Hummingbirds arrest their kidneys at night: diel variation in glomerular filtration rate in Selasphorus platycercus

Nectarivorous vertebrates face an osmoregulatory challenge. When feeding,they ingest astounding volumes of water(Martínez del Rio et al.,2001), yet they must prevent dehydration when they are not feeding(Powers, 1992). Therefore,achieving water balance requires the capacity to both eliminate and conserve water. Water conservation, however, requires different morphological characters and physiological processes from those necessary for water elimination (Dantzler, 1989; Goldstein and Skadhauge,2000). Hummingbirds, because of their small body sizes(Dunning, 1992) and high mass-specific metabolisms (Suarez,1992), are particularly challenged by this dilemma(Beuchat et al., 1990). How do hummingbirds meet these conflicting demands? In this article, we report the results of several experiments designed to shed light on the kidney's role in resolving this quandary.

As a consequence of ingesting food that is principally water(Baker, 1975), hummingbird water fluxes range from one to seven times their body mass(M_b) per day(Martínez del Rio et al.,2001). Because hummingbirds absorb essentially all ingested water that enters the gastrointestinal tract(McWhorter and Martínez del Rio,1999), the renal system must play a critical role in maintaining water balance. To avoid overhydration(Faenestil, 1977),hummingbirds must rapidly eliminate a large fraction of ingested water. How do hummingbird kidneys respond to these high water loads? Glomerular filtration rate (GFR) sets the pace of water reabsorption and/or elimination by the kidney. Although GFR appears to be less sensitive to water loading than to water deprivation (Williams et al.,1991), we hypothesized that hummingbirds would increase GFR to eliminate excess ingested water (McWhorter et al., 2004). A second complementary possibility is that renal fractional water reabsorption (FWR) would decrease as water load increases(Goldstein and Bradshaw,1998). Although the need to process large water loads may be, in part, ameliorated by high evaporative water loss (EWL) rates(Powers, 1992), these water losses can constitute a serious problem for hummingbirds when they are not feeding. Their inability to concentrate urine(Lotz and Martínez del Rio,2004) in combination with their high EWL rates suggests a potentially acute risk of dehydration for hummingbirds. Water conservation is therefore necessary when they are not feeding, for example at night and during extended periods of flight.

How do hummingbirds reduce urinary water losses during non-feeding periods?GFR decreases in response to water deprivation in several bird species(Yokota et al., 1985; Williams et al., 1991; Goldstein and Skadhauge,2000). Because hummingbirds do not feed at night, they are likely to be dehydrated in the early morning and need to conserve water(Fleming et al., 2004). We hypothesized that GFR would be lower during both the night and morning relative to the evening (Goldstein and Rothschild, 1993). We also predicted that hummingbirds would reduce GFR during an episode of water deprivation.

After mist-netting male broad-tailed hummingbirds (Selasphorus platycercus Swainson; M_b=3.60±0.40 g, N=10) in Albany County, Wyoming, USA (41°20′ N,106°15′ W), we housed them in individual cages(0.6×0.6×0.6 m) kept at 24±1°C on a 13 h:11 h photoperiod (photophase: 07:00–20:00 h MST). Hummingbirds fed ad libitum on two maintenance diets. Between 08:00 and 18:00 h, they fed on a 13.0% (mass percent) solution of Nektar-Plus (Guenter Enderle, Tarpon Springs, FL, USA) supplemented with vitamins (0.4%; Nekton-S; Guenter Enderle)and sucrose (5.0%). From 18:00 to 08:00 h, they fed on a 25% sucrose solution. Hummingbirds had to hover to feed and were acclimated to captivity for two weeks before experiments began.

We conducted two experiments. The first investigated diel variation in renal function in hummingbirds feeding naturally. The second experiment probed the effect of food (and thus water) deprivation on renal function. In experiment 1, we measured both renal FWR and GFR from roughly 18:00 to 19:59 h(`evening'). In the same experiment, we measured GFR from 20:00 to 06:59 h(`night') and from 07:00 to approximately 08:30 h (`morning'). Experiment 2 was conducted from approximately 11:00 to 15:00 h. In this experiment, we first measured GFR in hummingbirds feeding voluntarily (`midday') and then removed the sucrose solutions from their cages (`fast'). After this ∼1.5 h fast, we returned the sucrose solutions and continued measuring GFR in freely feeding hummingbirds.

