Hypertonic fluids are secreted by medial and lateral segments in duck (Anas platyrhynchos) nasal salt glands

Abstract

Indwelling catheters were used to collect fluid directly from the medial and lateral segments of duck nasal salt glands showing, for the first time, that the secretions are fully hypertonic before reaching the medial and lateral drainage ducts. Using this method it was possible to show that (a) there is a functional symmetry between the left and right salt glands, (b) the medial segment always secretes fluid at approximately twice the rate of the lateral segment and (c) fluid secreted by the medial segment has the same ionic composition but variable ion concentrations when compared with fluid from the lateral segment. A 12 % increase in post-segmental fluid osmolality was probably due to the evaporation of water from epithelial surfaces in the nasal cavities during breathing. A post-segmental outflux of Ca²⁺, Mg²⁺ and Cl⁻ in the medial and lateral collecting ducts and/or nasal epithelium may be of adaptive significance when birds inhabit calcium- and magnesium-rich marine environments.

Avian nasal salt glands are crescent-shaped organs that lie in shallow depressions in the frontal bones above the eyes in birds such as gulls, fulmars and petrels. In other birds they are found along the border of the frontal bone within the upper orbital membrane (curlews, geese and ducks), under the nasal bone (falcons and bustards), along the anterior border of the eye orbit (parrots), or under the eye beneath the orbital membrane (woodpeckers) (Marples, 1932). Schmidt-Nielsen and his colleagues were the first to show that supra-orbital glands in herring gulls (Larus argentatus) secrete hypertonic NaCl solutions (Fange et al. 1958; Schmidt-Nielsen, 1960) which enable these and other marine birds to drink brackish or full seawater, secrete salt and conserve osmotically free water. In Pekin ducks (Anas platyrhynchos) each salt gland is formed of separate medial and lateral segments (Butler et al. 1991) which have different embryonic origins (Jobert, 1869; Marples, 1932). This is of added interest because the medial segment in ducks is homologous with both the lateral nasal gland of reptiles and Stenson's gland in mammals whereas the lateral segment is a similar, but separate, septal nasal gland. These medial and lateral segments are compound tubular glands formed of two cell types. The first of these are cuboidal, peripheral cells located near the end of each secretory tubule. They are relatively small cells, often triangular in shape, lacking basal infoldings and having only a small number of mitochondria. The second type are principal cells which are found along the entire length of the secretory tubule (Peaker & Linzell, 1975). These relatively tall, mitochondria-rich cells possess lateral infoldings which form interstitial channels leading to the apical margin where tight junctions are formed with adjacent cells (Ernst et al. 1980). The plasma membrane at the base of the principal cell forms interdigitating processes that rest on the basal lamina (Holmes, 1972). It has been proposed that basolateral Na⁺-K⁺-ATPase provides energy for the co-transport of Na⁺ and Cl⁻ through the folded basal membranes. Chloride then moves passively through the apical membrane and Na⁺ flows between the principal cells, through the tight junctions and into the tubular lumen. Water movement may follow solutes via the cellular or paracellular route to yield a hypertonic fluid composed primarily of NaCl together with smaller amounts of K⁺ and traces of other ions (Ernst & Van Rossum, 1982; Lowy et al. 1989). Several other models have been proposed for the tubular secretion of hypertonic fluids (Butler et al. 1989). To date there have been no reports of differences in the length of secretory tubules or the cells within these tubules in the medial and lateral segments of duck salt glands. A mass of secretory tubules forms a compound tubular unit, the secretory lobule. Secretory lobules are arranged radially around, and drain into, the medial and lateral ducts, which pass along the entire length of each segment (Butler et al. 1991). When the medial and lateral ducts leave their respective segments, they become the medial and lateral collecting ducts (Fig. 1) The medial collecting duct empties into the nasal cavity at the base of the vestibular concha whereas the lateral collecting duct forms a loop on the surface of the nasal septum and opens there.

Figure 1. Diagram of a Pekin duck (Anas platyrhynchos) head, illustrating the medial and lateral segments of the right nasal salt gland.

Catheters were inserted into the medial and lateral drainage ducts, advanced to the edge of each segment and tied in place. MS, medial segment; LS, lateral segment; MD, medial collecting duct; and LD, lateral collecting duct.

