of white lupin grain (WLG) used to supplement Washera sheep fed on Rhodes grass hay
Full size table

Body weight change and feed conversion efficiency

Results of body weight (BW) change and feed conversion efficiency (FCE) are given in Table 5. Supplementation with WLG resulted in significantly higher (P < 0.001) final BW, average daily gain (ADG) and FCE. Among the supplemented groups, sheep on T4 had significantly higher (P < 0.001) ADG and final BW than on T2.

Table 5 Body weight change and feed conversion efficiency of Washera sheep fed on Rhodes grass hay supplemented with different forms of white lupin grains
Full size table

Carcass characteristics

Supplementation with WLG significantly (P < 0.001) improved mean values of slaughter weight (SW), empty BW (EBW), hot carcass weight (HCW), dressing percentage (DP) expressed as SW (DPSW) and rib-eye muscle area (REMA) (Table 6). However, the difference among supplemented groups (T2–T5) was not significant (P > 0.05), except for SW that T4 was significantly (P < 0.001) higher than T2.

Table 6 Effect of different forms of WGL supplementation on carcass parameters of Washera sheep fed on Rhodes grass hay
Full size table

Edible offal components

Supplementation with WLG significantly (P < 0.001) affected most of the edible offals (Table 7). Similarly, supplementation significantly (P < 0.001) increased the amount of abdominal and kidney fat.

Table 7 Effect of different forms of white lupin grain supplementation on edible offal of Washera sheep fed on Rhodes grass hay
Full size table

Inedible offal components

Total inedible offal component (TIOC) and each inedible offal component were significantly (P < 0.01) affected, except the spleen and penis, by supplementation with WLG (Table 8). All inedible offal components were not different (P > 0.05) among supplemented groups.

Table 8 Effect of different forms of white lupin grain supplementation on inedible offal of Washera sheep fed on Rhodes grass hay
Full size table

Discussion

Chemical composition of experimental feeds

Though hay was made from mature Rhodes grass, its CP content (81 g/kg DM) was relatively higher than most tropical feeds commonly available during the dry season (Bediye and Sileshi 1989; Bediye et al. 2007; Tolera 2007; Tegegne and Assefa 2010). The CP content of RGH was sufficient for maintenance of body weight and to maximize digestibility of DM and NDF of straw-based diets (Van Soest 1994; McDonald et al. 2002).

The CP content of the different forms of WLG (362–583 g/kg DM) was also comparable or higher than CP in oilseed cakes (noug, sunflower, cotton, linseed, rapeseed, sesame and groundnut) which are used as protein supplements in Ethiopia (Bediye and Sileshi 1989; Bediye et al. 2007). This is very important in situations where the major limiting factor for animal production is deficiency of N in basal feeds such as cereal; crop residues are important feed resources in Ethiopia (Tolera 2007; Tegegne and Assefa 2010). The difference in CP content between raw (385 g/kg DM) and roasted (362 g/kg DM) WLG (Table 1) could not be explained as roasting had no effect on the same component (Yeheyis et al. 2011). The improvement in CP content in soaked and dehulled WLG (Table 1) was in line with the finding of the latter authors.

Feed and nutrient intake

Rhodes grass hay DM intake of control group in the current study (414 g/day) was relatively higher than the value (360 g/day) reported for Adilo male yearling sheep fed ad libitum RGH (Gebregiorgis et al. 2012). Differences could be attributed to variations in chemical composition of experimental feedstuffs, season of the studies and sheep breed used, though they are indigenous to Ethiopia (Bishaw and Melaku 2008; Mekuriaw et al. 2012). The higher intake of RGH DM by sheep on the control treatment as compared to the supplemented group might be due to the higher consumption of the basal diet to fulfil the nutrient requirement. The low RGH DM intake in the supplemented sheep might be due to the substitution effect though it was not significant (P > 0.05) (Table 2).

