Gastrointestinal and renal responses to variable water intake in whitebellied sunbirds and New Holland honeyeaters

Bird nectars are generally dilute (Baker et al., 1998; Johnson and Nicolson, 2008; Nicolson, 2002; Pyke and Waser, 1981), dramatically influencing the physiology of nectarivores, which must consume large volumes of water to satisfy their energy requirements (Martínez del Rio et al., 2001; Nicolson and Fleming, 2003b). When birds feed on dilute nectar, they can consume up to five times their body mass in water daily (Collins, 1981; McWhorter and Martínez del Rio, 1999; Nicolson and Fleming, 2003a). These massive ingested water loads can potentially cause severe disruptions in water balance (Beuchat et al., 1990; McWhorter et al., 2003). Nectarivores also face a constant challenge in terms of fluctuations in water balance, having to switch between avoiding water loading and dehydration as they switch between food plants or between feeding bouts and fasting periods. During fasts (overnight or during disturbance during the day, e.g. due to storms), these birds do not feed and therefore have no water intake. Regulating osmotic balance requires that these birds be able to deal with both extremes (water-loading and dehydration) on a daily basis.

The kidneys are among the most metabolically active tissues in the vertebrate body. They consume a disproportionate amount of a vertebrate's daily energy expenditure to carry out water and waste excretion while ensuring that blood glucose and electrolyte balances are maintained (Silverthorn, 2004). We predict that the metabolic costs of kidney function will be especially high in nectarivorous animals, due to the high preformed water loads of their nectar diet. One way to avoid this high renal metabolic load would be to not absorb all preformed water from the intestine, instead shunting some of the excess water directly through. This was proposed as the ‘intestinal shunting hypothesis’, predicting that birds feeding on large volumes of dilute nectar could reduce the water load to be processed by the kidneys (renal loading) by reducing intestinal water absorption (fractional water absorption; f_A) (Beuchat et al., 1990). This intestinal shunting hypothesis has been examined for two hummingbird species to date, including broad-tailed hummingbirds, Selasphorus platycercus (McWhorter and Martínez del Rio, 1999), and green-backed firecrowns, Sephanoides sephanoides (Hartman Bakken and Sabat, 2006). These hummingbird species absorb ~80% and ~90% (respectively) of the water ingested; however, fractional water absorption was not correlated with dietary water intake. By contrast, a similar study in Palestine sunbirds (Cinnyris oseus) demonstrated a significant correlation between f_A and dietary preformed water intake, suggesting that these birds are able to regulate their absorption of water in relation to the amount of water consumed: as water intake increased, f_A decreased (McWhorter et al., 2003). These data suggest that there are interesting differences in the handling of water loads between these nectarivore lineages.

A second way to reduce renal metabolic costs of electrolyte and glucose retrieval may be to reduce glomerular filtration rate (GFR). Although this has not been found for feeding nectarivorous birds, reduction in renal water reabsorption (f_R) in response to increased water excretion has been recorded (McWhorter et al., 2004). Another way to avoid high renal metabolic load would be to shut down the kidneys because renal processing is not required when the birds are not feeding (i.e. overnight). Both hummingbird species examined to date apparently arrest kidney GFR overnight (Hartman Bakken et al., 2004; Hartman Bakken and Sabat, 2006). A similar finding has been recorded for a nectar-feeding bat (Pallas's long-tongued bats, Glossophaga soricina) during the daytime rest period (Hartman Bakken et al., 2008).

Evaporative water loss (EWL) is a third possible route that could be used to eliminate large volumes of preformed water. In birds, modulation of EWL either through the skin or respiratory surfaces (through panting) has been noted in response to heat stress (Dawson, 1982; Dawson and Whittow, 2000; Skadhauge, 1981; reviewed by Williams et al., 2012) and in relation to hydration state (Arad et al., 1987; Maloney and Dawson, 1998; Williams, 1996). However, there are few accounts linking modulation of EWL with water loading (Hartman Bakken and Sabat, 2006). Birds that consume nectar should be capable of higher rates of EWL than those consuming predominantly solid foods. Furthermore, nectarivores consuming dilute nectar should have higher EWL rates than those drinking more concentrated nectars.