During experiments, hummingbirds were housed individually in opaque Plexiglas® cages (0.3×0.3×0.3 m). One cage panel was a Mylar®-coated, one-way glass mirror. Each cage contained one perch that was fitted with an insulated Cu–Cn thermocouple (±0.1°C;ΩOmega Corporation, Stamford, CT, USA) and suspended from an electronic balance (±0.01 g; Scout II; Ohaus Corporation, Florham Park, NJ, USA). Hummingbirds were acclimated to these cages for 2 days before each trial.

Hummingbirds increase their food intake when the sugar concentration of their food decreases (Martínez del Rio et al., 2001). To vary ingested water loads, we fed hummingbirds 292 and 876 mmol l^–1 sucrose solutions. The fractional water contents of these solutions are 0.94 and 0.81, respectively. In this report, `food intake' is the volume of sucrose solution ingested;`water intake' is the ingested volume of preformed water; and `food/water'refers to the sucrose solutions. Hummingbirds fed ad libitum on these sucrose solutions for ∼4 h before a trial. We assigned trial order and sucrose concentration randomly for each hummingbird, and hovering was required to feed. All measurements were conducted at 24±1°C and the photoperiod held constant.

We estimated GFR using a single injection of[¹⁴C]l-glucose(Chang et al., 2004) and a modified version of the slope-intercept method(Hall et al., 1977; Florijn et al., 1994). Our sole modification was that the marker disappearance rate from plasma is matched by its rate of appearance in excreta. In addition to the assumption of constant GFR made by the slope-intercept method(Hall et al., 1977; Florijn et al., 1994), our modification assumes constant renal FWR. Therefore, our method of estimating GFR can only be applied with a single compartment model of marker clearance. This same modification was used by McWhorter et al.(2004). It allows the investigation of renal function in unanesthetized free-flying birds.

We injected each hummingbird in the pectoralis muscle with 9.25×10⁴ Bq of [1-¹⁴C]-l-glucose (Lot#345-058-050; Moravek Biochemicals, Brea, CA, USA) dissolved in 10 μl of deionized water. Injections were at ∼18:00 and ∼11:00 h for experiments 1 and 2, respectively. After injections, we collected excreta samples for >1 h. Following the initial excreta collection for experiment 1, we collected both a ureteral urine, using a close-ended polyethylene cannula (Goldstein and Braun,1989), and blood sample (∼10 μl). The blood sample was obtained by clipping a single toenail. We collected these samples between 19:40 and 19:59 h. We resumed collecting excreta the following morning. Fig. 1 illustrates our procedure for experiment 1. In experiment 2, we collected excreta samples before and after an ∼1.5 h food/water deprivation period. Excreta samples were collected, using glass capillary tubes, from the wax paper that lined the cage bottom. We counted d.p.m. (LS 6000IC; Beckman Coulter, Fullerton, CA,USA) after dissolving injectate aliquots, excreta, ureteral urine and plasma samples in 7.0 ml liquid scintillation cocktail (EcoLume; ICN Biomedicals,Costa Mesa, CA, USA). All analyses were corrected for ¹⁴C background, quench and chemiluminescence.

Hummingbirds can enter torpor (Calder and Calder, 1992). To find out if hummingbirds remained normothermic during our measurements, we obtained estimates of T_b using insulated Cu–Cn thermocouples affixed to each perch and digital thermometers (±0.1°C; HH506; ΩOmega Corporation). The length of the perches (20 mm) forced birds to sit atop the thermocouple so that it contacted the abdomen skin surface. We calibrated perching temperatures with cloacal temperatures. Our criterion for hypothermia was any T_b estimate lower than 39.0°C(Calder and Calder, 1992). During the 11 h night phase, we measured T_b every 0.5 h;for all other experiments, we monitored T_bcontinuously.

Because the relationships between food intake and sugar concentration for nectarivorous birds are well described by power functions(Martínez del Rio et al.,2001), we log_e-transformed food intake and sucrose concentration data. To determine the effect of food intake rate and subject on GFR, we used repeated-measures analysis of variance (RM-ANOVA). To test for differences among means, we used Tukey's Honest Significant Difference(Tukey's HSD). In all other cases, we used linear models on non-transformed data to assess significance. We report values as means ± 1 s.d.

