Considering the extent to which anadromous fish in the Columbia and Snake River systems in the United States were affected by GBT, it is not surprising that most of the research reported in the literature focuses on GBT in fresh water environments and, in particular, on anadromous salmon and trout species. Although research has been conducted on other fish species and aquatic invertebrates, it has been a small portion of the total effort. The work on fish has taken two distinct directions. There have been numerous experimental programs aimed at describing the signs of GBT and defining levels of DGS which are harmful to fish in terms of time to mortality. Other work has involved theoretical and experimental studies to define the physiological and biological phenomena causing GBT, accompanied by a variety of statistical studies of the available experimental data. These works are examined in the following sections, with emphasis on that information which is relevant to the derivation of water quality guidelines for DGS.

6.1 Pacific Salmon and Cutthroat Trout

For the most part, the studies of DGS and GBT which were conducted on fish in the Columbia and Snake Rivers relates time to mortality to variables such as fish age class (or size), species, water dissolved gas tensions, temperature, and depth. As mentioned earlier, some of these data were obtained with only dissolved nitrogen levels being reported and are of limited use. Another feature of much of the data is that control experiments were not always conducted in conjunction with experiments involving DGS. In some cases, even though control animals were used, the survival rate of control animals was not reported in the experimental results.

The volume of data (over 1000 records) is such that it would be impractical, in this document, to describe each experimental condition and associated results. Instead, the data (which are summarized in Table C3 of Appendix C) will be reviewed and analyzed in the context adopted by Jensen et al. (1985 and 1986) and by Fidler (1988).

In an effort to develop statistical models which describe time to mortality from GBT as a function of dissolved gas tensions and ancillary parameters, Jensen et al. (1985 and 1986) collected, summarized, and tabulated over 621 data records reported by 24 separate authors. The data were for three Pacific salmon species (chinook, coho, and sockeye) and two trout species (steelhead, and cutthroat). These data were subsequently reviewed by Fidler (1988) in work directed at defining dissolved gas thresholds for the major signs of GBT. As a result, additional data records were added to the tables covering conditions which were relevant to the threshold studies of Fidler (1988), but not to the statistical time to mortality studies of Jensen et al. (1986). For example, data describing exposures to DGS without mortality were as important in identifying thresholds as exposures involving mortalities. These data types were not included in the statistical analyses of Jensen et al. (1986). Table C3 of Appendix C includes the original data of Jensen et al. (1985) along with the additions by Fidler (1988). Also included in Table C3 is information on time to mortality as a function of TGP and other ancillary factors from more recent experimental studies, including information on fish species not considered in the work of Jensen et al. (1985 and 1986) or that of Fidler (1988). In total, Table C3 contains 1059 data records.

6.1.1 GBT Statistical Data Analyses

Jensen et al. (1986) applied a multivarient dose-response analysis to their data summaries. Of the data collected by Jensen et al. (1985), many experimental results were unusable because fundamental parameters, such as delta P, TGP, or fish size, were not reported in the original papers. Figure 8, based on 171 usable data records, shows Model 1 of Jensen et al. (1986) where time to 50% mortality (ET₅₀) is plotted versus TGP%. Also shown are the 95% confidence limits for the model. The model indicates that if the lower 95% confidence limit is used, TGP% levels below 105% are required to completely protect fish from mortalities associated with GBT. One of the qualifying conditions for this model is that, of the 171 data records used in the model, all but one data record involved water depths less than 0.6 m. The other single data record was for a water depth of one metre.

When Jensen et al. (1986) included ancillary factors such as water depth, fish size, water temperature, and percent O₂, in their statistical models, several important features of the multivarient analysis appeared. For example, their Model 3 suggested that ET₅₀ values were independent of water temperature for temperatures above about 9 degrees Celcius. Below that temperature, ET₅₀ increased markedly as temperature decreased. Their Model 4 indicated that ET₅₀ values were independent of fish length for fish over about 65 mm in length. Model 2 indicated that ET₅₀ values increased as water depth increased, while Model 8 indicated that ET₅₀ increased as the ratio of dissolved oxygen to nitrogen increased.

6.1.2 Graphical Analysis of Data

Fidler (1988) examined the data of Table C3, Appendix C by filtering the records based on variables such as fish size, species, pO₂ levels, etc. and summarizing the filtered results graphically. The objective was to determine if the data provided information on thresholds in water delta P which could be related to specific signs of GBT. This type of analysis was essential to the validation of threshold equations derived from biophysical studies of the causes of GBT (Section 6.1.3).

6.1.2.1 Lowest Apparent Threshold

The obvious means of detecting thresholds in water delta P is to graph time to mortality values as a function of water delta P. Figure 9 is a plot of these parameters using the data of Table C3. As noted in the figure and in subsequent figures, experiments in which the indicated level of mortality was not achieved over the exposure period are plotted as negative times to mortality. However, the values of these negative times to mortality represent the time of exposure divided by 10.0. The division by 10.0 was done to provide more detail in the mortality data. Clearly evident in the figure is the absence of mortalities at a water delta P less than 76 mm Hg. This would suggest that the lowest threshold for any lethal sign of DGS is 76 mm Hg. To a large extent, this is the basis for the US EPA guideline for DGS (US EPA 1986).

Before proceeding further, it should be noted that the data of Wright and McLean (1985) for chinook salmon have not been included in Table C3 because delta P levels varied somewhat over the experimental period (122 days) and there was also a significant difference in fish size over the experimental period. However, the results of Wright and McLean (1985) along with those of Cornacchia and Colt (1984) will be considered further in Section 6.1.4.

