Nitrogen Metabolism and Plasma Urea ... - CSIRO Publishing

Aust. J. Agric. Res., 1987, 38, 917-26

Nitrogen Metabolism and Plasma Urea Concentrations in Fleeceweight-selected and Control Romney Rams

S. N. McCutcheon, D. D. S. Mackenzie and H. T. Blair Department of Animal Science, Massey University, Palmerston North, New Zealand.

Abstract Nitrogen retention and plasma urea concentrations were examined in 2-year-old Romney rams from fleeceweight-selection and control lines. In four experimental periods rams were fed chaffed lucerne hay (1 10% of maintenance energy requirements) three times daily (Period I), twelve times daily (Period 11), twice daily (Period IV), or were fasted (Period 111). Nitrogen balance was measured in Period I, while plasma concentrations, urinary excretions and clearances of urea and creatinine were examined in Periods 11-IV. Water intake and urine output were measured in all periods. Plasma urea concentrations were also measured in the same rams at grazing. Differences between the lines in water intake, urine output, faecal and urinary nitrogen excretion and nitrogen retention were not significant. Control rams maintained significantly higher plasma concentrations of urea and creatinine than fleeceweight-selected rams but only under controlled feeding conditions (particularly twelve times daily feeding). Differences between the lines in plasma urea concentration could be accounted for by the (non-significantly) greater urinary urea excretion, and lower creatinine clearance rate, of control rams. Measurement of plasma urea concentration in sheep may provide a useful predictor of genetic merit for fleeceweight. However, it will be necessary to measure plasma urea concentration under controlled feeding conditions to accurately rank animals. Concurrent measurement of creatinine clearance rate and urinary urea excretion should also enhance the accuracy of prediction of genetic merit.

Introduction As a result of 30 years of single-trait selection, the Massey University fleeceweight-selected (FW) Romney line now exceeds its random-bred control (C) line in greasy fleeceweight by approximately 20% (Blair 1986). Patterns of wool growth in the two lines have been examined only in the hoggets. In grazing hoggets the difference between the lines in wool production is evident throughout the year, but is greatest during the period June to October (McClelland et al. 1987). However, the physiological basis for this difference is not known. Clark (1987) observed that, when fed at or near maintenance, rams and ram hoggets from the FW line exhibited significantly lower concentrations of urea and creatinine in plasma than those from the C line. McClelland et al. (1986) found that FW hoggets produced approximately 30% more urine than C hoggets (P < 0.05). Water intake was not measured but would likely have been greater in the FW hoggets, a result consistent with those from studies of Merino fleeceweight selection lines in Australia (Dolling and Carpenter 1962; MacFarlane et al. 1966). These observations suggest the possibility that differences in kidney function might be responsible for differences between the lines in plasma concentrations of urea and creatinine. Clark (1987) also administered intravenous methionine to rams from the two lines. Methionine infusion reduced plasma urea concentrations to a greater extent in C than in FW rams. If, as in Merino selection lines (see McGuirk 1979), the wool of FW sheep contains a lower 0004-9409/87/050917$02.00

S. N. McCutcheon et al.

proportion of sulfur-containing amino acids (principally cystine) than that of C rams, this may indicate that the high plasma urea concentrations of C rams reflect greater deamination of non-sulfur-containing amino acids. The differential plasma urea concentration in C and FW sheep has potential utility as a means of predicting genetic merit for fleeceweight in young animals. However, the conditions under which this difference is most reliably expressed have not been examined. The present study was therefore undertaken to compare the effects of different feeding patterns (indoors and at pasture) on plasma urea concentrations in the two lines and to examine the physiological basis for the between-line difference.