After injection, the decline in ¹⁴C concentration (hereafter[¹⁴C]) of excreta with time followed single-compartment,first-order kinetics (Fig. 1). Mean coefficient of determination (r²) values for log_e-transformed data during experiment 1 were 0.83±0.12(N=10) and 0.43±0.23 (N=10) for the evening and morning, respectively (Fig. 1). During experiment 2, r² values were 0.75±0.15(N=10) and 0.49±0.28 (N=10) before food/water was removed and when it was returned, respectively.

Our estimate of S_P in broad-tailed hummingbirds was 0.74±0.15 ml (N=9), which is approximately 20.6±4.2% of M_b. [¹⁴C]l-glucose equilibration time was 19±11 min (N=20). The integrals of the relationship between [¹⁴C] of excreta with time indicated that we recovered 97.3±1.1% of Q_i (N=20). Because subject was a nonsignificant parameter in all our models (P>0.2), we removed this factor from all analyses.

During the evening and morning, food intake rate increased significantly as the sucrose concentration decreased (RM-ANOVA: F_1,7=10.83, P=0.0133, N=9). During the evening, food intake rates were 1.17±0.37 (N=5) and 0.56±0.14 ml h^–1(N=4) on the 292 and 876 mmol l^–1 solutions,respectively. Food intake rates during the morning were 1.11±0.39(N=5) and 0.65±0.20 ml h^–1 (N=4) on the 292 and 876 mmol l^–1 solutions, respectively. GFR during these same time periods was not influenced by sucrose concentration (RM-ANOVA: F_1,7=1.54, P=0.25, N=9). We therefore removed sucrose concentration from the analyses described in this section.

There were significant differences among our GFR estimates (RM-ANOVA: F_2,7=59.9, P<0.0001, N=9), with Tukey's HSD tests revealing that GFR_EVENING,GFR′_NIGHT and GFR_MORNING were all different from each other (Fig. 2). GFR_EVENING was 2.3±0.5 ml h^–1(N=9), ∼110% of the allometric prediction(GFR=0.013M_b^0.76; Bennett and Hughes, 2003; Fig. 2). There were no differences in [¹⁴C] of excreta between the last evening and first morning samples (paired t-test: t₈=0.52, P=0.62, N=9; Fig. 1) and GFR′_NIGHT was –0.1±0.2 ml h^–1 (N=9), suggesting an overnight interruption of whole-kidney GFR (Fig. 2). Our GFR′_NIGHT estimate was not different from 0 (t-test: t₈=–0.83, P>0.2, N=9). GFR_MORNING was 0.9±0.6 ml h^–1(N=9) and was lower than GFR_EVENING by a factor of 2.6(Fig. 2).

Contrary to our prediction, water intake rate did not influence GFR during the evening or morning (linear regression: evening, P=0.27, N=9; morning, P=0.34, N=9; Fig. 3A). However, during the evening, renal FWR decreased linearly as water intake rate increased(y=–0.13x+0.89, r²=0.66, P=0.03, N=7; Fig. 3B).

At midday, food intake rate increased significantly as sucrose concentration decreased (RM-ANOVA: F_1,7=30.44, P=0.0009, N=9). These intake rates were 0.9±0.3(N=5) and 0.4±0.2 ml h^–1 (N=4) on the 292 and 876 mmol l^–1 solutions, respectively. GFR,however, was not affected by sucrose concentration (RM-ANOVA: F_1,7=0.75, P=0.42, N=9). Following the∼1.5 h food/water deprivation period, sucrose concentration did not affect food intake rate (RM-ANOVA: F_1,7=0.94, P=0.36, N=9) or GFR (RM-ANOVA: F_1,7= 0.00, P=0.9930, N=9). We removed sucrose concentration from our analyses presented in this section.