The present experiments were designed primarily to show whether the hypertonic secretions from the salt glands are fully elaborated within the glands or within the external ducts that drain them.

METHODS

Animals

White Pekin drakes (Anas platyrhynchos) were purchased from King Cole Duck Farms in Aurora, Ontario, Canada, when they were 8 weeks old. They were housed in the Department of Zoology, University of Toronto under a cycle of 12 h light−12 h dark and fed commercial duck grower feed mixed with freshwater and 0.9 % NaCl drinking water ad libitum for a further 2 weeks before they were used for the experiments. At that time they weighed between 3.4 and 3.7 kg. All experimental work was done in accordance with the terms of an Animal Use Protocol issued by the University of Toronto Animal Care Committee.

Experiments

Test for accuracy of nasal fluid collections

Nasal fluid from the left salt gland was routinely collected into 100 ml Pyrex beakers washed in distilled water and dried. This beaker method was compared to a second method wherein fluid was carefully withdrawn into a 1 ml plastic hypodermic syringe (23 gauge hypodermic needle fitted with a 1 cm length of PE-50 polyethylene tubing) as it dripped from the edge of the nostril. Losses of fluid were negligible. These two methods of collection were compared as follows: samples 1, 3, 5, 7 and 9 were collected using a beaker and samples 2, 4, 6, 8 and 10 were collected using a syringe in each of six ducks. Flow rates, and Na⁺ and K⁺ concentrations were compared for each collection method for all samples (n = 30).

Functional symmetry of the nasal salt glands

This was measured by collecting fluid samples from the left and right medial segments concurrently. Ten successive 5 min fluid samples from the right and left segments were collected from each of 10 ducks. Flow rates and Na⁺ and K⁺ concentrations were determined for each sample and regressions were fitted to values for left (X) vs. right (Y) segments.

Measurement of ion concentrations and rates of secretion by the right medial and lateral segments and the left nostril (LSG) of duck salt glands

Secreted fluids were collected simultaneously from the right medial segment (RMS) and right lateral segment (RLS) catheters and the left salt gland (nostril) during seven or eight successive 5 min periods in seven ducks. RMS and RLS fluids were collected in 3 ml Pyrex tubes and fluid from the left nostril drained into a clean 100 ml Pyrex beaker held at a slight angle without significant losses of fluid. There was no measurable water loss by evaporation during these collection periods. Fluid was withdrawn from the tubes or beakers into preweighed 3 ml hypodermic syringes fitted with size 23 G hypodermic needles. Volumes were estimated by weight difference. Samples were then transferred to 1.5 ml conical Eppendorf tubes and stored at −20 °C.

Surgical preparation

Each duck was placed on its right side on a holding board with its left wing extended and held in place with rubber tubing. Its head was covered with a cotton towel so it remained calm and quiet. The left ulnar region was injected with lidocaine hydrochloride (Xylocaine, Astra Pharmaceuticals, Montreal, Canada). A suitable length of heparin-filled (Hepalean, 1000 USP units ml⁻¹, Organon Teknika, Toronto, Canada) PE-50 polyethylene tubing (Intramedic, i.d. 0.58 mm, o.d. 0.97 mm, Clay Adams, NJ, USA) was inserted into the left ulnar vein, tied in place with 5–0 surgical silk and used as an infusion catheter. A second length of heparin-filled PE-50 tubing was inserted into the right brachial vein for use as a blood withdrawal catheter. The access incisions in both wings were closed with 5 mm Michel clips. Next, a general anaesthesia was induced with equithesin (3 ml kg⁻¹i.v.) delivered via the left ulnar catheter. An incision was made through the skin from the anterior edge of the salt gland to the tip of the triangle formed by orange beak skin, then extended along both sides of the triangle, between beak skin and feathered skin, for a distance of approximately 1.5 cm. A second 1.5 cm incision was made along the midline of the triangle (see Fig. 1). The cut edges of skin were separated from underlying connective tissue and retracted to reveal a ridge of dense prefrontal bone. It was removed to expose the underlying spongy bone which was then cut away to reveal a length of the medial and lateral drainage ducts. The ducts are opaque, covered with well vascularized connective tissue and run parallel to, and near, the lateral border of the olfactory nerve (Butler et al. 1991). The outer diameter of the medial drainage duct always appeared to be twice that of the smaller lateral duct. Therefore the medial duct was always much easier to locate and to fit with a collection catheter. Each duct was cut open about half-way along its length and a size PE-10, polyethylene catheter (i.d. 0.28 mm, o.d. 0.61 mm) was inserted into its lumen and pushed forward gently until it touched the medial or lateral segment of the right salt gland. There was virtually no bleeding or damage to adjacent tissues. Each catheter was held in place exactly at the opening of the gland segment with size 5–0 silk ligatures so that fluid drained directly from the medial and lateral segments without ever coming into contact with the medial or lateral collecting duct or the nasal epithelia. Ampicillin (50 mg ml⁻¹, penbritin-500, Ayherst Labs, Montreal, Canada) was applied to the surgical site, which was then covered with sterile gauze and bound with adhesive tape. Next, the catheters were trimmed to the correct length and taped to the surface of the beak with masking tape. This prevented ducks from pulling out the catheters during the 2 day period before the collection experiments began. Ducks regained consciousness within an hour after surgery was completed and they were eating well and showing no signs of discomfort the day after surgery. Droplets of nasal fluid were seen at the end of the two catheters and at the left nostril. The masking tape was removed shortly before nasal fluid collections were started and replaced at the end of each experiment. When the experiments were completed the ducks were killed with an overdose of sodium pentobarbital and autopsied to show that the catheters were placed exactly at the edge of the medial and lateral segments as shown in Fig. 1.