The dry matter intake (DMI) (%BW) was comparable to results obtained by supplementing Washera sheep with 290 g/day blue sweet lupin (Yeheyis et al. 2012b). The significantly high (P < 0.05) feed and nutrient intake recorded (Table 2) can be attributed to roasting, soaking and dehulling (Smith and Warren 1986). Yeheyis et al. (2011) reported that roasted WLG had lower alkaloid content (1.07 %) than the raw form (1.43 %). Thus, the higher WLG DM intake recorded in T4 than in other supplemented treatments could be attributed to the roasting process, which is more efficient in alkaloid reduction than the other processing techniques used (Ibid). However, comparatively lower WLG DM intake (133–192 g/day) was recorded in this experiment (Table 2) than the value reported (270 g/day) for sweet blue lupin (Lupinus angustifolius L.—imported) using the same sheep breed (Yeheyis et al. 2012b). This could be more explained by the palatability problem related to the alkaloid content of WLG. Local white lupin grown in northwest Ethiopia has relatively high alkaloid content (16,752 mg/kg DM) as compared to sweet white lupin (178 mg/kg DM) (Yeheyis et al. 2012a). Thus, whether the traditional treatment methods (roasting, soaking and dehulling), which were believed to reduce alkaloid content (Yeheyis et al. 2011), were enough to reduce the anti-nutritional factor to a level that may not negatively affect WLG intake is open to question.

From CP intake data (Table 2), one can see that its concentration in the supplemented groups ranged from 16 to 24 % of the total DM intake which is above the 7–8 % required to satisfy the maintenance requirement of ruminant animals for efficient utilization of feeds at the level of the rumen (Van Soest 1994; McDonald et al. 2002; Norton 2003). Therefore, the improvement in intake of most of the nutrients in the supplemented groups can be warranted by the relatively higher intake of CP. Yu et al. (2002) reported that dry roasting of lupin grain may increase bypass crude protein without having a greater impact on the rumen-degradable characteristics of starch-free organic matter. A previous study showed that roasting could increase protection of lupin oil from ruminal hydrogenation (Robinson and McNiven 1993).

Based on the DM requirement of sheep for maintenance (2–2.5 % BW) and growth (4–4.5 % BW) (ARC 1980; NRC 1985), all supplemented groups consumed more than maintenance requirements while DM intake of all the treatment diets failed to meet growth requirements for DM.

Raw lupin protein is highly degradable in the rumen (Freer and Dove 1984) which may result in poor animal performance (Guillaume et al. 1987). Silano et al. (1981) reported that moderate heat treatment not only enhances palatability and acceptance but also reduces ruminal degradation of protein supplement, and it improves nitrogen retention and supply of amino acids for intestinal absorption. Thus, roasting, in addition to increasing DM and nutrient intake, may improve the value of the protein supplement (Silano et al. 1981).

Feed and nutrient digestibility

The relatively low DM and OM digestibility coefficient of Rhodes grass hay used in this study (0.58 and 0.59, respectively) could be attributed to harvesting of the grass after seed setting. Denekew (2004) reported that OM digestibility declined from 0.72 to 0.49 when harvesting grass for hay making was delayed from early leaf to maturity stage.

The possible explanation for the lower CP apparent digestibility of Rhodes grass hay might be due to lower CP (81 g/kg DM) and higher NDF (813 g/kg DM) contents that negatively affect microbial growth and fermentation in the rumen (McDonald et al. 2002). Comparable to the present study, different research results conducted in Ethiopia, supplementation increased digestibility of DM, OM and CP (Abate and Melaku 2008; Tekletsadik 2008).

Body weight change

Though CP content of Rhodes grass hay should have fulfilled the minimum requirement (Van Soest 1994; McDonald et al. 2002), sheep on the control diet lost 22 g/day. This was comparable with the finding recorded when Arado sheep were fed with teff (Eragrostis tef) straw as basal diet in Tigray, Ethiopia (Gereslassie and Melaku 2009). This low performance could not be fully explained by the DMI (414.6 g DM/day) (Table 2) which in principle could suffice the minimum maintenance requirement (2–2.5 % BW) (ARC 1980; NRC 1985). The higher NDF (813 g/kg DM) and ADL (129 g/kg DM) contents of matured Rhodes grass hay might have limited the use and availability CP. Moreover, sheep in the control treatment had low apparent nutrient digestibility (Table 3).

Supplements result in improved animal performance in several ways, such as by providing essential nutrients for rumen microorganisms, enhancing the microorganism activities in the rumen and providing nutrients for the sheep (Van Soest 1994). The ADG observed in this study was lower than the findings of Hailu et al. (2011), Simachew (2009) and Yeheyis et al. (2012b) who reported 25.0–34.0, 38.9–55.6 and 73.7–91.3 g/day, respectively, for the same sheep breed (Washera) implying that the experimental sheep in this study performed below their potential. This could be attributed to limited feed intake (Table 2), which is much less than the requirement for growth (4–4.5 % BW) (NRC 1985; McDonald et al. 2002), that was probably caused by alkaloids. High alkaloid levels in the diet of experimental animal have been known to depress feed intake and growth (Carr and Pearson 1976). Given that ADG had strong and positive correlation with CP intake (0.92) (Table 9), the low growth performance of sheep on T2 was expected. However, this did not seem to apply for T3 in which sheep had the maximum CPI (Table 2).