Figure 8: Model 1 of Jensen et al. (1986)

Figure 9: Time to Mortality Versus Water Delta P

6.1.2.2 Effect of Fish Size on Time to Mortality

Although the delta P threshold (76 mm Hg) indicated in Figure 9 is an important benchmark in deriving DGS water quality guidelines, a further refinement of the data analysis provides additional insight into delta P thresholds. As described earlier, Model 4 of Jensen et al. (1986) predicted that fish less than about 65 mm in length were not as susceptible to GBT as larger fish. Figure 10 is the result of filtering the data of Table C3 based on recorded values of delta P (i.e., not on N₂% only) , time to mortality, and fish size. The data of the figure clearly indicate that 50 mm is a critical length for fish survival. That is, fish less than 50 mm survive significantly longer than do fish larger than 50 mm. Although this length differs slightly from that predicted by Jensen et al. (1986) (i.e., about 65 mm), the data of Figure 10 qualitatively support their predictions. The data further suggest that, because of the significant differences in time to mortality, the mechanisms responsible for mortality in small fish may differ from those responsible for mortality in larger fish. Another feature of the data in Figure 10 is the set of data highlighted by the box. These correspond to conditions of hyperoxia in the experimental environments. Later in this section, additional analysis will be presented which shows hyperoxia can dramatically increase fish survival. This is also a prediction of the GBT threshold equations of Fidler (1984 and 1988) which will be examined in Section 6.1.3.

6.1.2.3 Species Thresholds

A further refinement of the analysis can be achieved by separating the records of Table C3 into two groups, one containing data for fish up to 50 mm in length and the other containing data for fish over 50 mm in length. When the records for fish up to 50 mm in length are plotted as time to mortality versus water delta P, the results are as shown in Figure 11. Although different symbols have been used to distinguish fish species, no species variability is evident from the figure.

When the data records for fish over 50 mm are filtered by fish species and plotted as time to mortality versus water delta P, the results shown in Figures 12 through 16 are obtained. In these figures, data have been restricted to the range of 20% to 70% levels of mortality. The lower limit was set to assure a significant level of response. As was mentioned earlier, much of the data from the literature was generated without control experiments or the response of controls was not reported. It was anticipated that the 20% level of response would distinguish mortality as a result of DGS from mortality due to other causes, had controls been used and/or reported. In the case of the upper limit, there was very little data for mortality above the 70% level.

The data of Figure 12 are for sockeye salmon. It is evident in the figure that fish of this species over 50 mm in length have a delta P threshold for mortality at about 125 mm Hg (sea level TGP% about 116%). The existence of this threshold is supported not only by the absence of mortalities at lower water delta P values, but also by the large number of data points at delta P levels less than 125 mm Hg for which there were no mortalities.

When time to mortality data for cutthroat trout are plotted versus water delta P, the results are as shown in Figure 13. Although the number of data points are not as abundant as for sockeye salmon, a threshold in delta P for mortality is clearly indicated. For this species the threshold is slightly lower at a water delta P of 116 mm Hg (sea level TGP% about 115%).

For steelhead trout, time to mortality versus water delta P is shown in Figure 14. Again, a threshold with characteristics very similar to that for sockeye salmon and cutthroat trout is indicated at a water delta P of

Figure 10: Time to Mortality Versus Fish Length

Figure 11: Time to Mortality for Fish up to 50 mm in Length

Figure 12: Time to Mortality for Sockeye Salmon over 50 mm in Length

Figure 13: Time to Mortality for Cutthroat Trout over 50 mm in Length

Figure 14: Time to Mortality for Steelhead Trout over 50 mm in Length.

About 115 mm Hg. (sea level TGP% about 115%). However, there appears to be another threshold at a delta P of about 76 mm Hg (sea level TGP% about 110%). When time to mortality is plotted versus water delta P for chinook salmon, the result is as shown in Figure 15. There appears to be a water delta P threshold in the vicinity of 130 to 140 mm Hg (sea level TGP% about 117% to 118%), and there may be yet another delta P threshold at 76 to 78 mm Hg. Both of these conceivable thresholds are characterized by a widening span of time to mortality values as water delta P is decreased toward the threshold. Finally, when time to mortality is plotted versus water delta P for coho salmon the result is not as clear as for the previous species (Figure 16). In Figure 16 there may again be a threshold indicated at a water delta P of about 133 mm Hg (sea level TGP% about 117%). However, there are many mortalities indicated at a water delta P of 87 mm Hg which may also represent a lower threshold. There are no mortality data below this level or between this level and the higher 130 mm Hg level to support this hypothesis.

6.1.2.4 Effect of Water Depth

An additional graphical analysis of time to mortality is valuable and involves data of Table C3 obtained by Knittle et al. (1980) for cutthroat trout. In these experiments, fish were confined to cages and restricted to specific water depths for a range of water delta P values. Figure 17 shows the result of these experiments, again plotted as time to mortality versus water delta P. The open symbols of the plot are the original data plotted independent of water depth. The closed symbols are the data with delta P corrected for water depth (i.e. delta P_corrected = delta P_uncorrected - 73.89 times water depth in metres). It is clear that the corrected time to mortality data closely follow a classic dose-response curve for toxicants and other stress factors. Furthermore, there is a threshold clearly evident near a water delta P of about 145 mm Hg (sea level delta P about 118%). The vertical line in the figure is the threshold for cardiovascular bubble growth defined by Equation 6 which will be described in Section 6.1.3. The close correlation between theory and experimental data is apparent. Lower values of water delta P were not examined in these experiments and the presence of yet another threshold cannot be assessed. In addition to supporting the existence of theoretically derived thresholds for mortality, the data of Figure 17 show that time to mortality is directly dependent on water depth and that the correction factor for depth is approximately 74 mm Hg per metre.