Collection of nasal fluid samples

Ducks were given a 1 day recovery period following the surgical implantation of catheters. At the start of the experiment, the duck was placed on a holding board. If noise and movement were kept to a minimum it remained calm throughout the experiment and there was little, if any, struggling. Nasal gland secretion was induced with a bolus injection of 12 ml kg⁻¹ of 1000 mosmol (kg H₂O)⁻¹ NaCl solution into the left ulnar vein and sustained with a constant infusion of the same solution at a rate of 0.30 ml (kg bw)⁻¹ min⁻¹ for the entire collection period.

Collection of blood samples

Blood samples (1 ml) were withdrawn from the right ulnar vein before the injection/infusion of 1000 mosmol (kg H₂O)⁻¹ NaCl solution and again at the onset of nasal secretion. They were centrifuged at 4000 g (-2 °C) for 10 min whereupon the plasma samples were transferred to 1.5 ml Eppendorf tubes and stored at −2 °C until Na⁺, Cl⁻ and osmolal concentrations were determined by flame photometry, chloridometry and freezing point depression, respectively.

Concurrent measurements of nasal fluid secretion from the left and right medial segments to test for functional symmetry

Approximately 10 nasal fluid samples were collected concurrently from the left medial (Y axis) and right medial (X axis) collecting ducts of 10 ducks giving a total of 99 paired measurements. The relationships between right medial and left medial samples were plotted as linear regressions.

Analytical methods

Plasma and nasal fluid samples were thawed and diluted appropriately. Sodium and potassium concentrations were measured with a flame photometer (Instrumentation Laboratory S.p.A., model 943, Milan, Italy); chloride concentrations by a coulimetric method using a Haake Buchler digital chloridometer (Buchler Instruments Inc., NJ, USA). Total Ca and Mg concentrations were measured with an atomic absorption spectrometer (model 351; Instrumentation Laboratory, MA, USA) after the samples were diluted with 1 % lanthanum chloride to reduce interference from phosphate. Finally osmolality was measured by the freezing point method using a micro-osmometer (model 3MO, Advanced Instruments, MA, USA).

Statistical methods

Regression lines were fitted using a curve-fit program (Statgraphics Plus, Version 3.0; Manugistics, Rockville, MD, USA). As many as eight points for each of seven ducks were plotted for the regressions. To exclude any effects of pseudo-replication, 5 degrees of freedom were used to determine the statistical significance of the correlation coefficient r. Data in Tables 1 and 2 are expressed as arithmetic means ± s.e.m. Sample groups in Table 1 were compared with a one-factor analysis of variance (ANOVA) and if significant differences occurred, multiple comparisons were made using Fisher's method of least squares, also with Statgraphics Plus. The fiduciary limit was set at P = 0.05.

Table 1.