Materials and Methods The study involved five C and five FW 2-year-old rams weighing 59.7 + 2.3 and 62.1 + 2.8 (mean k s.e.) kg respectively. The rams had previously been used in the study of Clark (1987) , and so were allowed a 2-week readjustment period (during which they were fed chaffed lucerne hay) prior to the commencement of the present study. The controlled feeding study was conducted (during September) over four experimental periods, the first involving measurement of nitrogen balance and the remainder different patterns of feeding/fasting to induce vanable patterns of plasma urea concentrations. Rams were housed in metabolism crates and fed chaffed lucerne hay (except during fasting in Period 111). During each of Periods, I, I1 and IV, hay was fed at a rate calculated to provide 110% of energy requirement for maintenance (i.e. 0.55 MJ M E / ~ ~ O assuming .'~ the feed had 9.5 MJ ME/kg DM and 85% DM; Rattray 1986). Crude protein content of the hay was 16.1% on a DM basis. This level of feeding was chosen to avoid problems of diet selection which commonly occur at higher allowances of lucerne chaff, and because between-line differences in plasma urea concentration had previously been observed at similar allowances (Clark 1987). On alternate days, 2 g of a mineral supplement (94% sodium chloride, 6% sodium molybdate) was fed with the lucerne chaff to counteract possible copper toxicity. Water was available ad libitum. Period I During the 7 day period following readjustment, feed was offered in three equal portions at 0800, 1600 and 2400 hours daily. At each feeding time water intake was measured and faeces collected and stored at -20°C for subsequent nitrogen determination. Urine was collected into buckets containing 40 ml/day 5 M HC1 (divided pro rata between sampling intervals), and the urine output in each sampling interval weighed. Samples (20 ml) were frozen for determination of urinary urea, creatinine and total nitrogen. Period II Jugular cannulae were inserted under local anaesthetic at the end of Period I and 24 hours later the rams were placed on a 2-hourly feeding interval. Sampling began after an 8 h adjustment to this feeding regimen and continued for 24 h. Water intakes and urine outputs were measured at 4 h intervals and urine sampled as in Period I. Blood samples (8 ml) were withdrawn at 2 h intervals (immediately prior to feeding). Plasma was separated by centrifugation (using sodium citrate as the anticoagulant) and frozen for determination of urea and creatinine concentrations. Period III Rams were fasted for 52 h at the end of Period 11. During successive 8 h intervals, commencing 4 h after the last feed, water intakes and urine outputs were measured, and urine and plasma samples obtained for urea and creatinine determinations. Period IV At the end of Period 111 the rams were placed on a once-daily feeding regimen for 3 days. During the next 4 days they were fed half the daily ration at 0800 hours and the balance at 2000 hours to produce a biphasic pattern of plasma urea concentrations. Period IV measurements were made on the fourth day. Urine outputs and water intakes were recorded at 4 h intervals commencing 0800 hours and urine samples taken. Blood samples were withdrawn at hourly intervals for the first 8 h and then at 2 h intervals for the remaining 4 h prior to the next feed. This sampling pattern was repeated during the 12 h after the second feed. Field Measurements At the end of the main experiment rams were returned to pasture for 6 weeks, at the end of which period they were shorn. After a further 4 weeks on pasture they were brought indoors and fasted for 24 h. Blood samples were taken (by vacutainer using E m A as the anticoagulant) immediately off pasture; after 12 h and 24 h of fasting; and 3 h, 6 h after the animals were returned to pasture. Plasma was stored

Nitrogen Metabolism in Romney Rams

frozen for analysis of urea concentrations. Wool samples were clipped from the midside (over the last rib).

Chemical Analyses and Calculations Nitrogen content of feed, faeces and urine was determined by an automated Kjeldahl method on a Kjeltec Auto 1030 analyser (Tecator AB, Sweden). Urea and creatinine concentrations in plasma and urine were determined by the autoanalyser methods of Marsh et al. (1965) and Chasson et al. (1961), respectively. Wool sulfur content was assayed in scoured and regained 50 mg samples using the dry oxidation method described by Landers et al. (1983). For each sampling interval within period, urea clearance rates were calculated as the ratio of urea excretion rate (= rate of urine production x urinary urea concentration) to average concentration of urea in plasma. Equivalent calculations were used to derive creatinine clearance rates. Average plasma concentrations of urea or creatinine were derived from all blood samples taken during each interval of urine collection. Statistical Analyses Data were initially analysed by repeated measures analysis using a generalized linear model computer package ('REG'; Gilmour 1985) to test overall selection line effects and line x time interactions. Since the interactions were generally nonsignificant, data for each period were then reduced to a single mean for each ram and subjected to analysis of (co)variance.