6.1.2.5 Effect of Water pO₂

One further analysis of the data from Table C3 shows the importance of water pO₂ on time to mortality. Rucker (1975b) examined the effect of water pO₂ on time to mortality in coho salmon. Figure 18 shows the results of these experiments. As indicated in the figure, the water delta P is for the range 146 to 148 mm Hg which is well above the mortality thresholds in delta P which are indicated in Figures 12 through 16 and the theoretical thresholds for cardiovascular bubble growth (Section 6.3.1). As water pO₂ increases beyond normoxic values, there is a gradual rise in time to mortality. This rise indicates a dependency of time to mortality on water pO₂ and that higher levels of pO₂ extend the survival time of fish. This result is consistent with statistical Model 8 of Jensen et al. (1986), described earlier. However, between a water pO₂ of 250 and 270 mm Hg there is a dramatic increase in the time to mortality as well as an absence of mortality in some of the experimental animals. This sudden increase in time to mortality is a strong indicator that a threshold condition has been exceeded. In this case, the threshold is one which is dependent on water pO₂ and one beyond which the mechanism responsible for mortality at lower pO₂ levels is no longer effective. Thus, there is clear experimental evidence that thresholds in mortality are dependent on water pO₂ and that water quality

Figure 15: Time to Mortality for Chinook Salmon greater than 50 mm in Length

Figure 16: Time to Mortality for Coho Salmon greater than 50 mm in Length

Figure 17: Time to Mortality Data of Knittle et al. (1980)

Figure 18: Time to Mortality Versus Water pO₂

guidelines for DGS must account for this effect. The dependency of thresholds for mortality on water pO₂ is also predicted by theoretical models which will be described in Section 6.1.3. The high values of time to mortality beyond a water pO₂ of 250 mmHg are the same points which were outlined in the box in Figure 10.

6.1.2.6 Summary of Graphical Analyses

The above analyses support the hypothesis that the lethal signs of GBT are dependent on thresholds in water delta P, which in turn are dependent on water pO₂ and depth. There is evidence from Figures 12 through 16 that two distinct thresholds for GBT mortality may be present. There is clear evidence of a single threshold in delta P of 125 mm Hg for sockeye salmon. An additional threshold at a lower delta P may be present, but its absence could be due to insufficient data at lower delta P levels. Cutthroat trout and steelhead trout appear to have a threshold at a delta P of about 115 mm Hg. In the case of steelhead trout, there may be a lower threshold as well at a delta P of 76 mm Hg. This lower threshold at a delta P of 76 mm Hg may also be present for cutthroat trout. However, as with sockeye salmon, its presence may be obscured by a lack of data at the lower delta P levels. For chinook salmon there is evidence for a threshold in the range of delta P about 130 to 140 mm Hg and perhaps another threshold at delta P = 76 to 78 mm Hg. The situation involving coho salmon is unclear, but the evidence does not rule out the presence of two thresholds. However, additional experimental studies would be needed before this can be established.

The indication that there may be more than one threshold for mortality implies that different mechanisms may be responsible for mortality. In addition, the higher thresholds in the delta P range of 115 to 145 mm Hg appear to vary with species, suggesting that there are species differences in resistance to GBT. If so, species would be another variable which must be accounted for in assessing the impacts of DGS on fish and in the derivation of water quality guidelines for DGS.

Although the graphical analysis described above demonstrates the existence of thresholds, there is no indication of how the thresholds are associated with the various signs of GBT. In the sections which follow, it will be shown that the lower threshold indicated in Figures 14 and 15, (delta P of about 76 mm Hg) corresponds to that at which the growth of extra-corporeal inter-lamella bubbles and sub-dermal emphysema of skin surfaces begins. The higher thresholds of Figures 12 through 16 (delta P about 115 - 140 mm Hg) correspond to that at which bubble growth in the cardiovascular system begins.

6.1.3 Biophysical Studies

At the University of British Columbia, Fidler (1984 and 1988) and Shrimpton et al. (1990a and b) conducted both theoretical and experimental studies of the biophysics of GBT in rainbow trout. The effects of DGS on physiological parameters such as swim bladder pressures, intra-corporeal and extra-corporeal bubble formation, blood pH, blood pO₂, and blood catecholamines were examined in relation to water delta P and pO₂, water depth, and fish size. By combining the results of these studies with those from the graphic analysis described above (Figures 12 through 18), Fidler (1984 and 1988) and Shrimpton et al. (1990a and b) were able to establish parameters in a series of equations which predicted the thresholds in water delta P for specific signs of GBT in fish (i.e., bubble formation in the cardiovascular system, over-inflation of the swim bladder in young fish, extra-corporeal bubble formation in gill lamella, and sub-dermal emphysema on body surfaces including the lining of the mouth). These equations and the basis for their development and validation are described next.

Fidler (1984) derived the following equations which define thresholds in dissolved gas levels for the major signs of GBT.

Delta P_SB = 73.89· h + 0.15· pO₂ Eq. 4

Delta P_EW = 73.89· h + 83.0 Eq. 5

Delta P_CV = 73.89· h + 0.21· pO₂ + 83.0. Eq. 6

Where: Delta P_SB = water delta P required to initiate over-inflation of the swim bladder in rainbow trout.