Fluid and electrolyte secretion by the medial and lateral segments of the right nasal salt gland and by the left nasal salt gland of Pekin ducks (Anas platyrhynchos)

		Right nasal salt gland			Left nasal salt gland
Rate of secretion		A. Medial segment	B. Lateral segment	C. Medial + lateral segments	D. Left nostril
Nasal fluid	(ml kg−1 h−1)	4.58 ± 0.18†‡§	2.01 ± 0.11 ‡§	6.53 ± 0.26 §	5.76 ± 0.24
Na+	(mmol l−1)	559 ± 7.4 §	537 ± 6.6 §	—	601± 9.5
	(μmol kg−1 h−1)	2531 ± 88.5†‡§	1052 ± 49.1 ‡§	3583 ± 131	3392 ± 131
K+	(mmol l−1)	10.57 ± 0.22 §	11.04 ± 0.25 §	—	12.15 ± 0.3
	(μmol kg−1 h−1)	48.09 ± 2.02 †‡§	21.66 ± 1.13 ‡;§	69.75 ± 3.01	68.25 ± 2.95
Cl−	(mmol l−1)	549 ± 7.6 §	536 ± 8.3 §	—	599 ± 9.9
	(μmol kg−1 h−1)	2488 ± 87.6 †‡§	538 ± 8.36 ‡§	3026 ± 90.4 §	3367 ± 125
Ca	(mmol l−1)	0.46 ± 0.03 §	0.39 ± 0.02 §	—	0.62 ± 0.02
	(μmol kg−1 h−1)	2.24 ± 0.21†‡§	0.74 ± 0.01‡§	2.98 ± 0.22	3.41 ± 0.15
Mg	(mmol l−1)	0.11 ± 0.01 §	0.09 ± 0.005 §	—	0.28 ± 0.03
	(μmol kg−1 h−1)	0.47 ± 0.04†‡§	0.17 ± 0.01 ‡§	0.64 ± 0.05 §	1.48 ± 0.13
Osmoles	(mosmol l−1)	985 ± 15.1 §	966 ± 14.9 §	—	1072 ± 19.1
	(μosmol kg−1 h−1)	4471 ± 176 †‡§	1903 ± 91.7 ‡§	6374 ± 255	6524 ± 264

Nasal fluid samples were collected from the medial and lateral segments via indwelling catheters to bypass the drainage ducts and nasal cavity. Fluid from the left salt gland was collected as it flowed from the nostril. Values are means ± s.e.m. for 55 samples (7–8 samples from each of 7 ducks).

^†P < 0.05 compared with group B lateral segment.

^‡P < 0.05 compared with group C (A+B).

^§P < 0.05 compared with group D collection from the left nostril using Fisher's least significant difference procedure.

Table 2.

Plasma electrolyte and osmolal concentrations in Pekin ducks (Anas platyrhynchos) before salt loading and at the onset of nasal fluid secretion

	Na+ (mmol l−1)	K+ (mmol l−1)	Cl− (mmol l−1)	Ca (mmol l−1)	Mg (mmol l−1)	Osmolality (mosmol kg−1)
Before salt loading	150 ± 1.9	2.87 ± 0.14	110 ± 2.1	2.6 ± 0.06	0.78 ± 0.03	295 ± 3.1
Onset of secretion	157 ± 1.5 *	2.9 ± 0.12	124 ± 2.7 *	2.41 ± 0.08	0.72 ± 0.03	312 ± 4.5 *

Values are means ± s.e.m. Means were compared using Student's t test.

^*P < 0.05 compared with ducks before salt loading; n = 7 ducks.

RESULTS

Accuracy of the method of nasal fluid collection

The flow rate from the left salt gland (nostril) was 0.47 ± 0.02 ml (kg bw)⁻¹ (5 min)⁻¹, [Na⁺] was 478 ± 32 mmol l⁻¹, and [K⁺] was 13.2 ± 0.30 mmol l⁻¹ using the syringe method (n = 30) compared with a flow rate of 0.48 ± 0.02 ml (kg bw)⁻¹ (5 min)⁻¹, [Na⁺] of 484 ± 26 mmol l⁻¹, and [K⁺] of 12.49 ± 0.40 mmol l⁻¹ using the beaker method (n = 30). Data collected using the syringe method were not significantly different from data collected using the beaker method.

