Aquatic amphibian eggs frequently encounter hypoxic conditions that have the potential to limit oxygen uptake and thereby slow embryonic development and hatching. Oxygen limitation might be avoided if egg capsule surface area and oxygen conductance increased in response to hypoxia. We investigated this possibility in two salamander species, Ambystoma annulatum and Ambystoma talpoideum. The effective surface area of egg capsules increased in response to hypoxia, which increased the conductance for oxygen and enhanced oxygen transport. The ability of amphibian eggs to adjust their conductance in response to oxygen availability may increase survival in hypoxic environments.
Aquatic amphibian eggs frequently experience hypoxic conditions (Barth, 1946; Gregg, 1962; Moore, 1940; Savage, 1935), which can result in slowed development and decreased survival (Mills and Barnhart, 1999). Oxygen partial pressures below 1.5 kPa have been reported within egg masses of several species (Bachmann et al., 1986; Pinder and Friet, 1994; Seymour et al., 1995; Seymour and Roberts, 1991). There are two major causes of hypoxia in amphibian eggs. First, many species utilize eutrophic wetlands and ponds as breeding sites (Collins and Wilbur, 1979; Stebbins and Cohen, 1995). Such habitats are likely to have a high biological oxygen demand (BOD) and low oxygen pressures, particularly at night. Second, respiratory gases must pass through substantial diffusion barriers to reach or leave the developing embryo. These barriers generally consist of a gelatinous egg capsule, the vitelline membrane and the capsular and/or perivitelline fluid (Salthe, 1963). However, in ambystomatids and most other salamanders, the embryo exits the vitelline membrane early in development and completes development in the capsular chamber (Salthe, 1963).
Convection created by cilia on the surface of the embryo enhances oxygen transport in the capsular fluid (Burggren, 1985). Therefore, it can be assumed that most resistance to oxygen transport is in the egg capsule (Seymour, 1995). Oxygen must diffuse through the egg capsule because the gelatinous nature of the egg capsule prevents convection. The diffusive conductance of oxygen across the egg capsule can be described by the equation:
Oxygen conductance (GO2; cm3 min–1 kPa–1) is dependent on (i) Krogh’s coefficient of oxygen diffusion (KO2; cm2 min–1 kPa–1), which describes the permeability of the egg capsule to oxygen, (ii) egg capsule thickness (L; cm) and (iii) effective surface area (ESA; cm2), which is defined as the area of a sphere having the geometric mean of the internal and external surfaces of the egg capsule (Seymour, 1994; Seymour and Bradford, 1995).
The GO2 of amphibian eggs increases during embryonic development (Seymour and Bradford, 1987; Seymour et al., 1991). Absorption of water into the capsular chamber increases the volume of the egg, causing an increase in ESA of the egg capsule and a decrease in capsule thickness because of distention. The increased ESA and decreased thickness of the egg capsule increase GO2 (Seymour and Bradford, 1987; Seymour et al., 1991). In Pseudophryne bibroni, the increase in GO2 during development matched the increasing rate of embryonic oxygen consumption, so that oxygen partial pressure (PO2) inside the egg changed little with development (Seymour and Bradford, 1987). GO2 depended on developmental stage and was not altered by environmental factors such as temperature, PO2 or water potential (Seymour et al., 1991). In Rana pipiens, however, egg volume increased as temperature increased (Salthe, 1965). This finding raises the possibility that the capsule chamber could expand and that GO2 could thereby increase in response to environmental hypoxia. Such a response would enable eggs to adapt to local availability of oxygen as well as to embryonic oxygen demand.
Ambystomatid salamanders are pond breeders whose eggs can be expected to experience hypoxic conditions. We exposed the eggs of Ambystoma annulatum and Ambystoma talpoideum to several levels of external PO2 and measured changes in capsule ESA and thickness. Our working hypothesis was that the capsule chamber would expand, increasing ESA and decreasing thickness, thus leading to increased GO2 and increased oxygen availability for the developing embryo.
Materials and methods
Experiment 1: Ambystoma annulatum
Six Ambystoma annulatum Cope egg masses were collected from a small, semi-permanent pond near the town of Reeds Spring in Stone County, Missouri, USA. Each A. annulatum egg mass consisted of 20–50 eggs enclosed within a common outer jelly matrix. The jelly matrix was cut into pieces containing single eggs so that each egg would be similarly exposed to experimental PO2. A portion of the outer jelly matrix was left around each egg to avoid damaging the egg capsule. The thickness of the jelly matrix around each egg ranged from 0.3 to 0.4 cm.