Results

Period I Parameters relating to nitrogen metabolism during the 7 day balance period are presented in Table 1. Daily excretion rates of urea and creatinine were significantly ( P < 0.05) related to body weight and so have been expressed on a per unit body weight basis. There were no significant differences between the selection lines in digestibility (of nitrogen or dry matter) or urinary nitrogen excretion. Although the rams had been fed at a rate calculated to provide 110% of maintenance energy requirements, they were in a slightly negative nitrogen balance. Water intake, urine output and urinary excretion of urea and creatinine were not significantly different between the two lines. Table 1. Nitrogen balance, urinary excretion of urea and creatinine, and water intake in fleeceweight-selected and control rams (Period I) Variable Nitrogen intake (g/day) Digestibility of nitrogen Digestibility of dry matter Faecal nitrogen excretion (g/day) Urinary nitrogen excretion (g/day) Nitrogen balance (g/day) Water intake (ml/day) Urine output (ml/day) Urea excretion (mmol kg-' day-') Creatinine excretion (mmol kg-' day-')

Selection line ( n = 5) Fleeceweight Control (C) (FW)

Pooled s.e.

33.4 0.723 0.639 9.2 25.9 -1.75

34.3 0.718 0.639 9.7 25.6 -.98

1.1 0.006 0.006 0.3 1.5 0.9

4843 2794 12.9 0.24

4798 2588 12.1 0.22

446 355 0.3 0.01

Period 11 Plasma urea and creatinine concentrations during Period I1 are shown in Fig. 1. It is clear from this figure that the feeding regimen adopted was successful in producing steady-state conditions with respect to plasma concentrations of these metobolites. Control rams maintained concentrations of urea and creatinine which were consistently greater than those of


FW rams. Patterns of water intake, urine output and urinary excretion of urea/creatinine were less consistent over time. Changes in urea and creatinine clearance with time paralleled the urinary excretion patterns and were presumed to be a consequence of intermittent voiding of urine rather than true temporal patterns in clearance. 9.0

r Fig. 1. Plasma concentrations of urea (upper panel) and creatinine (lower panel) in five fleeceweightselected (-) and five control (---) rams during 2-h feeding (Period 11). Bars represent standard errors about the mean. Rams were fed their daily ration in 12 equal portions (immediately after each blood sample).

2400

0400

0800

1200

1600

2000

Sampl~ngtime (hours)

Relationships among these variables (calculated over all 10 rams) are in Fig. 2. Regression analysis showed that urea excretion, creatinine excretion, urea clearance and creatinine clearance were again all significantly ( P < 0.05) related to body weight. Hence these variables were expressed on a per unit body weight basis prior to deriving the relationships shown in Fig. 2. As is shown in Fig. 2, water intake and urine output were highly correlated (r = 0.96, P > 0.01), a relationship which also held in the first experimental period (r = 0.90, P > 0.01). In addition to being significantly correlated with plasma creatinine, the concentration of urea in plasma was positively associated with urea excretion rate and negatively associated with creatinine clearance rate. When the urea excretion and creatinine clearance rates were used as covariates to take account of variation between rams within lines, the difference between selection lines in plasma urea concentration was significant at P < 0.01. These effects together accounted for 88% of the variation in plasma urea concentration. Differences between the lines in plasma creatinine concentration were also significant ( P < 0.01, Table 2) when the creatinine excretion rate and urea clearance rate were fitted as covariates (model r = 86%). Neither the urea excretion rate nor the creatinine clearance rate were significantly different between the selection lines (Table 2). However, they together accounted for the line difference in plasma urea concentration. Thus line effects were non-significant when fitted after the regressions on urea excretion and creatinine clearance. When adjusted to common urea excretion and creatinine clearance rates, the mean plasma urea concentrations in control and fleeceweightselected sheep were 7.6 + 0.2 and 7.7 k 0.2 mM respectively. Period I11 The effects of a 52 h fast on plasma concentrations of urea and creatinine are shown in Fig. 3. Plasma urea concentrations declined by about 25% during the fast, but creatinine con-


centrations remained stable. In each case differences between the selection lines paralleled the corresponding differences in Period I1 (i.e. C > FW), but were not significant. Water intake exhibited a distinct biphasic pattern with peak intakes occurring 20 h and 44 h after commencement of the fast, whereas urine output declined steadily during the 52 h period. Differences between the lines in water intake and urine output were not significant. Urinary urea excretion essentially paralleled urine output, declining from 200 mmol in the first 8 h sampling interval to 100 mmol in the last. By contrast, creatinine excretion remained constant, a reflection of the fact that urinary creatinine concentration doubled during the 52 h fasting period. Over the whole period urinary urea excretion (per unit bodyweight) in the C rams (17.8 0.5 mmol kg-' day-') was significantly ( P < 0.05) greater than that in the FW rams (15.6+ 0.6 mmol kg-' day-'). Plasma urea concentration was again negatively correlated with creatinine clearance rate (r = -0.81, P < 0.01), but differences between the lines in creatinine clearance were not significant.