In this study, we examined water handling in two nectarivore species: whitebellied sunbirds [Cinnyris talatala (Smith 1836)] and New Holland honeyeaters [Phylidonyris novaehollandiae (Latham 1790)]. Based on previous work showing that Palestine sunbirds could modulate their fractional water absorption, we predicted that these two passerines would similarly be able to modulate intestinal water absorption in repose to increased preformed water load. We predicted that these nectarivores would also vary renal function in response to diet concentration: GFR would increase and renal water reabsorption would decrease with increasing water load, but when these birds were not feeding overnight, we predicted that GFR would slow or stop to reduce renal metabolic expenditure. Finally, we predicted that these birds would modulate evaporative water loss in response to increasing water load.

Eight whitebellied sunbirds were captured in Jan Cilliers Park, Pretoria, and eight New Holland honeyeaters were captured on the Murdoch University campus, Perth, using mist-nets. The birds were housed in individual cages (27×31×21 cm) in controlled environment rooms maintained at 21±1°C with an 11 h photoperiod from 07:00 to 18:00 h. During captivity, sunbirds were fed a maintenance diet consisting of 20% w/w sucrose and 2% Ensure®, a nutritional supplement (Abbott Laboratories, Johannesburg, South Africa); honeyeaters were fed 20% w/w sucrose with 15% Wombaroo® powder (Wombaroo Food Products, Adelaide, Australia). Birds received the maintenance diet in inverted, stoppered syringes. Bird body mass (M_b) at the start of the experiments was 8.07±0.45 g for sunbirds, 22.6±1.65 g for honeyeaters.

During experiments the birds were housed in individual experimental cages (42×54×50 cm) made of Perspex with a one-way mirror in the front. Birds were fed from inverted syringes fixed to the inside of the back wall of the cage.

The routine animal care procedures and experimental protocols used in this study were reviewed and approved by the University of Pretoria (Animal Use and Care Committee EC013-07) and Murdoch University (Animal Ethics Committee R1137/05). Licenses permitting the possession and use of radiolabelled substances were obtained from the Nuclear Energy Corporation of South Africa (reference number 7710245246084) and from the Radiological Council of Western Australia (license number LS 345/2006). The Gauteng Directorate of Nature Conservation granted permits to capture and house the sunbirds, and the Australian Department of Environment and Conservation approved our use of honeyeaters.

We varied food intake rate by feeding birds three diet sugar concentrations (0.25, 0.5 and 1 mol l⁻¹ sucrose solutions) in separate feeding experiments. The order of trials and order of treatment given were both randomly assigned.

Before each trial, birds had fed ad libitum from a syringe containing their allocated experimental diet for 15 h. We injected each bird (intramuscular, i.m.) with a combined dose of ¹⁴C-l-glucose and tritiated water (³H₂O). At 16:00 h, sunbirds were weighed and then injected in the pectoralis muscle with ~15 μl of solution containing 140 KBq ¹⁴C-l-glucose and 150 KBq of ³H₂O, while honeyeaters were injected with ~50 μl containing 330 KBq of ¹⁴C-l-glucose and 360 KBq of ³H₂O. The mass of solution administered by i.m. injection was measured by weighing the syringe before and after administration. Aliquots of the i.m. solutions were saved for radioactivity analysis; samples were transferred to a vial of known mass (±0.00001 g), which was then re-weighed to calculate sample mass.

We examined the elimination of these radiolabelled markers in excreta. Cloacal fluid (CF) samples were collected for 2 h commencing immediately from the time of i.m. administration (16:00 to 18:00 h; afternoon samples; PM) and then again the following day (07:00 to 09:00 h; morning samples; AM). CF samples were collected from wax paper rolled through the cage floor (to minimise disturbance) by pipette immediately after the bird excreted, with the exact time noted. Samples were transferred to a vial of known mass, which was then re-weighed to calculate sample mass.