Tissue culture plates (B-D Falcon) were cut in half to make six-compartment trays. A single egg was randomly assigned to each of the six 21 mm×17 mm compartments in each tray. The compartments were left open to permit water to circulate, and the eggs were restrained only by thin strips of plastic to prevent them from drifting out of the compartments. Each tray of six eggs was placed in one of six pools through which water circulated. PO2 levels in the pools were 2.1, 3.7, 6.4, 8.7, 12.7 and 17.4 kPa (details below).
The developmental stage (Harrison, 1969) and internal and external capsule diameters of each egg were recorded on days 0, 3, 7, 9 and 12 of the experiment. Diameters were measured using a Zeiss stereomicroscope and ocular micrometer. Measurements were taken while the eggs were submerged in water to prevent light refraction in the jelly and to avoid compression due to gravity. Egg capsule thickness was calculated from the internal and external capsule diameters by converting the diameters to radii and subtracting the internal radius from the external radius. Effective surface area was calculated from egg capsule internal and external diameters by converting the diameters to radii and using the equation:
where ri and ro are the internal and external radius, respectively. There were no significant differences in initial capsule ESA and thickness among the six treatments based on an analysis of variance (ANOVA; ESA, F5,26=0.62, P=0.6868; thickness, F5,26=0.64, P=0.6677). All embryos were at Harrison stages 31–35 (Harrison, 1969) when this experiment began. Embryos hatched at Harrison stages 39–40, which they reached 9–12 days after they had been placed in the PO2 treatments.
Control of oxygen levels
Water was deoxygenated using a gas-stripping column (Barnhart, 1995) and then reoxygenated by passing over a series of partitions and pools (aeration ladder). Water was continuously recycled through the system at a flow rate of approximately 0.5 l min–1. Homogeneity of PO2 within each pool was measured prior to experiments using a Cameron oxygen meter (model OM-201) with a semi-micro oxygen electrode (Microelectrodes, Inc., model MI-730). There was no measurable spatial variation in PO2 within each pool. Oxygen levels in each pool were checked every 2–3 days throughout the experiments using an Orion (model 820) oxygen meter. The Cameron oxygen meter was calibrated by bubbling water with air and nitrogen until the electrode reached equilibrium. The Orion oxygen meter was calibrated by bubbling water with air until the electrode reached equilibrium. All oxygen measurements were converted from percentage of air saturation to kPa on the basis of the daily mean barometric pressure, temperature and water vapor pressure (Dejours, 1981). Water temperature was controlled by thermostat and ranged from 14.6 to 14.9°C during the experiment.
Experiment 2: Ambystoma talpoideum
Ambystoma talpoideum (Holbrook) eggs were collected from Rainbow Bay, a temporary pond located at the US Department of Energy’s Savannah River Site, Barnwell County, South Carolina, USA. The A. talpoideum eggs were not enclosed within a common outer jelly matrix, as is common in most other Ambystoma species. Therefore, the eggs were easily separated from each other at their point of contact with surrounding eggs. Each egg was submerged in approximately 8 mm of water in a shallow plastic cup. Twelve cups were placed in each of six plastic containers in which PO2 was controlled (details below). At the beginning of the experiment, all embryos were staged and photographed using a Nikon Optiphot-2 and a Videoscope black-and-white CCD camera (Videoscope International). Measurements of egg capsule external diameter and thickness were taken from the photographs using MetaMorph image-analysis software (Universal Imaging). Measurements taken with the Metamorph image-analysis software were calibrated using photographs of a micrometer taken at the time the eggs were photographed. Egg internal radius was calculated by converting the external diameter to a radius and subtracting the capsule thickness. ESA was calculated from the egg capsule internal and external radii using equation 2. There were no initial differences in ESA between treatments (ANOVA, F1,50=2.36, P=0.1311) or among containers within treatments (ANOVA, F3,50=0.56, P=0.6453) and no initial differences in capsule thickness between treatments (ANOVA, F1,50=2.14, P=0.1497) or among containers within treatments (ANOVA, F3,50=1.61, P=0.1999). All eggs were at Harrison stages 12–15 at the beginning of the experiment. Each egg was staged daily until it reached Harrison stage 37–38, at which time the eggs were again measured. Hatching generally occurred within 1–2 days of when the final measurements were taken (i.e. 5–7 days after the experiment began).