Delta P_EW = water delta P required to initiate sub-dermal emphysema and extra-corporeal bubble growth between gill lamella.

Delta P_CV = water delta P required to initiate bubble growth in the cardiovascular systems of rainbow trout.

H = water depth at which the fish is located - metres.

pO₂ = partial pressure of dissolved oxygen (mm Hg) in the environmental water.

The equations were derived from analyses of bubble growth processes associated with decompression, cavitation, nucleate boiling, and other physical processes (Fidler 1984). The basis for the equations centres on the concept that nucleation sites are involved in phase changes between liquids and gases (Harvey et al. 1944, Fox and Herzfeld 1954, Hlastala and Fahri 1973, Yount 1979). Because surface tension and other surface phenomena impose restrictions on the stability of these nucleation sites, thresholds in water delta P are an immediate consequence. The application of these stability criteria to bubble growth in fish and to over-inflation of the swim bladder involves additional considerations in terms of gas exchange between the fish and the water environment. Diffusive and convective resistance at the gill reduce blood dissolved gas tensions from those of the environmental water (Randall and Daxboeck 1984). As a result, the thresholds for bubble growth in fish differ from those for bubble growth in the environmental water. These principles were incorporated into the derivations presented by Fidler (1984 and 1988) which resulted in Equations 4, 5, and 6.

In Equations 4, 5 and 6, the factor 73.89 converts water depth to hydrostatic pressure in mm Hg. As the equations imply, water depth is a major factor in establishing the thresholds for signs of GBT. Every metre of depth requires approximately 74 mm Hg of additional delta P to initiate a particular sign. Thus, water depth, if available and used by fish, can play an important protective role for fish exposed to high levels of DGS.

The 0.21 coefficient in Equations 6 accounts for the reduction of dissolved oxygen in arterial blood from that in the environmental water. The value 0.21 was established through an analysis of experimental data from the scientific literature which describes blood pO₂ levels in adult rainbow trout (Fidler 1988). In Equation 4, the coefficient multiplying the pO₂ term is 0.15. This value was established from the work of Shrimpton et al. (1990a and b) and is based on studies of swim bladder over-inflation in juvenile rainbow trout. The 0.21 and 0.15 coefficients in the equations imply that the delta P required to initiate swim bladder over-inflation and cardiovascular bubble growth increases as water pO₂ increases. Again, this is in agreement with the statistical modeling studies of Jensen et al. (1986) and the graphical analysis of GBT data presented earlier. It will be noted that Equation 5, which describes the threshold for extra-corporeal bubble growth and sub-dermal emphysema, is independent of water pO₂. In the case of extra-corporeal bubbles, there is no reduction in dissolved oxygen levels, while for the swim bladder and cardiovascular bubbles there is a reduction. Sub-dermal emphysema appears to involve direct diffusion of gases from the water to nucleation sites just beneath the skin surface. Again, there is no reduction in dissolved oxygen levels (Fidler 1988).

The 83.0 parameter in Equations 5 and 6 accounts for the combined effects of blood or water surface tension, blood pressure, and the size of microscopic nucleation sites upon which bubble growth in the vascular system or in the environmental water is initiated. In the case of the swim bladder, this parameter is zero due to the large size of the swim bladder (Fidler 1988). It was through a series of laboratory experiments using rainbow trout and the graphical analysis of experimental data, described earlier, that a value of 83.0 was established for this parameter (Fidler 1988).

Water temperature does not appear explicitly in Equations 4, 5, and 6. Although the statistical models of Jensen et al. (1986) indicated that ET₅₀ values were dependent on water temperature for values below 9° C, it is not clear that this should be the case for bubble growth thresholds. There are several reasons why temperature should not have a strong effect on bubble growth thresholds and yet affect time to mortality. First, thresholds for bubble growth are related to conditions of static stability at the gas - water or gas - blood interface (Harvey et al. 1944, Fox and Herzfeld 1954, Hlastala and Fahri 1973, Yount 1979, Fidler 1984 and 1988). As such, mass transfer operations are not involved until bubble growth or collapse actually begins (Fidler 1984). There is a weak dependency of GBT delta P thresholds on temperature as a result of the effects of surface tension. However, Fidler (1984) showed that the surface tension of fish blood was very nearly the same as that of water and neither varied significantly over the range of biologically significant temperatures. There is also a small effect of water temperature on GBT delta P thresholds as a result of variations in the vapour pressure of water with temperature (Figure 1). As related in Section 4.1, water vapour is included as a component of delta P and it is always considered to be in a saturated state (Colt 1986, Fidler 1988). By examining Figure 1 it can be seen that even at a temperature of 25 °C, water vapour would account for only 3% or less of any positive delta P. Thus, its effects on thresholds should be small.

On the other hand, ET₅₀ values are related to the actual bubble growth or swim bladder inflation processes. These, in turn, are dependent on the rate at which dissolved gases are transferred from water or blood to a bubble or to the swim bladder. These rates are controlled by diffusion coefficients, convective mass transfer coefficients, and oxygen demand of the animal, all of which are strong functions of water temperature.