A single ~15 μl blood sample was collected by micro-haematocrit capillary tube from the brachial vein 2 h after i.m. administration. Microcapillary tubes were sealed with clay tube sealing compound (Vitrex Medical, Herlev, Denmark) and centrifuged for 2–3 min at ~9000 g to separate plasma from blood cells. At the same time as blood sampling, a small sample of ureteral urine was collected by catheter. The plasma and ureteral urine were each transferred to a vial of known mass, which was then re-weighed to calculate sample mass.

Injection aliquot, CF, plasma and ureteral urine samples were each mixed with 3 ml of scintillation fluid (sunbirds: Ultima Gold XR, Packard Bioscience, Groningen, The Netherlands; honeyeaters: Ecolite+, MP Biomedicals Australasia, Seven Hills, New South Wales, Australia) and then counted in a scintillation spectrometer (sunbirds: Packard Tri-Carb Liquid Scintillation Spectrometer; honeyeaters: Beckman LS6500 Liquid Scintillation Counter, Beckman Coulter, Fullerton, CA) for disintegrations per minute (d.p.m.) for ³H and ¹⁴C.

Diet consumption was measured gravimetrically (±0.001 g; measured at the commencement and end of each experimental phase) and after correcting for leakage (cups of paraffin were placed under each feeder to collect any spilt food, which was taken into account in the calculations), these values were used to estimate sucrose (; g h⁻¹) and water (; g h⁻¹) intake rates. Intake rates were calculated as a function of the actual time spent feeding, as we noted that many individuals would not return to feeding immediately.

Model fits were then compared by F-tests (Motulsky and Ransnas, 1987), where the residual sum of squares and the numbers of parameters in each model are used to compute the F-ratio, which tests for significant differences in the goodness-of-fit of the two models to the same data. The largest F-values and smallest P-values of each species are reported in each case.

A second assumption of the pharmacokinetic method is that the birds are feeding at a steady rate. Not all birds commenced feeding immediately after they were returned to the cage after injection of the radioisotopes. We previously have shown that the pharmacokinetic calculations are extremely sensitive to this assumption of steady-state feeding, and any time that the animal is not feeding needs to be taken into account in the calculations, especially for intake rates (Napier et al., 2012). To do this, the intake rates were adjusted for actual time spent feeding; this was done by re-setting t=0 to the point when the birds started to defecate regularly (and were thus feeding regularly). In order to handle this data issue objectively, we adjusted the data for each individual trial separately. While the honeyeaters would generally return to feeding almost immediately (39 trials; 9 trials had to be adjusted by 18.3±8.8 min, range 10–31), the sunbirds would spend longer before returning to feed (returned to feed immediately for 25 trials, 23 trials had to be adjusted by 22.7±15.4 min, range 4–77). Please see supplementary data for graphs with or without the correction (supplementary material Figs S1 and S2).

A third assumption is in regard to data accuracy. Data editing is an important but also very unreliable aspect of handling pharmacokinetic data (Napier et al., 2012). The first excreta samples are likely to have a low concentration of ³H and ¹⁴C, because these samples may reflect CF produced before the i.m. administration of the radioisotope markers, or before the equilibrium from i.m. (rather than intravenous) administration. Calculations of S and K are both extremely sensitive to inclusion of these erroneously low values and they do need to be removed (Napier et al., 2012). This method is supported in the pharmacokinetics literature for intravenous injections; even with intravenous injections there is some small lag to complete equilibration (Pappenheimer, 1990). Initial samples where the isotope concentration was <75% of subsequent samples were therefore eliminated from calculations.