+

0.96"

WATER INTAKE

URINE OUTPUT

UREA EXCRETION

PLASMA

UREA

CREATlNlNE EXCRETION

0.72'

CONCENTRATION

PLASMA CREATlNlNE CONCENTRATION

1 -0.61t

UREA CLEARANCE RATE

b CREATlNlNE CLEARANCE RATE

Fig. 2. Relationships among parameters associated with urea and creatinine metabolism in Period 11. Numbers are simple correlation coefficients with significance levels. t P < 0.10. *P < 0.05. **P < 0.01. nary excretion rates. b~xpressedper unit body weight. Table 2. Water intake, urine output and plasma concentrations, urinary excretion and clearance of urea and creatinine in fleeceweight-selected and control rams (Period 11) Variable

Selection line (n = 5) Control (C) Fleeceweight (FW)

Pooled s.e.

Water intake (ml/day) Urine output (ml/day) Urea excretion (mmol kg-' day-' Creatinine excretion (rnmol kg-') Urea clearance (ml min-' kg-') Creatinine clearance (ml min-' kg-') Plasma urea concn. (mM) Plasma creatinine concn. (mM) **P < 0.01.

Period IV The twice-daily feeding regimen used in Period IV induced a biphasic pattern in plasma


concentrations of urea and, to a lesser extent, creatinine (Fig. 4). As in the previous periods, C rams maintained higher plasma concentrations of urea and creatinine than FW rams. Differences between the selected lines in plasma urea concentration were again significant (P < 0.05) using the same convariate model described for period 11, but differences in creatinine concentration were not significant. Water intake also exhibited a marked biphasic response, intakes peaking during the 4 h immediately after feeding and remaining low thereafter. Diurnal variation in urine output and the urinary excretion of urea and creatinine was less marked, and differences between the selection lines were not significant.

Field Measurements Plasma urea concentrations during fastinghefeeding in the field are shown in Table 3. Variation between rams (within lines) was generally greater than that observed during indoor feeding trials, and differences between the lines were not significant. Nor was there any significant association between plasma urea concentration in the field and period mean concentrations during the main experiment. Sulfur content of the wool in controls rams (4.40k 0.17%) was not significantly different from that in their fleeceweight-selected counterparts (4.33 k 0.18%). Within the controls, there was no relationship between sulfur content and plasma urea concentration. However, within the FW rams wool sulfur content was positively related to plasma urea concentration in all three experimental periods (correlation coefficients of 0-65, 0.90, and 0.96 in period 11, 111 and IV respectively, the last two correlations being significant at P < 0.05).

\

-

Fig. 3. Plasma concentrations of

1 I

fE

I

t

I

I

0 09

V

-&---A 0 07

Fasting period (h)

urea (upper panel) and creatinine (lower panel) in five fleeceweightselected ( -) and five control (---) rams during a fast (period 111). Bars represent standard errors about the mean.


Discussion The results of this study are consistent with the view that selection for increased greasy fleeceweight does not lead to correlated changes in digestive function (at least as indicated by crude measures such as apparent digestibility). McClelland et al. (1986) found no consistent differences between ram hoggets from these selection lines, and similar results have been obtained in studies of Australian Merino selection lines (Piper and Dolling 1969; Williams 1979). Nor was there any evidence from this study that fleeceweight-selected and control rams differ in water intake. Although fleeceweight-selected sheep have previously been reported to have high water intakes (Dolling and Carpenter 1962; MacFarlane et al. 1966) and urine volumes (McClelland et al. 1986), the mature C and FW rams maintained similar levels of each throughout the four experimental periods.