Concurrent secretion from the medial segments of the left and right salt glands as evidence of functional symmetry

Concurrent samples were collected directly from the left (Y) and right (X) medial segments (Fig. 1). Flow rates and Na⁺ and K⁺ concentrations in the left and right medial segments were nearly equal. Linear regressions were plotted as follows: flow rates (ml (kg bw)⁻¹ h⁻¹), Y = 1.00X plus; 0.024; r = 0.83; [Na⁺] (mmol l⁻¹), Y = 0.87X plus; 76.6; r = 0.75; and [K⁺] (mmol l⁻¹), Y = 0.97X plus; 0.92; r = 0.84. None of these slopes differed significantly from 1.0.

Rates of nasal fluid secretion

Figure 2 shows the relationship between the rate of flow from the left nostril (LSG) and the rate of flow from the right medial segment (RMS) plus right lateral segment (RLS) (Y = 0.75X plus; 0.84; r = 0.80). Out of a total of 56 points, 38 fell below, 4 on and 12 above the line of equivalence indicating that post-segmental water reabsorption/evaporation occurred in 69 % of the samples, an average of 12 % (Table 1). The slope of the regression was significantly (P < 0.01) less that 1 (Student's t test, t = 3.27 calculated as slope - 1/standard error of the regression). The mean rate of fluid secretion from the RLS was 41 % of that from the RMS. This relationship was constant over the entire range of flows (Fig. 3).

Figure 2. Rate of fluid secretion (ml kg⁻¹ h⁻¹) from the right medial segment (RMS) + the right lateral segment (RLS) vs. the rate of flow from the left nostril (LSG) of Pekin ducks.

n = 55 samples collected from seven ducks plotted in relation to a line of equivalence.

Figure 3. Rate of fluid secretion (ml kg⁻¹ h⁻¹) from the right medial segment (RMS) or right lateral segment (RLS) relative to total secretion from both segments of Pekin duck salt glands.

n = 55 samples collected from seven ducks.

Electrolyte and osmolal concentrations in RMS fluids compared with RLS fluids

Ion and osmolal concentrations in RMS fluid samples were plotted against those in RLS fluid samples collected concurrently: Na⁺ (Y = 0.45X plus; 282.8; r = 0.51); Cl⁻ (Y = 0.72X plus; 141.8; r = 0.66), Ca (Y = 0.11X plus; 0.34; r = 0.16), Mg (Y = 0.38X plus; 0.05; r = 0.52), and mosmoles (Y = 0.57X plus; 406.0; r = 0.57). None of these relationships were statistically significant indicating that the medial and lateral segments secrete fluids simultaneously but with measurably different ion and uncorrelated concentrations. Only K⁺ concentrations in the RMS and RLS fluids were significantly correlated and therefore, predictably, the same (Y = 0.81X plus; 2.46; r = 0.73; P < 0.05 using 5 degrees of freedom). The relative differences in ion and osmolal concentrations are not evident when the average ion and osmolal concentrations in all RMS fluid samples were compared with all of those in the RLS fluid samples (Table 1).

Comparison of electrolyte and osmolal concentrations in RMS fluids with LSG fluids

Table 1 shows that the mean concentrations of Na⁺, K⁺, Cl⁻, Ca, Mg and mosmoles were all significantly greater (P < 0.05) in LSG fluid than in RMS fluid.

Total electrolyte and osmolal excretion from the RMS plus RLS compared with the LSG

Rates of electrolyte and osmolal excretion in fluid collected from the left nostril (LSG) were plotted vs. the rates of excretion from the RMS + RLS in concurrent samples (Fig. 4). The following number of points were above the lines of equivalence: Na⁺, 23 (41 %); K⁺, 24 (43 %); Cl⁻, 35 (63 %); Ca, 38 (68 %); Mg, 44 (79 %); and mosmoles, 21 (38 %). This suggested that the rate of Ca, Mg and Cl⁻ excretion from the left salt gland via the nostril is greater than from the right medial plus lateral segments. When the rates of excretion were averaged for all samples, Mg and Cl⁻ but not Ca were secreted at a higher rate via the left nostril (LSG) than via the RMS + RLS (P < 0.05).

Figure 4. Secretion of electrolytes and mosmoles by the right medial segment (RMS) + the right lateral segment (RLS) vs. the left nostril (LSG) of Pekin ducks.

n = 55 samples collected from seven ducks plotted in relation to a line of equivalence.