Two-way repeated-measures ANOVA (RM-ANOVA) was carried out to examine the effects of diet concentration and time (PM or AM) on water intake rate (Statistica, Statsoft, Tulsa, OK, USA). One-way RM-ANOVA was used to test the effects of time upon GFR. Where data were missing for an individual (one whitebellied sunbird), that animal was deleted from the repeated-measures analyses. These analyses were followed by Tukey's honestly significant difference test for differences among means. To compare slopes of linear relationships, we used StatistiXL (Nedlands, Western Australia, Australia). For all other data, we used a mixed-model linear analysis of effects comparing the dependent factor (each water handling parameter) against total water gain (independent factor), including bird ID (random factor; these analyses therefore took into account the repeated measures on each individual), time (fixed factor; AM or PM) and body mass (covariate) in the analysis.

Values are means ±1 s.d. throughout. Statistical significance was accepted at P<0.05.

For afternoon values, the relationships of ln[CF³_H] and ln[CF¹⁴_C] with time were well described by negative linear functions (Table 1; see the example for one honeyeater individual shown in Fig. 1), with significant values (P<0.05) for the coefficient of determination (r²) for honeyeaters (³H: r²=0.88±0.14; ¹⁴C: r²=0.87±0.06) and sunbirds (³H: r²=0.73±0.24; ¹⁴C: r²=0.89±0.08). The afternoon elimination rate of ³H₂O and [¹⁴C]-l-glucose in CF did not violate the assumptions of one-compartment, first-order kinetics for either species. In all 24 sunbird ³H trials (F<0.01, P>0.990), 18 out of 24 honeyeater ³H trials (F<1.76, P>0.185), 22 out of 24 sunbird ¹⁴C trials (F<3.16, P>0.062) and five out of 24 honeyeater ¹⁴C trials (F<0.47, P>0.635), a bi-exponential model did not fit elimination significantly better than a mono-exponential model.

For morning values, coefficients of determination averaged 0.90±0.15 and 0.60±0.24 for ³H and ¹⁴C in sunbirds, respectively, and 0.90±0.11 and 0.28±0.25 for ³H and ¹⁴C in honeyeaters, respectively. Therefore there were a number of trials where the relationships were not statistically significant and therefore comparison between mono- and bi-exponential models could not be performed. In sunbirds, for 22 of the 23 trials that could be tested, a bi-exponential model did not fit ³H elimination significantly better than a mono-exponential model (F<0.01, P>0.990). In honeyeaters, for 16 of the 19 trials that could be tested, a bi-exponential model did not fit elimination significantly better than a mono-exponential model (F<3.708, P>0.050). There were only three ¹⁴C trials for sunbirds and five ¹⁴C trials for honeyeaters where both the mono-exponential and bi-exponential relationships were statistically significant; therefore, statistical comparison between the different model fits was not robust. The parsimonious option was therefore to use a mono-exponential model fit for all data.

The estimate of TBW (calculated from ³H₂O dilution to estimate distribution space, S³_H) for sunbirds was 51±11% of M_b and for honeyeaters 45±13% of M_b. The distribution space of ¹⁴C-l-glucose (S¹⁴_C) in sunbirds was 11.25±7.57% of their M_b while that of honeyeaters was 17.19±1.22% of their M_b.

Both sunbirds and honeyeaters drank significantly more of the dilute than the concentrated diets, and consequently water intake rates were higher on the more dilute sucrose diet concentrations (RM-ANOVA diet: sunbirds, F_2,20=38.77, P<0.001; honeyeaters, F_2,21= 73.50, P<0.001). However, there was no significant difference in water intake rates between afternoon and morning (RM-ANOVA time: sunbirds, F_7,15= 0.243, P=0.967; honeyeaters, F_7,16=0.134, P=0.994).