0800

1200

1600

2000

2400

0400

0800

Sampling time (hours)

Fig. 4. Plasma concentrations of urea (upper panel) and creatiand five nine (lower panel) in five fleeceweight-selected (-) control (---) rams during twice-daily feeding (Period IV). Bars represent standard errors about the mean. The daily ration was fed in two equal portions at 0800 hours and 2000 hours. Table 3. Plasma urea concentrations during fasting and refeeding in fleeceweight-selected and control rams (field measurements) Time of $ample Immediately off pasture 12 h fasting 24 h fasting 3 h refeeding 6 h refeeding

Plasma urea concentration (mM) Control (C) Fleeceweight (n = 5) (FW)

8.42 8.49 9.35 9.63 9.36

7.90 8.43 9.5 1 9.13 8.39

Pooled s.e.

0.64 0.63 0.57 0.50 0.59


The main purpose of this study was to examine factors controlling variation in plasma urea concentration. Sumner (1977) reported breed differences in plasma urea, Border Leicester lambs maintaining significantly higher concentrations than Super-Fine Merino lambs. This effect was due to a greater voluntary intake per unit metabolic body weight in the Border Leicester lambs. Conversely, Clark (1987) observed that, even when fed in proportion to metabolic body weight, mature rams and ram hoggets from the Massey University control line maintained signficantly greater urea concentrations (by about 1 mM) than those from the fleeceweight-selected line. In the present study differences were of a similar magnitude, but were significant only under controlled feeding conditions and when within-line variation could be accounted for by fitting the weight-corrected creatinine clearance and urinary urea excretion rates as covariates. High creatinine clearance rates were associated with low plasma urea concentrations in all three periods in which they were examined. Creatinine clearance rate is widely used as a measure of the rate at which plasma is filtered by the kidneys (i.e. the glomerular filtration rate). Animals with high glomerular filtration rates would be expected to have low plasma concentrations of substances such as creatinine and urea which are not actively secreted or reabsorbed in the nephron, because if the glomerular filtration rate is high, more of these substances will be filtered and removed from plasma. This presumably accounts for the significant correlation between plasma concentrations of urea and creatinine, particularly in Periods I1 and I11 when changes in plasma concentrations were relatively.slow (i.e. when discrete plasma samples adequately represented mean daily concentration).

-

0

A

A

Fig. 5. Relationship between plasma urea concentration (corrected for creatinine clearance rate) during 12 x daily feeding (Period 11) and nitrogen balance (Period I) in control ( A ) and fleeceweightselected (@) rams.

A

-

0

7

8

9

Corrected plasma urea concentration (mM)

High weight-corrected urinary urea excretions were associated with high plasma urea concentrations in Periods I1 and IV when the rams were fed in proportion to metabolic body weight ( ~ e i ~ h t ~Since " ~ ) .there was little variation in digestibility of dietary nitrogen, and hence in the amount of nitrogen absorbed, this implies that rams with high plasma urea concentrations excreted a greater proportion of their nitrogen intake as urea than did those with low plasma concentrations. Moreover, urinary urea excretion was highly correlated with total urinary nitrogen excretion (r = 0.93, P < 0.01), suggesting that rams with high plasma urea concentrations would have had an increased ratio of urinary nitrogen excretion : nitrogen intake (i.e. a lower nitrogen retention). Hence the relationship between plasma urea concentration (Period 11) and components of nitrogen balance data from Period I was examined. Plasma urea concentrations were corrected for creatinine clearance rate, so removing a component of variation in plasma urea independent of that associated with urinary urea excretion. The correlation between corrected plasma urea concentration and nitrogen retention (Fig. 5) was


r = -0.85 (P < 0.01). Thus the high plasma ureas of some rams appear to reflect increased deamination of amino acids consequent upon their inability t o retain a high proportion of dietary nitrogen in wool and/or body tissues. During fasting, FW rams maintained lower concentrations of urea in plasma than C rams and had significantly lower rates of weight-corrected urinary urea excretion. Thus low plasma urea concentrations appear to be associated with low rates of urea excretion both when rams are fed and when they are fasted. No differences was observed between the C and FW rams in the sulfur content of wool. This is in contrast to the results of studies with Australian Merino lines (McGuirk 1979). It also leaves unexplained the observation that plasma urea is depressed to a greater extent in C than in FW rams when intravenous methionine is administered (Clark 1987). However, the wool samples were taken 10 weeks after the main experiment when the rams were at pasture. Further studies of wool sulfur content under controlled conditions will be required, particularly given the observation that this trait was positively correlated with plasma urea concentration in the fleece weight-selected rams. This study has confirmed that the measurement of plasma concentration has potential utility as a means of predicting genetic merit for fleeceweight in Romney sheep. However, discrimination between the lines on the basis of plasma urea concentration is best achieved under conditions which approach continuous feeding. Thus measurement of plasma urea could not be used for large-scale screening of sheep in the field. It is therefore important that the physiological basis for differences in plasma urea be established with a view to identifying genetic markers for fleeceweight which are less sensitive to environmental effects. Results obtained here provide preliminary evidence that between-line differences in plasma urea can be attributed to the combined effects of variation in weight-corrected urinary excretion rates and creatinine clearance rates. More extensive trials will be required to detect significant between-line differences in these parameters, and to establish their physiological relationship with fleece production. Acknowledgments