Plasma electrolyte and osmolal concentrations

Both were measured before and immediately after the onset of nasal secretion in each of the seven ducks used in the present study. Plasma Na⁺, Cl⁻ and osmolal concentrations increased significantly (P < 0.05) at the onset of nasal fluid secretion (Table 2).

DISCUSSION

Evaluation of methods for nasal fluid collection

In earlier studies of avian salt glands, fluid was collected in a beaker as it dripped from the nostrils (Holmes et al. 1961; Butler, 1984, 1985, 1987; Butler et al. 1989) onto washed cotton wool held over the nostrils (Wright et al. 1966), or through tubing by aspiration (Hanwell et al. 1971; Kaul et al. 1983). Water evaporation may have led to errors in the measurement of electrolyte concentrations. Therefore, results using the beaker method were compared with those obtained following the direct withdrawal of nasal fluid into preweighed 3 ml plastic syringes fitted with 23 gauge hypodermic needles and 10 mm lengths of PE-50 tubing. There was no statistically significant difference between fluid volumes and ion concentrations measured with each of these methods so the beaker method was used throughout.

Functional symmetry in the left and right salt glands

Experimental evidence for the post-segmental flux of electrolytes and water is based on the assumption that there is a functional symmetry between the left and right salt glands. To test this assumption, fluid was collected concurrently from indwelling catheters in the left (X) and right (Y) medial collecting ducts. When the lines of best fit for flow rate, [Na⁺] and [K⁺] were determined none of the slopes differed significantly from zero thereby demonstrating functional symmetry between the left and right salt glands.

Dehydration of post-segmental fluid

Figure 2 illustrates the relationship between the rate of secretion of fluid from the medial plus lateral segments and the concurrent secretion of fluid from the left nostril. It is possible but unlikely that water is reabsorbed along the length of the medial and lateral collecting ducts (Butler et al. 1991) because the segmental fluids are almost fully hypertonic before leaving the salt glands. Water evaporation in the nasal passages during breathing provides the most likely explanation for the observed 12 % increase in post-segmental osmolality.

Evidence for the formation of hypertonic fluids within the medial and lateral segments

Marshall et al. (1985) used X-ray microanalysis to measure the distribution of electrolytes in frozen sections of Pekin duck nasal salt glands. Ion concentrations in the lumenal fluid within secretory tubules were as follows (mmol l⁻¹): Na⁺, 73 ± 15; Cl⁻, 98 ± 18; K⁺, 27 ± 6; Mg, 10 ± 4; Ca, 11. This lumenal fluid was slightly hypotonic relative to duck blood plasma. Ion concentrations were also measured at the point where fluid drains from the salt gland and enters the ‘main duct system running from the glands to the external nares’ or what we refer to as the medial or lateral collecting ducts (Butler et al. 1991). At the point of egress from the gland proper, ion concentrations were as follows (mmol l⁻¹): Na⁺, 118 ± 11; Cl⁻, 143 ± 13; K⁺, 12 ± 2; Mg, 7 ± 1; Ca, 4 ± 1. Therefore, Marshall et al. (1985) concluded, ‘nasal fluid becomes increasingly hyperosmotic in the duct system, most concentration occurring in the main ducts running from gland to external nares’. These observations may have been affected negatively by (a) administration of the anaesthetic sodium pentobarbital and its known attenuation of secretion and (b) changes during the surgical excision of salt glands from anaesthetized ducks, subsequent preparation of tissue slices and freezing in liquid nitrogen prior to X-ray probe analysis. The present experiments show that nasal fluid collected from conscious ducks was fully hypertonic as it emptied from the medial and lateral segments. This segmental fluid was modified only slightly as it flowed through the main collecting ducts, into the nasal passages and out through the nostrils (Table 1).