Total body water flux () was positively correlated with total water gain in both sunbirds and honeyeaters (mixed-model linear analysis of effects: P<0.001) for both afternoon and morning data (equations for regression lines shown in Fig. 2A and Fig. 3A). There was no significant difference in between afternoon and morning in sunbirds, but honeyeaters showed a different relationship for afternoon and morning data (P=0.015). Comparing between the two species, not surprisingly the intercepts of the data against total water gain were significantly different (PM: P=0.001; AM: P=0.032) which would reflect the greater TBW and intake rates of the honeyeaters compared with the sunbirds. However, the slopes comparing and total water gain were not significantly different between the two species (P>0.05).

Fractional intestinal water absorption (f_A) in sunbirds (Fig. 2B) did not differ between afternoon and morning (P>0.05), and was significantly correlated with total water gain (r²=0.78, P=0.002); sunbirds absorbed all the water ingested on the lowest water gain diets, but only half (average of 50%) the water ingested on the highest water gain diets. New Holland honeyeaters (Fig. 3B) had different f_A responses for afternoon and morning (P=0.010): there was a significant correlation between f_A and total water gain for the afternoon (r²=0.78, P=0.004), but this relationship did not reach statistical significance for the morning data (r²=0.06, P=0.057). Therefore, f_A in honeyeaters feeding in the afternoon was as low as 0.70 on the highest water gain diets (i.e. these birds were absorbing only 70% of the water in their intestine; up to 30% of the ingested water would pass through the intestine without being absorbed).

PM rate of water excretion was significantly positively correlated with total water gain in sunbirds (P=0.002; Fig. 2C) and honeyeaters (P=0.017; Fig. 3C).

There was a significant effect of time of day on estimates of GFR in both sunbirds (RM-ANOVA sunbirds: F_1,7=124.32, P<0.001) and honeyeaters (F_1,7=63.77, P<0.001). For both bird species, GFR was significantly higher in the afternoon than in the morning, and overnight GFR′ was negligible (Fig. 4). For both species, GFR was not correlated with total water gain (P>0.05; Fig. 2D, Fig. 3D). Estimates of afternoon kidney fractional water reabsorption (f_R) were similarly insensitive to water loading in both sunbirds and honeyeaters (P>0.05; Fig. 2E, Fig. 3E).

The estimates of TEWL were extremely variable for both species, which may largely be due to the number of pharmacokinetic calculation steps involved in these estimates. The cumulative error was likely to influence the calculations, where even slight differences in estimates of the parameters involved had substantial effects upon calculated values. Many of the estimates were less than zero (Fig. 2F, Fig. 3F). Assuming these values were zero, estimates of TEWL for sunbirds (0.56±0.38 ml h⁻¹, range 0–1.55 ml h⁻¹) were substantial (i.e. 7% of M_b hourly). TEWL was significantly positively correlated with total water gain in sunbirds (P=0.024; Fig. 2F): TEWL increased with water loading. The honeyeater data had a high proportion of erroneous values (N=10 of 24 trials yielded TEWL estimates <0 ml h⁻¹) and were highly variable (0.63±0.78 ml h⁻¹). TEWL was estimated as 3% of M_b hourly (range 0–2.76 ml h⁻¹, calculated by substituting erroneous data for values with 0 ml h⁻¹). There was no correlation between TEWL and total water gain for honeyeaters (P=0.216; Fig. 3F), but these estimates cannot be considered reliable.

We found that sunbirds and honeyeaters handle their water loads similarly for the most part. Both species showed modulation of intestinal water absorption (f_A) but no modulation of GFR or renal water reabsorption (f_R) with varying water intake. Sunbirds were more sensitive to the disruption caused by i.m. administration and would often not return to feed immediately, but when they did feed, they fed at a fairly steady rate in both the afternoon and the morning, with similar water intake, water flux, intestinal absorption, turnover and excretion. Honeyeaters showed a greater range of water gains for morning data, and differences between afternoon and morning data for water flux, intestinal absorption and excretion. First we will discuss the findings of this study and then assumptions and limitations of the steady-state feeding pharmacokinetics method.

