Discussion and Conclusions


In order to draw conclusions regarding the behavior of cyanobacterial colonies in Upper Klamath Lake based on RB and vertical velocities measured by ADCPs, one must be confident that the scatterers reflecting the acoustic signal back to the instruments are primarily cyanobacterial colonies. There is no way to differentiate among actively growing AFA, other organisms, and resuspended organic and inorganic material in RB data. However, most of the suspended material collected from the ADCP sites in September 2006 was organic material: 62–95 percent at the shallow site and 48–85 percent at the deep site; the organic fractions near-surface generally were somewhat higher than organic fractions near-bottom (Gartner and others, 2007). The organic fraction comprises both phytoplankton and zooplankton. AFA completely dominates the phytoplankton species from June to October, reaching as much as 50 mg/L (wet weight) (Kann, 1998). The dominant zooplankton in the lake is Daphnia pulicaria; typical maximum seasonal concentrations are about 100 individuals per liter, and dry weights seldom exceed a few milligrams per liter (Kann, 1998). The D. pulicaria have the potential to affect the measured backscatter (and vertical velocity), especially because of their large size—the average length of the largest size class of D. pulicaria was 1.74 mm in 2005, and the average dry weight per individual was 20.75 µg/animal (A. St. Amand, Phycotech, unpub. data, 2005). The number of cyanobacterial colonies in the water column, however, typically is at least an order of magnitude higher than the number of D. pulicaria, suggesting that D. pulicari and other zooplankton contribute a minor component to the measured backscatter and vertical velocity.


With regard to the inorganic fraction, resuspension of the bed material would be consistent with the observed positive and increasing correlation of RB with horizontal currents with distance beneath the water surface at all measurement sites except site ADCP6. However, resuspended bed material in Upper Klamath Lake would be expected to be at least partially organic silt (Laenen and LeTourneau, 1996; Wood, 2001). Northwest winds greater than 4.5 m/s are sufficient to resuspend bed sediment (Laenen and LeTourneau, 1996). Typical SSC values were calculated to be about 200 mg/L at low lake elevations of 4,137 ft (1,260.96 m, corresponding to a mean lake depth of 1.5 m) but only about 5 mg/L at a lake elevation of 4,140 ft (1,261.87 m corresponding to a mean lake depth of 2.4 m) (Laenen and LeTourneau, 1996). Lake elevations ranged from 4,143.1 to 4,139.5 ft (1,262.82 to 1,261.72 m) during the 2005 ADCP deployments. (Lake elevations are from USGS water level gage 11507001.) Thus, with deeper lake levels, even with average diel winds exceeding 4.5 m/s for several hours during late evening and early morning, typical values of resuspended material were expected to be low in 2005. In addition to predictions for relatively low levels of mobilized bed material, the measured RB was usually less near-bottom than it was near-surface on both daily and seasonal time scales at each of the five ADCP sites. Although wind speeds were periodically high enough to resuspend bed materials and distribute mass throughout the water column, evidence of resuspended material should be evident with increased RB measurements near-bottom. Correlations of RB with current speeds were weaker than for RB with wind or RB with ΔT°; correlations were particularly poor at sites ADCP3 and ADCP6. In addition, the timing between RB and wind and RB and current speeds varied greatly among the five ADCP sites. If resuspended bed material were a substantial portion of the suspended material in the water column, it would be expected that the extrema of diel cycles of RB generally would coincide with the extrema of diel cycles of near-bottom currents, which was not the case except at sites ADCP3 and ADCP5 where the extrema in RB and near-bottom currents occurred within about an hour of each other. Mass was infrequently concentrated in the lower part of the water column, although it did happen occasionally, particularly at site ADCP1 where the current speeds were highest (fig. 9). Thus, the observations tended to support the hypothesis that the correlation of RB with currents near- bottom was related more to mixing of suspended material downward from the surface than to resuspension of bed materials. Finally, the computed SSC was correlated with dissolved oxygen and chlorophyll a, other water quality indicators of bloom dynamics. It seems reasonable to conclude, therefore, that RB is a useful surrogate for cyanobacterial mass (dominated by AFA) in Upper Klamath Lake although the presence of unknown, but generally small amounts of inorganic and organic sediments, detrital AFA particles, and zooplankton may decrease accuracy of results. 


The theoretical model of light-driven vertical migration does not take into account the entrainment into turbulent eddies that occurs in wind-driven surface mixed layers (Reynolds, 2006). In Upper Klamath Lake, this mixed layer can extend to the bottom when winds are strong or to less than a meter from the surface when winds are light and the top-to-bottom temperature difference in the water column provides resistance to mixing. Because Upper Klamath Lake is shallow and mixes on an almost daily basis, it is likely that the vertical migration pattern of the colonies is controlled by the concurrent physical turbulence and thermal (density) characteristics of the water column as well as light availability. 