Figure 19 shows Equations 4, 5, and 6 plotted in terms of GBT delta P's versus threshold water depth for a water temperature of 10 °C and a water pO₂ of 157 mm Hg (sea level normoxic). In some cases, the depths of Figure 19 can be interpreted as compensation depths or those depths below which the particular GBT sign may or may not occur. However, it is important that care be taken in applying this interpretation. For example, depending on the initial inflation pressure in the swim bladder, which in many situations is determined by the fish independently of DGS (Section 8.1.1.5), the swim bladder would over-inflate when a fish moves above the compensation depth. When the fish moves below the compensation depth, the swim bladder would deflate. Thus, the threshold line for swim bladder over-inflation of Figure 19 is a true compensation threshold.

Figure 19: Threshold Depths for GBT Signs Versus Water Delta P

For the growth of intra-corporeal and extra-corporeal bubbles, a slightly different interpretation of the thresholds of Figure 19 is required. If bubble growth has not been initiated and a fish stays below the depth corresponding to the particular bubble growth threshold, bubble growth would not be initiated. However, once the fish moves above the threshold depth and bubble growth begins, moving back below the threshold depth would not necessarily stop bubble growth. This is because once the bubble radius has increased, growth can continue at delta P values lower than those required to initiate growth (Harvey et al. 1944, Fox and Herzfeld 1954, Hlastala and Fahri 1973, Yount 1979, Fidler 1984). Bubble growth would stop or reverse only after the fish moves to a depth below that at which growth began. In the case of extra-corporeal water bubbles this could be up to 0.8 metres below the threshold depth and up to 1.2 metres below the threshold depth for cardiovascular bubbles.

Some general observations are important to the application of Equations 4, 5, and 6 and the threshold depths shown in Figure 19. First, in the experimental studies and data analysis which were performed to validate Equations 4, 5, and 6 (Fidler 1988, Shrimpton et al. 1990a and b, White et al. 1991), it was found that the threshold delta P required for initiation of sub-dermal emphysema was very close to the delta P at which extra-corporeal bubble growth between gill lamella began and the delta P at which swim bladder rupture occurred in small fish. Thus, there is an area of overlap in terms of thresholds and signs of GBT. It should also be noted that mortalities associated with the sign of extracorporeal bubble growth between gill lamella, in combination with sub-dermal emphysema of the lining of the mouth (Fidler 1988), were observed with captive, restrained fish in shallow water environments. It is not known if these signs would produce mortalities in wild fish where swimming activity could periodically dislodge bubbles from between gill lamella and possibly prevent death. This note also applies to the data of Table C1 of Appendix C. Most of the data listed in the table were collected under conditions where fish were not able to swim freely. A final consideration in relation to bubble growth in the environmental water (i.e., extra-corporeal bubbles between gill lamella) is the effect of nucleation site size. This threshold was established in laboratory water which was considered "clean" (i.e., free of suspended particulate matter). In natural environments which contain high concentrations of suspended particulate matter there may be nucleation sites considerably larger than those present in "clean" water supplies. In these environments, it should be expected that bubble growth can begin at delta P levels lower than those predicted by Equations 5.

6.1.4 Chronic GBT and Swim Bladder Overinflation

The graphical analyses presented in Section 6.1.2 imply that water delta P levels below 76 mm Hg should protect fish from the effects of DGS. Yet, Wright and McLean (1985) found that, over a 122-day exposure period, there is increased mortality in juvenile chinook salmon held in a shallow water environment having delta P levels ranging from 0 to 46 mm Hg. In these experiments, no direct relationship was established between the observed mortalities and any specific sign of GBT. Also, Cornacchia and Colt (1984) found increased mortality in striped bass (Morone saxatilis) at delta P levels of 42 mm Hg. The mortality appeared to be caused by swim bladder over-inflation and bubbles in the gut. Dannevig and Dannevig (1950), Henly (1952), Peterson (1971), and Kraul (1983) have also observed similar effects in other fish species exposed to low levels of DGS. Over-inflation of the swim bladder was a common sign and was suspected as being indirectly responsible for mortality. It is hypothesized that uncompensated over-buoyancy caused by swim bladder over-inflation imposes additional swimming demands on fish and that the resulting stress eventually leads to increased mortality. To date, there are no data which confirm this hypothesis directly. The data which are available (Cornacchia and Colt 1984, Wright and McLean 1985, and Shrimpton et al. 1990a and b) provide only circumstantial evidence. Nevertheless, considering the extent to which experimental conditions were monitored by Wright and McLean (1985), it is unlikely that the increased mortality they reported was due to any other cause.

Accepting that swim bladder over-inflation is a chronic cause of mortality in fish, the water delta P thresholds at which this occurs become crucial to the development of water quality guidelines. As indicated in Figure 19, the delta P threshold for swim bladder over-inflation predicted by Equation 4 is as low as 25 mm Hg. In order to define the conditions under which swim bladder over-inflation occurs and the effects of over-buoyancy on fish, Shrimpton et al. (1990 a and b) conducted an extensive series of experiments using rainbow trout. The results of these experiments along with relevant background information are examined in the next section.

6.1.5 Thresholds for Swim Bladder Over-inflation

Like most fresh water fish species, rainbow trout are more dense than water and they possess a swim bladder which is used to control buoyancy. In physostome fish, the swim bladder is connected to the esophagus by a small-diameter pneumatic duct (Fänge 1983). The pneumatic duct serves as a path for filling the swim bladder with atmospheric air (Harvey 1963). Physoclist fish also have swim bladders, but do not have a pneumatic duct. In these fish, the swim bladder is filled by way of a complex gas gland which secretes oxygen directly into the swim bladder (Fänge 1983). When physostome fish are frightened, the pneumatic duct can be used to vent air as a means of quickly reducing buoyancy (Harvey 1963). Presumably, this enhances the fish's ability to swim to deeper water and seek cover. Shrimpton et al. (1990 a and b) found that in supersaturated water the swim bladder can become over-inflated as a result of dissolved gases diffusing from the water to the swim bladder by way of the gills and vascular system. When this happens, fish become overbuoyant. It appears that rainbow trout are unable to control the venting of gas through the pneumatic duct under conditions of DGS. This is probably true of other trout and Pacific salmon species as well.