The authors gratefully acknowledge Mr W. B. Parlane (technical assistance), Dr R. M. Greenway (assay of urea and creatinine) and Dr M. J. Hedley (assistance with assay of wool sulfurs). References Blair, H. T. (1986). Response to 27 years of selection for greasy fleeceweight. Proc., 3rd World Congr. on Genetics Applied to Livestock Production, Vol. 12, pp 215-20. Chasson, A. L., Grady, A. J., and Stanley, M. A. (1961). Determination of creatinine by means of automatic chemical analysis. Am. J. Clin. Path. 35, 83-88. Clark, C. M. (1987). Physiological responses to selection for greasy fleeceweight in Romney sheep. M. Agr. Sc. Thesis, Massey University, Palmerston North. Dolling, C. H. S., and Carpenter J. T. (1962). Water consumption at pasture of Merino sheep selected for high wool production. Proc. Aust. Soc. Anim. Prod. 4, 172-4. Gilmour, A. R. (1985). REG: A generalized linear models programme. Misc. Bull. No. l., Div. Agric. Serv. N.S.W. Dept. Agric. Landers, D. H., David, M. B., and Mitchell, M. J. (1983). Analysis of organic and inorganic sulfur constituents in sediments, soils and water. Znt. J. Environ. Anal. Chem. 14, 245-56. MacFarlane, W. V., Dolling, C. H. S., and Howard, B. 1966. Distribution and turnover of water in Merino sheep selected for high wool production. Aust. J. Agric. Res. 17, 491-502. McClelland, L. A,, Blair, H. T., Wickham, G. A., and Brookes, I.M. (1986). A comparison of fleeceweight selected and control Romney rams for intake, liveweight gain, wool growth and feed utilization. Proc. N.Z. Soc Anim. Prod. 46, 215-18. McClelland, L. A,, Wickham, G. A., and Blair, H. T. (1987). Efficiency of Romney hoggets from a fleeceweightselected flock. Proc. 4th Anim. Sci. Congr. Asian-Australas. Assoc. Anim. Prod. SOC.:330 (Abstr.).


McGuirk, B. (1979). Selection for wool production in Merino sheep. I n 'Selection Experiments in Laboratory and Domestic Animals'. (Ed. A. Robertson.) pp 176-97. (Commonwealth Agric. Bureaux: Farnam Royal, U.K.) Marsh, W. H., Fingerhut, B., and Muller, H. (1965). Automated and manual direct methods for the determination of blood urea. Clin. Chem. 11, 624-27. Piper, L. R., and Dolling, C. H. S. (1969). Efficiency of conversion of food to wool. V. Comparison of the apparent digestive ability of sheep selected for high clean wool weight with that of sheep from a random control group. Aust. J. Agric. Res. 20, 579-87. Rattray, P. V. (1986). Feed requirements for maintenance, gain and production. In 'Sheep Production 11: Feeding, Growth and Health'. (Eds S. N. McCutcheon, M. E McDonald and G. A. Wickham.) Ch. 5. (New Zealand Institute of Agricultural Science.) Sumner, R. M. W. (1977). Nutritional factors affecting production in Border Leicester and Super-Fine Merino sheep. Ph.D. Thesis, University of New England, Armidale. Williams, A. J. (1979). Speculations on the biological mechanisms responsible for genetic variation in the rate of wool growth. In 'Physiological and Environmental Limitations to Wool Growth'. (Eds J. L. Black and P. J . Rees.) pp. 337-54. (University of New England Publishing Unit: Armidale, N.S.W.) Manuscript received 7 October 1986, accepted 25 May 1987