Body water turnover rate increased linearly with water intake in both sunbirds and honeyeaters. When birds were feeding on the most dilute diets (0.25 mol l⁻¹ is an ecologically relevant concentration for nectar solutions), sunbirds were turning over up to 80% of their TBW every hour, while honeyeaters were turning over up to 50% of their TBW. This is a dramatic water turnover rate, which is similar to water turnover rates experienced by aquatic vertebrates (Beuchat et al., 1990). How these birds deal with these massive amounts of preformed water is therefore an important aspect of their physiology.

Water loading puts an immense burden on the renal system. The two species of hummingbirds tested to date appear to deal with water loading by relying on their renal system, absorbing the majority of ingested water across the intestine and showing no regulation of intestinal water absorption on dilute diets (Hartman Bakken and Sabat, 2006; McWhorter and Martínez del Rio, 1999). By contrast, Palestine sunbirds regulate their f_A, avoiding 64% of ingested water by shunting this water straight through the intestine when intake rates are high (McWhorter et al., 2003), confirming the intestinal shunting hypothesis (Beuchat et al., 1990). Our study supports the findings for Palestine sunbirds, with whitebellied sunbirds also modulating intestinal water absorption, avoiding 50% of the ingested water when water intake rates are high and thereby reducing renal load. New Holland honeyeaters also modulated f_A, avoiding up to 30% of ingested water when water intake rates are high in the afternoon. However, in the morning, honeyeaters showed extremely variable responses and, therefore, their f_A was not significantly correlated with total water gain (P=0.057). This variability is likely due to individual responses to dehydration overnight when the birds are fasting, thus requiring different levels of rehydration in the mornings, but may also indicate problems with the assumptions of the pharmacokinetic method in this case (i.e. some honeyeaters may not be in a steady feeding state during the morning and may be rehydrating, given that they show lower water flux for corresponding total water gain values measured in the afternoon).

Interestingly, GFR did not vary with different levels of water loading for either sunbirds or honeyeaters. A similar lack of response of GFR to varying water gain was also recorded in S. sephanoides hummingbirds (Hartman Bakken and Sabat, 2006). While the hummingbirds had a GFR that was 10% lower in the morning compared with the afternoon (Hartman Bakken and Sabat, 2006), this difference between afternoon and morning GFR values was even more pronounced for sunbirds (74% lower) and honeyeaters (86% lower). The extremely low morning GFR values for honeyeaters are especially puzzling, and may be related to rehydration processes.

Neither sunbirds nor honeyeaters showed a relationship between water gain and f_R. This is unexpected, as hummingbirds (S. sephanoides) and nectar-feeding bats (G. soricina) decrease f_R with increasing water gain as their mechanism of countering water-loading (Hartman Bakken et al., 2008; Hartman Bakken and Sabat, 2006; McWhorter and Martínez del Rio, 1999). The lack of modulation of f_R in sunbirds and honeyeaters supports the suggestion that modulation of intestinal water absorption is likely to be the important physiological mechanism used by these passerines.

When feeding on dilute diets, nectarivores excrete greater volumes of urine (Goldstein and Bradshaw, 1998; Nicolson and Fleming, 2003b), but could potentially also adjust the volume of water that is lost by evaporation. Birds that consume nectar should be capable of higher rates of EWL than those consuming predominantly solid foods, and ideally should be able to modulate their TEWL according to their preformed water load. However, TEWL for S. sephanoides was not different than predicted from an allometric expectation and was not affected by water intake (Hartman Bakken and Sabat, 2006). We used the same prediction based on our data and allometric equations (Williams, 1996) and found that the TEWL allometric calculations for both sunbirds (2.11 ml day⁻¹ or 0.09 ml h⁻¹) and honeyeaters (3.34 ml day⁻¹ or 0.14 ml h⁻¹) were much lower than the values calculated in the present study (0.56±0.3 and 0.63±0.78 ml h⁻¹, respectively). In sunbirds, two studies have demonstrated a possible link between diet and EWL (Fleming et al., 2004b; Lotz and Nicolson, 1999). Similarly, for two honeyeater species, gravimetrically measured EWL was affected by diet concentration (Collins, 1981). Pallas's bats (G. soricina) have also been shown to increase TEWL with increasing water intake (Hartman Bakken et al., 2008). While these data suggest that nectar-feeding animals may respond to increased preformed water load by increasing TEWL, it is also important to consider what happens when these animals stop feeding. Estimations of TEWL in hummingbirds (S. sephanoides) predict that these birds would not have any problem replacing the amount of water lost through evaporation (~2% of body water per hour) while feeding, but that, unchecked, this would amount to a loss of ~28% of their TBW when they are not feeding overnight (Hartman Bakken and Sabat, 2006).