Comparison of function of the medial and lateral segments

There are no evident differences in tubule length or in types and distribution of secretory cells within the medial and lateral segments even though they have different embryonic origins. It was of interest to learn if they function differently by secreting fluids at different rates and/or with different ionic concentrations. Retrograde injections of coloured silicon rubber into the medial and lateral drainage ducts and segments showed that the medial segment was always approximately twice the size of the lateral segment in both the left and right duck salt glands (Butler et al. 1991; also Fig. 1). Therefore it was not surprising to find that the average flow rate from the RMS (4.86 ± 0.18 ml kg⁻¹ h⁻¹) was slightly more than twice the average flow rate from the RLS (2.01 ± 0.11 ml kg⁻¹ h⁻¹; Table 1). This relationship was constant over the entire range of secretion rates showing that there were no independent adjustments of flow rates in either the medial or lateral segments (Fig. 3). It should be noted that the rate of NaCl infusion (0.30 ml kg⁻¹ min⁻¹) of 1000 mosmol (kg H₂O)⁻¹ NaCl was not balanced by the secretion (0.2 ml kg⁻¹ min⁻¹) of hypertonic fluids (650 mosmol (kg H₂O)⁻¹) that are primarily NaCl. Therefore the salt glands were operating at maximum secretory capacity. At non-maximal capacity the relative contributions of the medial and lateral segments may be different.

Na⁺, K⁺, Cl⁻, Mg, Ca and osmolal concentrations in RMS fluids were plotted against those in RLS fluids using a linear regression model (Fig. 4). The large degree of scatter showed that the medial and lateral segments simultaneously secrete fluids with different ion and osmolal concentrations. These relatively minor and variable differences in concentration between the two fluids were masked when the average ion and osmolal concentrations were calculated for 55 medial and lateral fluid samples from all of the seven ducks (Table 1).

Ionic and osmolal concentrations were always significantly greater in the fluid collected from the left nostril than in fluid collected directly from the right medial segment (Table 1). These elevated ion concentrations in LSG fluid (nostril) may be due largely to the approximately 12 % post-segmental loss of water (Table 1). Nevertheless, it was also important to show if there was a significant post-segmental secretion of ions.

Mammalian salivary glands first secrete an isoosmotic fluid. NaCl is then reabsorbed by the collecting ducts as it drains from the glands thereby lowering the NaCl concentration to about 5 mmol l⁻¹ (Sanford, 1992). In this way salt is reclaimed from the more that 1 litre of saliva formed daily in humans. Thus it seemed possible that salt gland collecting ducts secrete ions into the post-segmental fluid as it drains. Post-segmental ion secretion could be measured by comparing the rates of ion secretion from the RMS + RLS with the rates of secretion from the left nostril (LSG). Table 1 shows that rates of Na⁺, K⁺ and osmolal secretion from the RMS + RLS and LSG were nearly equal whereas there was a borderline (0.05 > P < 0.10) increase in Ca excretion and a clearly significant increase in the rate of secretion of both Mg²⁺ and Cl⁻ (P < 0.05). Moreover, Fig. 4 shows that when the rate of secretion of Na⁺, K⁺ or mosmoles from the left nostril (LSG) was plotted against the rate of secretion from the RMS + RLS there was a fairly even distribution of points above and below the line of equivalence. Yet the majority of the Ca (68 %), Mg (82 %) and Cl⁻ (63 %) points fell above the equivalence lines illustrating post-segmental secretion. This capacity to secrete divalent ions and Cl⁻ may be an important adjunct to the primary secretion of Na⁺ and K⁺ and Cl⁻ within the segments and may be of adaptive significance during adaptation to marine environments where dietary intakes of these ions are relatively large. Salt glands are known to secrete fluids with relatively high calcium concentrations albeit in a terrestrial bird, the ostrich (Struthio camelus), in response to dehydration (Schmidt-Nielsen et al. 1963).

Plasma Na⁺ (4.8 %), Cl⁻ (12.7 %) and osmolality (5.8 %) increased significantly (P < 0.05) at the onset of nasal secretion (Table 2). These changes are similar to those observed at the start of secretion in other experiments with domestic ducks even though the NaCl loads and routes of delivery differed (Holmes et al. 1961; Butler, 1985). The correlation between increased plasma osmolality and avian salt gland secretion was first observed in gulls more than 40 years ago (Fange et al. 1958).

Finally, by using indwelling catheters it has been possible to collect fluid directly from the medial and lateral segments of duck nasal salt glands and to show that the primary concentrating power of salt glands lies within secretory tubules in the medial and lateral segments. Physiologists may now use this experimental methodology to make more accurate measurements of salt gland function leading eventually to a better understanding of the way in which hypertonic fluids are formed.