Table 9 Correlation between nutrient intake, apparent nutrient digestibility, ADG and carcass parameters of Washera sheep fed on grass hay supplemented with different forms of white lupin grain
Full size table

According to Kung et al. (1991), heat treatment decreases rumen degradability of lupin protein and thereby increases the supply of dietary protein to the lower gut, improving body weight gain, feed efficiency and nitrogen retention in lambs. The highly significant effect (P < 0.001) of supplementing with roasted WLG on ADG (Table 5 ), without affecting digestibility coefficients, and FCE (P > 0.05) is in agreement with the latter authors. The effect of roasting in reducing the alkaloid content and consequently on growth performance (Rahma and Rao 1984) could not be conclusive as alkaloid’s negative effect on DMI could not be ruled out.

Carcass characteristics

Dressing percentage (DP) is one of the important carcass parameters (Massae and Mtenga 1992) which could also be influenced by nutrition. Better quality feeds could increase growth and in turn increase DP (Díaz et al. 2002). Positive effect of WLG supplementation had been observed on carcass characteristics (Table 3 ). Comparable to the present study, Mulu (2005) reported DP ranging from 27.3 to 38.4 expressed as percent SW when Wogera sheep were supplemented with brewery dried grain. Higher DP values for yearling Arsi-Bale, Horro, and Tigray highland sheep were reported (Feyera and Animut 2011; Gebretsadik and Kebede 2011) which could be attributed to breed difference. However, higher DP for yearling Washera sheep (breed used in the present study) supplemented with graded levels of concentrate mix (42.4 to 46.8 and 49.8 to 54.2 on slaughter BW and EBW basis, respectively) (Hailu et al. 2011). In addition, Simachew (2009) reported DP of 30–40 and 46–52 % on slaughter BW and EBW basis, respectively, for Washera sheep weighing 17.3–23.0 kg at slaughter (SW) (fed grass hay basal diet supplemented with 300 g noug seed meal, maize bran and their mixture). The differences within the same breed could possibly be associated with the relatively higher SWs, slaughtering conditions and type and quality of feeds used in the latter experiments.

The mean EBW of Washera sheep in the present study (Table 3) was within a range of 9.6–17.1 kg which was reported for Hararghe highland sheep fed urea-treated maize Stover and concentrate mix (Yeheyis et al. 2012a, b), but lower than 19.9–23.3 kg who reported for Tigray highland sheep fed iso-nitrogenous oil seed cakes in cactus (Opuntia ficus-indica)–tef straw (E. tef)-based feeding (Degu et al. 2009).

The highest (P < 0.001) REMA observed in supplemented sheep than in the non-supplemented sheep (Table 3 ) might be due to the efficient utilization of feeds offered for growth of useful muscle component in the former group. A comparable result to the present study in REMA (3.1–6.9 cm2) was reported for Arsi-Bale sheep (Tafa et al. 2010). However, REMA from the current study is lower than the values reported for different Ethiopian sheep breeds (Feyera and Animut 2011; Hagos and Melaku 2009).

Supplementation increased the amount of visceral and kidney fat, heart, head, LTO and SLI which agreed with previous reports (Hailu et al. 2011; Simachew 2009), except that the latter authors reported higher TIOC in supplemented sheep. Contrary to the current finding, TIOC in Afar rams decreased as supplementation level increased and might depend on slaughtering conditions (Hagos and Melaku 2009).

Conclusion

Among the three processing methods used, roasting could be recommended for ruminant producer smallholder farmers who are threatened by the dwindling access to expensive concentrate. The relevance of dehulling, which is a labour-intensive and time-consuming task, can be questioned. And yet, the improvement in CP content due to dehulling cannot be ignored. Supplementation of Washera sheep with different forms of white lupin grain improved feed and nutrient intake, digestibility, growth and carcass quality parameters. In a country where livestock lose weight during the long dry seasons, white lupin grain can be recommended not only for maintenance, which can be seen as a strategy in dryland areas, but also for optimum performance of ruminants. In terms of CP content, however, high performance could be expected from white lupin grain supplementation. Thus, since this kind of study on local white lupin grain is the first of its kind in Ethiopia, there has to be further investigations on the processing methods as well as animal evaluation.