Unfortunately, using the pharmacokinetic technique to calculate TEWL has proven to be unreliable in this study for sunbirds and honeyeaters. The values needed for the many calculations all include some error in estimation, and minute variations in each component of the final equation may compound to result in large errors. We estimated values for honeyeater TEWL that were extremely variable and close to (or below) zero, making it difficult to draw any substantial conclusions. TEWL in sunbirds were similarly highly variable, but the TEWL estimates were significantly correlated with total water gain.

Although GFR did not change with varying levels of water loading, it is sensitive to water deprivation: both sunbirds and honeyeaters arrested kidney function at night. Shutting down the kidneys overnight appears to be an important mechanism used by hummingbirds (Hartman Bakken et al., 2004; Hartman Bakken and Sabat, 2006), as well as sunbirds and honeyeaters (present study), to help avoid potential dehydration during the overnight fast. Although we recorded no changes in GFR with water intake, what did change with varying water loads was intestinal water absorption, which was higher for the most concentrated diets and declined with diet dilution for both sunbirds and honeyeaters.

Certain assumptions are made in the steady-state feeding pharmacokinetic protocol used. While some assumptions are supported by previous studies, others have the potential to cause variations and inconsistencies (Napier et al., 2012).

The first assumption is that the estimates of K and S are derived from correct modelling of the numbers of distribution pools (i.e. the relationship between isotope concentration and time reflects dispersal through a single compartment, rather than more than one body compartment). In both species, single compartment, first-order kinetics could be applied to ³H₂O elimination for both afternoon and morning data. Elimination of [¹⁴C]-l-glucose in the afternoon was clearly single compartment; however, elimination of [¹⁴C]-l-glucose in the morning was less well described by a linear relationship. This may be due to the pattern of CF excretion after fasting overnight – both sunbirds and honeyeaters arrested kidney function at night, and the first excreta samples in the morning, which were smaller in volume and more concentrated than those produced later in the morning, were likely to represent CF that had been retained until the bird recommenced feeding in the morning (Fleming et al., 2004b). Consequently, the relationship with time was lost for these early samples (i.e. the time that the CF was produced was not the time recorded as excreted). This was not observed for ³H₂O excretion because water would continue to be reabsorbed and handled overnight through EWL and cloacal reabsorption.

The second assumption is that the animals are feeding at a steady rate. This assumption is valid for the afternoon data but is potentially violated in the morning due to the overnight fast and rapid rehydration and feeding (Fleming et al., 2004a); conclusions about morning data should be made with careful consideration of these potential errors. Additionally, response to the experimental method was also a cause for concern in regard to the assumption of steady-state feeding. Because the honeyeaters mostly resumed feeding within minutes, these birds did not confound the assumption of steady-state feeding. However, some whitebellied sunbirds did not commence steady-state feeding immediately after being captured and injected, and for half of the experimental trials with sunbirds, the time calculations had to be adjusted accordingly (compared with ~20% of trials with the honeyeaters). Other species differences in feeding and excretion behaviour were also identified. The first excreta after i.m. administration for the honeyeaters showed higher [¹⁴C]-l-glucose concentrations than subsequent values (Fig. 1), while the initial values for the sunbirds were lower than subsequent excreta. This difference suggests that sunbirds probably reduced GFR in response to disturbance, but the honeyeaters continued to eliminate [¹⁴C]-l-glucose through glomerular filtration and reduced frequency of excretion (i.e. stored CF and reabsorbed water in the distal intestine) until they started feeding normally. When honeyeaters started to feed, the concentration of ³H₂O in excreta dropped as urine flow rate increased. But the sunbirds are a different matter; if they retained water then effectively they were a closed system and the pharmacokinetic model would not apply. This is sufficient justification to adjust the intake data by re-setting t=0 to the point when the birds started to defecate regularly (and were thus feeding regularly).