Through their experimental studies, Shrimpton et al. (1990 a and b) found that in fish larger than 200 g in weight, swim bladder over-inflation was generally not a problem, with the pressure in the swim bladder seldom exceeding 10 mm Hg. However, in fish much less than 200 g in weight, swim bladder pressures approaching 70 mm Hg were observed and swim bladder rupture was frequently present at these high pressures. Figure 20 shows the results of these studies where swim bladder venting pressure (or pneumatic duct release pressure) is plotted as a function of fish weight. Clearly, as weight (or fish size) decreases, the venting pressure (or swim bladder overpressure) increases sharply. As a possible explanation for this response, it was hypothesized that for fish much less than 200 g in weight, the small diameter of the pneumatic duct created high tension surface forces at the gas - water interface in the duct. This effectively blocked the flow of gas through the duct, allowing the observed high pressures to develop within the swim bladder (Fidler 1984 and 1988).

The results shown in Figure 20 help explain the differences in time to mortality between large and small fish which were identified by Jensen et al. (1986) and in the graphical analysis presented in Section 6.1.2. That is, swim bladder over-inflation is a problem for small fish only and the mortalities resulting from this sign are generally chronic in nature. For larger fish, mortality is the result of other signs of GBT (i.e., extra-corporeal inter-lamella bubbles, sub-dermal emphysema, cardiovascular system bubbles, etc.) which tend to be more acute in nature.

Other results from the experiments by Shrimpton et al. (1990a and b) appear to support the validity of the threshold equation for swim bladder over-inflation (Equation 4). This is based on measurements of the rate of swim bladder inflation or deflation in rainbow trout as a function of water delta P and pO₂ levels. The experimental results are shown in Figure 21 where conditions for increasing or decreasing swim bladder pressure are plotted as a function of water delta P and pO₂ levels. The solid symbols indicate conditions of swim bladder inflation while the open symbols indicate conditions of swim bladder deflation.

Also plotted in the figure is Equation 4, the threshold equation for swim bladder over-inflation. With the coefficient of the pO₂ term set to 0.15, the threshold line forms a boundary between those delta P levels which cause swim bladder inflation and those which lead to swim bladder deflation. Thus, the experimental data support the use of Equation 4 to define the threshold for swim bladder over-inflation in small rainbow trout. Because swim bladder over-inflation occurs at the lowest level of DGS, Equation 4 is a useful relationship for describing the critical delta P and pO₂ for signs of GBT in small trout and Pacific salmon.

Shrimpton (1985) and Shrimpton et al. (1990b) addressed one further question related to the sign of swim bladder over-inflation. That is, do fish such as trout and Pacific salmon, use water depth to compensate for over-buoyancy resulting from DGS? Shrimpton (1985) found that for delta P levels up to about 90 mm Hg, coho salmon did compensate for over-buoyancy. However, there was no compensation apparent at higher delta P levels. In subsequent studies using rainbow trout, Shrimpton et al. (1990b) established that small fish exposed to DGS would tend to descend in the water column to a location where they were either neutrally buoyant or slightly under-buoyant. That is, as delta P increased, fish would move deeper in the water column. In these studies, there was no indication of a limit in delta P (up to 150 mm Hg) for which compensation took place.

Figure 20: Pneumatic Duct Release Pressure

Figure 21: Swim Bladder Inflation/Deflation Conditions.

It should be noted that although fish moved to deeper water, this was not a sounding response. That is, fish tended to use the entire water column; however, on the average, they spent more time at or below the compensation depth corresponding to the prevailing levels of DGS. These results demonstrated that, when available, small fish utilize water depth a means of compensating for the effects of swim bladder over-inflation. In doing so, they may also avoid the other signs of GBT.

These studies also found that as fish grew in size, there was less and less tendency to use depth as a means of compensating for DGS. Presumably, this was a consequence of large fish not experiencing swim bladder over-inflation because of the lower venting pressures of the pneumatic duct (Figure 20). This result implies that without the need for buoyancy compensation, larger fish may stay near the water surface where they would be more prone to other signs of GBT.

6.2 Other Fish Species

In addition to information on the response of trout and salmonid species to the effects of DGS, there is some data for other North American fresh water fish species. Time to mortality data which can be used for the development of water quality guidelines are available for carp (Cyprinus carpio), black bullhead (Ictalurus melas), channel catfish (Ictalurus punctatus), mountain whitefish (Prosopium williamsoni), cutthroat trout (Salmo clarki), largescale sucker (Catostomus machrocheilus), torrent sculpin (Cottus rhotheus), and the northern squawfish (Ptychocheilus oregonensis).