Control of oxygen levels
Six plastic containers (30 cm×20.5 cm×8.5 cm) with tightly fitting lids were equipped with a 3 mm gas inlet coupling at one end of each container and a 1.5 mm gas outlet at the opposite end. Three of the containers were connected to a tank of compressed medical-grade gas consisting of 5 % oxygen and 95 % nitrogen, while the other three containers were connected to a tank of compressed medical-grade gas consisting of 21 % oxygen and 79 % nitrogen. The two sets of containers were interspersed on the shelf to prevent any effect of shelf position. Gas flow through each container was continuous for the duration of the experiment at a flow rate of approximately 100 ml min–1. Twelve egg-cups and a shallow dish of water were placed in each container. PO2 in each container was checked daily by opening the container and measuring PO2 in the dish using a YSI (model 58) oxygen meter. The oxygen meter was calibrated by bubbling water with air until the electrode reached equilibrium. All oxygen measurements were converted from percentage of air saturation to kPa on the basis of the daily mean barometric pressure, temperature and water vapor pressure (Dejours, 1981). The containers were kept at room temperature (23±1°C).
The effects of PO2 on capsule ESA and thickness were analyzed using ANOVA (SAS 8.0, SAS Institute, 1999). All tests were conducted using α=0.05. In experiment 1, differences in capsule ESA and thickness were analyzed at a common stage of development (stage 37–38) and also after a common time of exposure (12 days). In experiment 2, differences in capsule ESA and thickness between treatments and among containers within treatments were analyzed at a common stage of development (stage 37–38). Initial egg dimensions and initial developmental stage were used as covariates in all analyses to remove any variability in the data resulting from initial differences among eggs. Residuals from all analyses were inspected for normality and homogeneity of variance. In experiment 1, egg capsule thickness data were inverse-transformed to meet assumptions of normality and homogeneity of variance.
GO2 was calculated from capsule ESA and thickness using equation 1 and assuming that KO2 was a constant. KO2 values of 2.68×10–7 and 2.91×10–7 cm2 min–1 kPa–1 were used for experiments 1 and 2, respectively (Seymour, 1994). Differences in GO2 among treatments were analyzed using ANOVA. Initial GO2 and developmental stage were included in the model as covariates.
Experiment 1: Ambystoma annulatum
Results on day 12
The ESA increased in all treatments over the 12 days of the experiment. The ESA of eggs in low-PO2 treatments increased significantly more than the ESA of eggs in high-PO2 treatments (Table 1; Fig. 1B; F5,20=3.25, P=0.0261). The ESA of eggs in the lowest-PO2 treatment (2.1 kPa) increased by a factor of 3.5, whereas the ESA of eggs in the highest-PO2 treatment (17.4 kPa) increased by a factor of 2.5 (Table 1). Initial ESA accounted for a significant amount of variability in ESA at day 12 (F1,20=67.26, P<0.0001), but initial developmental stage did not account for a significant amount of variability (F5,20=1.73, P=0.1734).
Egg capsule thickness decreased in all treatments by factors ranging from 0.59 to 0.75 (Table 1; Fig. 1A). However, there were no significant differences among treatments on day 12 of the experiment (F5,20=2.01, P=0.1204). Among the covariates, initial thickness accounted for a significant amount of variability in capsule thickness at day 12 (F1,20=8.67, P=0.0080), but initial developmental stage did not account for a significant amount of variability (F5,20=0.24, P=0.9401).
On the basis of observed egg capsule ESA and thickness and assuming that KO2 was constant, GO2 did not differ significantly among treatments on day 12 of the experiment (Table 1; Fig. 1C; F2,20=2.47, P=0.0673), although there did appear to be an inverse relationship between GO2 and PO2 (Fig. 1C). GO2 across the egg capsule increased by a factor of 3.6 at 17.4 kPa but increased by a factor of 5.3 at 2.1 kPa.
Results at stage 37–38
Eggs in the lowest-PO2 treatment developed more slowly than eggs in the other PO2 treatments. Embryos in the 2.1 kPa treatment took 12 days to reach developmental stages 37–38, whereas all eggs in the other treatments had reached developmental stages 37–38 by day 7 of the experiment (Table 2) and were at developmental stages 39–40 by day 12 of the experiment (Table 1). Developmental stage is known to influence GO2 (Seymour et al., 1991). Therefore, capsule ESA, thickness and GO2 were also analyzed at a common stage of development (Harrison stage 37–38).