Before food/water removal, GFR_MIDDAY was 1.8±0.4 ml h^–1; (N=9; Fig. 4). During the food/water deprivation period,GFR′_FAST (0.9±0.5 ml h^–1; N=9) was 50% lower than GFR_MIDDAY(Fig. 4). When we returned the food/water, GFR_RETURNED was 1.4±1.0 ml h^–1(N=9; Fig. 4). Our GFR estimates differed significantly (RM-ANOVA: F_2,7=9.79, P=0.0094, N=9), but Tukey's HSD tests showed that these differences were only between GFR_MIDDAY and GFR′_FAST; both GFR_MIDDAY and GFR′_FAST were not significantly different from GFR_RETURNED (Fig. 4). GFR_MIDDAY increased significantly as water intake rate increased (y=0.78x+1.36, r²=0.52, P=0.03, N=9; Fig. 5A). However, GFR_RETURNED was not influenced by water intake rate (linear regression: P=0.71, N=9; Fig. 5B).

Hummingbirds were normothermic throughout all experimental trials except at night, where they spent 10.4±5.3% of the 11 h dark phase hypothermic(N=10). The rate of change in M_b(ΔM_b) during the night was –0.04±0.01 g h^–1 (N=10) and decreased linearly as time spent hypothermic increased (y=–0.02x+0.06, r²=0.69, P=0.0028, N=10; Fig. 6). During the food/water deprivation period, ΔM_b was –0.25±0.11 g h^–1 (N=8) and was significantly higher than overnight ΔM_b (paired t-test: t₇=4.94, P=0.0017, N=8).

GFR in broad-tailed hummingbirds varied throughout the day and in response to food/water deprivation. Perhaps the most surprising result of this study is the seeming cessation of GFR by hummingbirds during the night. Here, we first consider the diurnal variation in GFR displayed by hummingbirds; then we discuss the renal responses to food/water deprivation, paying particular attention to the observation of an overnight interruption in whole-kidney GFR. We conclude by using a phylogenetic approach to determine whether diurnal GFR in nectarivorous birds conforms to the allometric expectation.

Broad-tailed hummingbirds displayed significant diurnal variation in GFR. They had a low GFR in the morning (0.9±0.6 ml h^–1; Fig. 2), an intermediate GFR at midday (1.8±0.4 ml h^–1; Fig. 4) and a high GFR in the evening (2.3±0.5 ml h^–1; Fig. 2). It is likely that hummingbirds filter slowly in the morning to conserve water and hydrate after a night of water losses (Fleming et al.,2004). Because intake rates during the day are sufficient for birds to hydrate within a few hours(Collins, 1981), the observation of a gradual increase in GFR throughout the day is perplexing but seems to be a pattern shared by other birds. Goldstein and Rothschild(1993) reported a similar pattern in song sparrows (Melospiza melodia).

When hummingbirds were deprived of food/water, they reduced mean GFR(Fig. 4). This finding is consistent with the responses to water deprivation observed in other birds(Williams et al., 1991; Goldstein and Skadhauge,2000). There is, however, one notable difference. In most of the other species examined, the reduction in GFR occurs progressively over a period of several days (Williams et al.,1991). Yet, hummingbirds modulated GFR within 1.5 h of deprivation(Fig. 4). This observation is not surprising, but it illustrates that GFR in hummingbirds is particularly sensitive and responsive to food/water deprivation. Although broad-tailed hummingbirds reduced mean GFR significantly during the deprivation period,they displayed a wide range of responses(Fig. 7). The reduction in mean GFR ranged from moderate (∼25%; Fig. 7B) to almost complete (∼90%; Fig. 7C). This variation may be explained by differences in water balance status among birds prior to food/water removal.

Although our observation is not the first evidence of intermittent renal filtration in birds (Braun and Dantzler,1972; Goldstein,1993), it represents the first account of what appears to be interrupted whole-kidney GFR in a normothermic bird. Our observation of arrested nighttime renal filtration in broad-tailed hummingbirds (Figs 1, 2) is noteworthy for two reasons. First, because hummingbirds were normothermic for ∼90% of the night, the cessation of renal filtration was not a result of reduced pressure in the renal arteries due to hypothermia(Glahn et al., 1993). We cannot, however, rule out a nocturnal dip in systemic blood pressure(Miyazaki et al., 2002). Second, a sudden decrease in whole-kidney GFR disrupts homeostatic processes and can have pathological consequences(Anderson and Schier, 2001). How do hummingbirds cope with arresting whole-kidney GFR? This is an intriguing question, but one that is presently open. The ability to interrupt GFR, however, is better understood.