6.2.1 Carp and Black Bullhead

Gray et al. (1982 and 1983) examined the survival of carp (Cyprinus carpio) under various conditions of DGS. Carp from Italy, weighing 20.5 g, survived exposures to a delta P of 107 mm Hg for 96 hours, but suffered a 50% mortality after about 9.9 hours at a delta P of 404 mm Hg. The 96-hour LC₅₀ for this species of carp was estimated to be 171 mm Hg. Based on the probit data provided, it appears that the upper limit for delta P which would assure 100% survival is about 114 to 130 mm Hg. Contrasting with these results, Fickeisen et al. (1975) found that Columbia River carp of the same species experienced no mortalities after 96 hours at a delta P of 266 mm Hg.

Gray et al. (1982) also found that black bullhead (Ictalurus melas) from Italy, weighing 27.7 g, survived exposures to a delta P to 55 mm Hg for 96 hours, but experienced a 50% mortality after exposure at a delta P of 435 mm Hg for 5.6 hours. The 96-hour LC₅₀ for the black bullhead was estimated to be 109 mm Hg. Based on the probit data provided, it appears that the upper limit for delta P which would assure 100% survival of this species is about 76 to 90 mm Hg. Again, contrasting with these results, Fickeisen et al. (1975) reported that the 96-hour LC₅₀ for the Columbia River black bullhead was a delta P of 185 to 202 mm Hg.

6.2.2 Squawfish

Bentley et al. (1976) examined the effects of DGS on squawfish (Ptychocheilus oregonensis Richardson). Fish weighing 534 g (364 mm in length) were held in water at a delta P of 76 mm Hg without mortalities over a 12-day period. However, when water delta P was increased to 129 mm Hg, a 10% mortality occurred in 4.8 days. After 12 days exposure, mortalities had not reached the 50% level. As water delta P was increased to 152 mm Hg, the 10% mortality level decreased to 41 hours and the 50% mortality level was achieved in 9.7 days. At a water delta P of 198 mm Hg, the 10% mortality level decreased to 19 hours while the 50% level dropped to 20 hours.

6.2.3 Channel Catfish

Colt et al. (1985) studied the response of channel catfish (Ictalurus punctatus) to DGS. These studies determined that juvenile fish (4.80 to 5.03 g) were fairly resistant to delta P levels up to 76 mm Hg. At this level, a 1% mortality occurred in a 35-day exposure period. As delta P was increased to 117 mm Hg, the level of mortality rose to 56% in 35 days.

6.2.4 Whitefish, Cutthroat Trout, Largescale Sucker, and Sculpin

Fickeisen and Montgomery (1978) examined the effects of DGS on mountain whitefish, cutthroat trout, largescale sucker, and torrent sculpin. Their experimental design involved a large tank with a constant water delta P of 222 mm Hg (sea level TGP% about 129%). Variations in delta P were obtained by holding experimental animals at various depths in the experimental tank (i.e., hydrostatic pressure was used to achieve various effective delta P levels). No information was provided on the size or age class of the fish tested. However, all fish which died during the experiments were examined to ensure that signs of GBT were present. Because the technique for achieving various levels of delta P differed from that of all other experimental data reported in Section 6.1, and because of the lack of information on fish size, the data on cutthroat trout were not included in the analyses of Jensen et al. (1986) or that of Fidler (1988).

For the mountain whitefish (Prosopium williamsoni), they found LT₅₀ values of 12 hours at a delta P of 222 mm Hg, 14 hours at a delta P of 192 mm Hg, 50 hours at a delta P of 161 mm Hg, and 48 hours at a delta P of 130 mm Hg. Although this species was the most sensitive to the effects of DGS, Fickeisen and Montgomery (1978) reported that many of the experimental animals developed signs of piscine tuberculosis. Thus, these results are probably not representative of healthy mountain whitefish.

For cutthroat trout (Salmo clarki), Fickeisen and Montgomery (1978) found LT₅₀ values of 12 hours, 17 hours, 34 hours, and 89 hours at delta P values of 222, 192, 161, and 130 mm Hg, respectively. For largescale sucker (Catostomus machrocheilus), they reported LT₅₀ values of 34 hours, 67 hours, and 103 hours at delta P values of 222, 192, and 161 mm Hg, respectively. At a delta P of 130 mm Hg, 90% of the fish tested were alive at the end of 10 days. They reported that, of all species tested, torrent sculpin (Cottus rhotheus) was the most resistant to the effects of DGS. An LT₅₀ value was reached only at a delta P of 222 mm Hg after 10 days. However, loss of equilibrium due to swim bladder over-inflation was observed at most levels of delta P. They reported ET₅₀ values for loss of equilibrium as 127 hours, 185 hours, and 233 hours at delta P values of 222, 192, and 161 mm Hg, respectively.

6.2.5 Summary - Other Fish Species

The studies described above show that carp, black bullhead, squawfish, channel catfish, mountain whitefish, largescale sucker, and torrent sculpin are either less sensitive or as sensitive to the effects of DGS as the Pacific salmon and cutthroat trout which were described in Section 6.1. However, because of the short duration of the experiments, chronic responses were not evident in the data. Nevertheless, the threshold criteria developed in Section 6.1 should allow the development of DGS water quality guidelines which are protective of these species as well. The data described above have been entered into Table C3 of Appendix C and will be considered in the derivation of fresh water quality guidelines for DGS.

Although several other studies reported in the literature have dealt with the effects of DGS on species other than cutthroat trout and Pacific salmon (Jones and Lewis 1976, Montgomery and Becker 1980, Kolbeinshavn and Wallace 1985, Boon et al. 1987, Tucker 1989, and Backman et al. 1991), the information provided is qualitative only and of little value in the derivation of numerical water quality guidelines. For example, Blahm et al. (1976) reported on the effects of DGS on largemouth bass, rainbow trout, crappie, squawfish, smelt, and mountain whitefish. However, all dissolved gas levels were reported in terms of nitrogen supersaturation and, as such, are incomplete.