The third assumption of the steady-state pharmacokinetic method is in regard to data accuracy, assuming that there is immediate distribution of the marker from the site of injection, that concentrations in the CF reflect those in the blood, and that isotope concentrations leaving the body are equal to those in body water at that moment in time. However, previous research has identified differences in isotope concentration between body water and excreted fluids, which occur due to physical and biological fractionation (Lifson and McClintock, 1966), a process that is believed to occur in nectar-feeding birds (McWhorter and Martínez del Rio, 1999). Thus, for better accuracy, we estimated the proportion of ingested water contributing to the turnover of TBW according to the model presented herein. This calculation makes the assumption that the rate of appearance of isotope in the excreted fluid is equal to the disappearance of isotope from TBW. As an aside, although the estimates of TBW (sunbirds: 51±11%; honeyeaters 45±13% of M_b) may appear to be lower than would be expected, these values are only marginally lower than values for green-backed firecrowns [56.6±2.0% (Hartman Bakken and Sabat, 2006)] or Palestine sunbirds [63.6±0.7% (McWhorter et al., 2003)].

In conclusion, this study shows that both sunbirds and honeyeaters use modulation of intestinal water absorption as an important component of water regulation to help deal with massive preformed water loads. Shutting down GFR during the natural overnight fast is another way of reducing the energy required by the kidneys and avoiding dehydration. Sunbirds and honeyeaters maintain osmotic balance very effectively on diets that can vary markedly in preformed water load by making use of a combination of mechanisms, varying water absorption and excretion through the intestine, kidneys and EWL.

Our thanks to Craig Symes for helping to capture birds, Angela Köhler for helping with animal care, and three reviewers for helping improve the manuscript.

 
• CF

  cloacal fluid

 
• f_A

  fractional water absorption in the gut

 
• f_T

  fractional turnover rate of body water

 
• f_R

  fractional water reabsorption in the kidneys

 
• GFR

  glomerular filtration rate (ml h⁻¹)

 
• GFR′

  estimated overnight GFR (ml h⁻¹)

 
• I¹⁴_Ct=0

  intercept concentration of ¹⁴C in plasma (d.p.m. ml⁻¹)

 
• ln[CF³_H]

  ln-transformed ³H₂O concentration in cloacal fluid

 
• ln[CF¹⁴_C]

  ln-transformed [¹⁴C]-l-glucose concentration in cloacal fluid

 
• K

  elimination rate constant

 
• K³_H

  fractional water turnover (h⁻¹)

 
• K¹⁴_C

  fractional l-glucose turnover (h⁻¹)

 
• M_b

  body mass (g)

 
• S

  distribution space

 
• S¹⁴_C

  [¹⁴C]-l-glucose distribution space (ml)

 
• S³_H

  water distribution space (ml)

 
• 

  sucrose intake rate (g h⁻¹)

 
• TBW

  total body water (ml)

 
• TEWL

  total evaporative water loss (ml h⁻¹)

 
• 

  water intake rate (ml h⁻¹)

 
• 

  water excretion rate (ml h⁻¹)

 
• 

  metabolic water production rate (ml h⁻¹)

 
• 

  water flux (ml h⁻¹)