In birds, the reduction in GFR is believed to result from vasoconstriction of the pre-glomerular arterial vessels that supply `loopless' nephrons(Dantzler, 1989). This vasoconstriction is mediated by arginine vasotocin(Braun, 1976; Giladi et al., 1997; Goecke and Goldstein, 1997). In hummingbirds, more than 99% of all nephrons are loopless(Casotti et al., 1998). Consequently, hummingbirds cannot concentrate urine(Lotz and Martínez del Rio,2004), but they can reduce urinary water losses by decreasing GFR. This mechanism has a potential drawback. In mammals, the cessation of filtration due to vasoconstriction of afferent arterioles can lead to damage of renal cells from ischemia (Hays,1992). How do hummingbirds nourish these cells when GFR is suspended?

Birds, like other vertebrates with intermittent glomerular filtration, have a renal portal system (Dantzler,1989; Smith et al.,2000). Dantzler(1989) hypothesized that this renal portal circulation may perfuse nonfiltering loopless nephrons in the absence of a post-glomerular blood supply. Additionally, other researchers have noted glomerular bypasses in the arterial vasculature of the avian kidney(Siller and Hindle, 1969; Kurihara and Yasuda, 1975). Although these features may allow the perfusion of renal cells when filtration is suspended, their relative importance is unknown.

One would expect high GFRs in animals with the astounding water intakes that characterize nectarivorous birds(Yokota et al., 1985; McWhorter et al., 2004). Accordingly, our estimate of GFR in broad-tailed hummingbirds exceeded the allometric prediction (Table 1). The other available data for nectarivorous birds, however,suggest that GFRs are lower than expected(Table 1; Bennett and Hughes, 2003; McWhorter et al., 2004). To find out if diurnal GFR is higher or lower than expected from M_b in nectarivorous birds(Calder and Braun, 1983), we used phylogenetically independent contrasts (PICs) and the method proposed by Garland and Adolph (1994). We used the DNA–DNA hybridization tapestry of Sibley and Ahlquist(1991) as a hypothesis for the phylogenetic relationships and evolutionary distances of birds(Fig. 8A). Briefly, we constructed a regression through the origin with all the standardized phylogenetic contrasts of log₁₀(M_b) and log₁₀(GFR). This regression excluded the nectarivorous species. We then determined whether the contrasts including nectarivorous birds were within or outside the 95% confidence interval for this relationship.

Before phylogenetic correction, the relationship between M_b (g) and GFR (ml h^–1) was described by a power function (y=–0.85x^0.74, r²=0.90, N=28; Fig. 8B). The exponent obtained using PICs (0.72±0.10; N=23; Fig. 8C) was similar to that obtained from the phylogenetically uncorrected regression (0.74±0.26; Fig. 8B). The points for the contrasts that included the clades of nectarivorous birds fell within the 95%confidence interval for the regression line. Despite the high water fluxes that characterize nectarivorous birds, these animals do not seem to have unusual rates of glomerular filtration. This conclusion, however, must be treated with caution. An overnight mean GFR of 0 may qualify broad-tailed hummingbirds as outliers. If labile GFRs(Goldstein and Rothschild,1993; present study) are common among birds, the time of GFR measurement cannot be ignored in comparative analyses.

Hummingbird illustrations are by Annie Hartman Bakken. We owe a debt to two anonymous reviewers, whose keen criticisms helped us to measurably improve an earlier draft of this manuscript. We thank Dr Graham Mitchell for his insightful comments and mammalian bias, which reminded us of the important observation that birds are not mammals. Our capture, care and experimental protocols were approved by the University of Wyoming Animal Use and Care Committee. Hummingbirds were collected under United States Fish & Wildlife and Wyoming Game & Fish permits issued to C.M.R. Support for this research was provided by grants from the National Science Foundation (NSF; IBN-0110416)and the National Institutes of Health/National Center for Research Resources(RR-16474) awarded to C.M.R. and B.H.B., respectively; T.J.M. was supported by an NSF grant to William H. Karasov (IBN-0216709). B.H.B. extends his thanks to Dr Lloyd A. and Patricia L. Bakken.