ESA differed significantly among treatments at developmental stage 37–38 (F5,24=7.09, P=0.0003) and the differences were even more pronounced than when ESA was compared after a common period of exposure to the treatments (Table 2; Fig. 1B). Whereas the ESA of eggs at 2.1 kPa had increased by a factor of 3.5, that of eggs at 17.4 kPa had increased by a factor of 1.9 (Table 2). Initial ESA accounted for a significant amount of variability in ESA at developmental stage 37–38 (F1,24=27.01, P⩽0.0001), but initial developmental stage did not account for a significant amount of variability in ESA (F5,24=1.38, P=0.2686).
Capsule thickness did not vary significantly among treatments when comparisons were made among eggs at a common stage of development (Fig. 1A; F5,24=1.15, P=0.3634). Change in thickness ranged from a factor of 0.83 to a factor of 0.62 (Table 2). Neither covariate accounted for a significant amount of variability in thickness at developmental stage 37–38 (initial thickness, F1,24=0.54, P=0.4683; initial developmental stage, F5,24=0.75, P=0.3957).
GO2 differed significantly among treatments at developmental stage 37–38 (F5,24=5.68, P=0.0014) and generally increased as PO2 decreased (Table 2; Fig. 1C). GO2 across the egg capsule increased by a factor of 2.7 at 17.4 kPa but increased by a factor of 5.3 at 2.1 kPa. Initial GO2 accounted for a significant amount of variability in GO2 at stage 37–38 (F1,24=6.97, P=0.0143), but initial developmental stage did not account for a significant amount of variability in GO2 (F5,24=0.44, P=0.8173).
Experiment 2: Ambystoma talpoideum
The eggs in one container assigned to the low-PO2 treatment were eliminated from all analyses because of problems in controlling gas flow, which caused large fluctuations in PO2 throughout the experiment.
The ESA of eggs in both the low- (6.3 kPa) and high- (18.8 kPa) PO2 treatment increased during development. However, the ESA of egg capsules in the low-oxygen treatment increased more than the ESA of egg capsules in the high-oxygen treatment (F1,43=28.79, P<0.0001). ESA increased by factors of 1.9 and 1.5 in the low- and high-oxygen treatments, respectively (Table 3). Significant differences also occurred in ESA among containers after accounting for treatment differences (F3,43=3.61, P=0.0207). Initial ESA and developmental stage both explained significant amounts of variation in ESA at the conclusion of the experiment (initial ESA, F1,43=8.23, P<0.0064; initial developmental stage, F6,43=2.78, P<0.0224).
Egg capsule thickness changed relatively little during development (Table 3), and there were no significant differences in thickness between treatments (F1,43=0.49, P=0.4857) or among containers after accounting for treatment differences (F3,43=1.51, P=0.2242). Initial egg capsule thickness and initial developmental stage both explained a significant amount of variation in capsule thickness at the conclusion of the experiment (initial thickness, F1,43=70.30, P<0.0001; initial developmental stage, F6,43=12.17, P<0.0001).
GO2 increased by a factor of 1.8 at 6.3 kPa and by a factor of 1.4 at 18.8 kPa (Table 3). The difference between treatments was significant (F1,43=12.70, P=0.0009). Initial GO2 accounted for a significant amount of variability in GO2 at developmental stage 37–38 (F1,43=10.65, P=0.0022). Initial developmental stage did not account for a significant amount of variability in GO2 at developmental stage 37–38 (F6,43=2.22, P=0.0596).
The design of both experiments can be criticized. In experiment 1, the treatment effects are potentially confounded with other effects of position along the aeration ladder, although we tried to ensure that no other variables (i.e. temperature, pH or light) varied with position along the aeration ladder. In experiment 2, independent containers were used so that the containers could be interspersed on the shelf. However, all containers in a given treatment received a gas mixture from the same compressed-air tank. Although neither experimental design was ideal, we feel that together the two experiments provide convincing evidence that PO2 influenced egg capsule GO2.
GO2 was significantly higher in lower-PO2 treatments in both A. annulatum and A. talpoideum when measured at a common developmental stage. The differences in GO2 among treatments were due primarily to differences in ESA, which consistently showed an inverse relationship with PO2. Egg capsule thickness did not differ consistently among treatments.
GO2 did not differ significantly among treatments after a standard treatment period (12 days) in A. annulatum. This result was probably due to the fact that embryos exposed to low PO2 developed more slowly than embryos exposed to high PO2. GO2 is known to increase in parallel with development (Seymour, 1995). Thus, it is more appropriate to compare GO2 at a common stage of development than after a particular time of exposure.