6.3 Invertebrates

In addition to the studies of DGS and GBT in fish, there has also been research directed at defining responses to DGS in aquatic invertebrates. Nebeker et al. (1976c) conducted experiments on several species including a Cladoceran (Daphnia magna), western crayfish (Pacifastacus leniusculus), and three stoneflies (Acroneuria californica, A. pacifica, and Pteronarcys californica). The experiments with D. magna involved suspending test animals in small wire cages in a large 0.6 m deep test tank at selected levels of DGS. One aspect of the data which was not reported was the depth at which cages were located in the main tank. The experimental results are summarized in Table 5. The three test series shown reflect differences in the location of the wire cages within the main tank. The authors thought water currents through the cages might affect the response of the animals. This may be the case, as shown by the data in the table; however, it may also be the result of differences in depth at which the cages were located. Examination of the test animals after exposure to the various levels of DGS showed clear evidence of bubbles in the gut and most animals were floating on the water surface. Extrapolated lethal thresholds were given as TGP% = 111% (delta P about 84 mm Hg).

In experiments with western crayfish, Nebeker et al. (1976c) obtained the results shown in Table 6. The variations in the test series are due to a variety of factors including animal size and caged versus free animals in the test tank. Again, it is not clear what depth the cages were placed in the large test tank. At the highest delta P levels, most animals had bubbles within the distended membrane between the carapace and abdominal segments. Bubbles were also present in the gills and body fluids. Extrapolated lethal thresholds were given as TGP% about 120 to 127% (delta P about 152 to 205 mm Hg)

Table 5: Time to Mortality Data for Daphnia magna
(Nebeker et al. 1976c)

Test Series	Nominal delta P	Time to 50% Mortality - hr	Time to 20% Mortality - hr
1	152	91	38
1	228	65	45
1	304	71	48
2	152	210	131
2	228	130	92
2	304	123	72
2	380	101	82
3	76	No mortality	No mortality
3	114	No mortality	137
3	152	93	49

Although their studies of aquatic insects produced no mortalities for water delta P's ranging from 114 to 266 mm Hg, Nebeker et al. (1976c) found bubbles adhering to ventral tracheal gill masses and in internal body fluids at the higher delta P levels. Several animals were distended like balloons due to internal bubbles.

White et al. (1991) examined the effects of DGS on river invertebrate communities of the Bighorn River below the Yellowtail Afterbay Dam. Although the study results were not in a form from which mortality - delta P relationships could be developed, several observations are worth noting. The general effect of DGS on all taxa of invertebrates examined was the presence of both internal and external bubbles at river delta P levels ranging from 76 to 150 mm Hg. As a result, most organisms were trapped at the water surface by excess buoyancy. Advanced signs of GBT included protraction of the head from the thorax, separation of abdominal segments, and loss of torsal mobility.

Table 6: Time to Mortality for Western Crayfish
(Nebeker et al. 1976c)

The above results indicate that fresh water invertebrates are either less sensitive or as sensitive to some of the signs of GBT as fresh water fish. Therefore, the threshold information developed in Section 6.1 should provide a conservative basis for the development of DGS water quality guidelines which are protective of fresh water invertebrates.

6.4 Amphibians

Colt et al. (1984a) exposed bullfrog tadpoles (Rana catesbeiana) to delta P levels of 160 to 170 mm Hg for four days with no apparent effect. When exposure was increased to ten days, mortalities increased along with a systemic bacteria infection. The intestinal tract and gallbladder were also filled with gas bubbles. Colt et al. (1987) exposed adult bullfrogs (R. catesbeiana) to several levels of DGS. At the highest level (delta P = 240 mm Hg), a 40% mortality occurred in a 24-hour period. At all levels of delta P above 128 mm Hg, animals had extensive blistering of external skin surfaces and bubbles in the vascular system. Colt et al. (1984b) also exposed adult African clawed toad (Xenopus laevis) to DGS. The authors reported that extensive bubble formation occurred in inter-digital webbing and sub-cutaneously on body surfaces. Death resulted from bubble formation in the vascular system and secondary bacterial infections. None of these data allowed estimates of time to mortality as a function of DGS levels.

6.5 Plants and Algae

No data were found in the literature which describe the effects of DGS on aquatic plankton, algae, or vascular plants. Nevertheless, based on the signs of GBT in fish and other aquatic organisms, in conjunction with an understanding of the processes involved in bubble formation and growth, some effects of DGS on aquatic plants can be anticipated. For example, if bubbles form in the environmental water and become attached to plankton and algae, these plants may float to the water surface. Since bubble formation in "clean" water appears to occur at a delta P of about 76 mm Hg, this is a threshold which may apply to aquatic plants.

Still, the detrimental effects of excess buoyancy to plants is unknown. One effect which might appear at high levels of DGS is a concentration of plants at the water surface which could enhance oxygen production in the surface water layers through photosynthesis. This would increase the delta P to even higher levels. Although this might cause problems for fish and invertebrates, it may still not affect the survival of the plants themselves. As pointed out earlier, the presence of suspended particulate matter could lower the delta P thresholds at which bubbles would form in the environmental water. However, there is no information presently available on how this would affect aquatic plants and algae. Another sign which could be anticipated in vascular plants is the formation of bubbles internal to the plant. At present, there is no information on the delta P levels at which this would occur.