We predicted that swelling of the capsular chamber of eggs in the hypoxic treatments would stretch, and therefore thin, the egg capsules. In A. annulatum, the thickness of the egg capsule decreased measurably over the course of the experiment. However, we did not detect differences in thickness between treatments, possibly because the precision of our measurements was inadequate. In any event, the significance of capsule thickness in A. annulatum is problematic, because the eggs are normally enclosed in a jelly matrix and can be up to 1 cm from the surface of the egg mass. Therefore, egg capsule thickness may be only a small part of the total diffusion distance between the embryo and the outside environment. In this situation, it is useful to think of the jelly matrix as the environment of the individual eggs and the internal capsule surface area as the respiratory exchange surface of the embryo. Swelling of the capsule chamber increases the internal surface area, thus increasing GO2 (Table 1, Table 2). Analyses indicate that GO2 is very sensitive to changes in internal surface area but relatively insensitive to changes in capsule thickness unless the capsule is very thin (Seymour, 1994).
Although A. talpoideum eggs are generally also enclosed in a common outer jelly matrix, some populations lay individual eggs without a thick outer jelly matrix (Semlitsch and Walls, 1990). The outer jelly matrix is reduced to a thin, somewhat sticky, outer layer that can be considered part of the egg capsule. We intentionally chose to use eggs from these populations to fit more closely the model presented in equation 1. Thus, any changes in egg capsule thickness would have a significant effect on egg capsule GO2 and should be straightforward to interpret. However, the effects of hypoxia on egg capsule thickness in the A. talpoideum experiment were also difficult to interpret. Egg capsule thickness did not change substantially, and possibly even increased slightly, during development irrespective of PO2. Microscopic examination revealed that this was probably due to decreased cohesiveness of the outermost capsule layer, which increased interstitial space both within the outermost layer and between it and the underlying layers of the egg capsule. The loss of integrity in the outermost layer may have implications for egg capsule GO2, and needs to be investigated further. Not only does it allow for the possibility of convective transport of oxygen through the outermost layer of the egg capsule but it may also indicate changes in the egg capsule that would affect KO2.
We are aware of only a few previous studies that have explicitly investigated the effects of hypoxia on GO2 in either individual egg capsules or egg masses in amphibians; e.g. Seymour et al. (Seymour et al., 1991) and Seymour and Roberts (Seymour and Roberts, 1991). In contrast to the results from the present study, the GO2 of the egg capsule of Pseudophryne bibroni showed no response to hypoxia (Seymour et al., 1991). The reasons for this difference are in need of further investigation. However, it seems likely that the ability to adapt to changes in external PO2 in single, terrestrial eggs, such as those of P. bibroni, would have little selective advantage because the eggs are incubated in air, where PO2 is consistently high. In contrast, the eggs of most Ambystoma species are quite likely to be exposed to hypoxic conditions and there would be a distinct selective advantage in being able to increase GO2.
The ability to increase GO2 in response to hypoxic conditions has important implications for fitness. Hypoxia generally slows development and induces hatching at a premature stage of development (Mills and Barnhart, 1999). Larvae that are less developed and smaller at hatching show decreased survival (Mills and Barnhart, 1999), a decreased ability to compete with conspecifics (Smith, 1990) and a decreased ability to escape from predators (Petranka et al., 1987; Sih and Moore, 1993). Furthermore, many amphibian larvae inhabit ephemeral aquatic habitats in which they must complete larval development and metamorphosis before pond drying occurs (Semlitsch, 1987; Semlitsch et al., 1996). Therefore, the ability of the embryo to adjust GO2 in response to oxygen stress, thus facilitating growth and development, has the potential greatly to enhance fitness.
This study demonstrates that GO2 increases substantially in response to hypoxia, primarily as a result of changes in ESA. The ability to alter GO2 could have a strong influence on the survival of the resulting larvae and, thus, may be of general significance. This study raises a number of questions that should be examined further. The physiological mechanisms for increasing capsule GO2 and the response of capsule GO2 to intermittent hypoxia are both in need of further examination. Furthermore, the ability to respond adaptively to hypoxia through increased GO2 should be investigated in species with different habitat and life-history requirements and in other taxonomic groups to gain an understanding of the universality of this response.
Funding was provided in part by a Southwest Missouri State University faculty grant received by M.C.B. and an EPA grant (827095-01) received by R.D.S. T. Ryan provided A. talpoideum eggs from the Savannah River Site, and K. Yamada and E. Norton provided valuable assistance collecting data. This manuscript was improved considerably by the helpful comments of T. Green, J. Mills, B. Rothermel and two anonymous referees.