The seasonally averaged diel cycles in measured RB, surface vertical velocity, and computed SSC provided insight into the behavior of AFA colonies on diel time scales. The diel pattern in RB showed a maximum near the surface of the water column on a daily basis at all sites as the thermal stability of the water column increased in the afternoon, the result of a combination of rising colonies accumulating there and growth. The afternoon setup of thermal stability also resulted in a minimum in the measured RB near the bottom at each site as floating colonies rose to the surface and sinking colonies settled out of the water column. A minimum in colonies near the surface and the most uniform distribution of colonies through the water column coincided with the period from late night to mid-morning associated with the minimum in temperature stratification (figs. 3 and 9).


Interpretation of the seasonally averaged diel cycle in vertical velocities was complicated by the fact that the measured velocity is an average of the motion of all suspended particles sampled by the ADCP acoustic beams, which may be a mix of positively, negatively, and neutrally buoyant particles. At the three shallowest sites, there was little evidence of any diel cycle in the surface vertical velocities, which generally were negative, and between -0.1 and -0.2 cm/s, except at site ADCP3 where positive velocities as much as 0.1 cm/s dominated the diel pattern. There were small diel fluctuations that were not consistent from site to site, but there was no indication of a daily switch between positively and negatively buoyant colonies consistent with light-driven migration. Only at the two deepest sites, ADCP1 and ADCP5, was there clear evidence of a diel cycle. At site ADCP5, that pattern was manifested as smaller sinking velocities in the afternoon; near‑surface vertical velocity ranged from about -0.4 cm/s in the early morning to about -0.15 cm/s in the mid-afternoon. The diel cycle was strongest at site ADCP1, where the measured vertical velocity ranged from -0.15 cm/s in the early morning to +0.15 cm/s in the late afternoon. Both the floating and sinking velocities were large, implying that mass could be depleted or could accumulate at the surface in a matter of a few hours, which is supported by the highly dynamic diel cycles in surface RB. This is consistent with observations by Moreno-Ostos and others (2009), who concluded that diatom and AFA mass at the surface responded to changes in the wind stress within about 30 and 80 minutes, respectively. 


When RB was converted to SSC, it was possible to calculate the movement of mass vertically through the water column. At all sites, based on the seasonally averaged diel cycle, mass was concentrated in the upper water column through the afternoon and evenly distributed between the upper and lower water column in the early morning hours, consistent with growth in the upper water column in the afternoon, and mixing of the water column in the early morning. As was the case with the RB data, there was limited occurrence of mass concentration in the lower water column that would indicate potential resuspension of sediments. The depth-integrated mass at each site peaked in the late afternoon to early evening, usually a few hours later than the maximum in the upper water column. The difference between the daily minimum and maximum in depth-integrated mass at each site was large, ranging from 19 percent of the average at site ADCP1 to 92 percent of the average at site ADCP6. The highly dynamic nature of the diel cycle required large loss rates due to sedimentation and mortality within the water column, and large growth rates. 


On subseasonal time scales, negative correlations of RB and air temperature decreased from near-bottom upward and became positive at near-surface in the water column at most of the ADCP sites. Positive correlations of RB and current speeds decreased from near-bottom upward and became negative near-surface at three ADCP sites. Because resuspended bed sediments were a minor component of suspended material, this correlation probably resulted primarily from a dispersal of colonies throughout the water column when the water column mixed more easily. The negative correlation near-surface at three ADCP sites indicated that colonies accumulated near‑surface under conditions of increased water column stability. Moreno-Ostos and others (2009) observed the same relation between the vertical distribution of buoyant AFA and wind-induced mixing, and demonstrated that this behavior of positively buoyant colonies was easily distinguished from the behavior of negatively buoyant diatoms, which became depleted at the surface and accumulated in the lower water column when mixing was suppressed. At the two deepest sites, the surface vertical velocities were positively correlated with air temperature, consistent with decreased entrainment of floating colonies when air temperature increased. These correlations were consistent with increased concentration of AFA colonies near the surface when water column stability increased and dispersal of colonies throughout the water column when the water column mixed more easily. The depth-integrated mass at the two deepest sites was negatively correlated with air temperature and positively correlated with horizontal current speed. Thus, although the total water column SSC peaked with the highest air temperatures of the day on a daily basis, higher air temperatures were associated with lower total water column SSC on a subseasonal time scale. At site ADCP7, located at the mouth of Howard Bay, the correlations were inconsistent, but the maximum horizontal current speeds occurred near-bottom rather than near-surface or at mid-depth as they did at the other sites (Gartner and others, 2007; fig. 3). The near-surface currents and advection of suspended material were complicated by the movement of water across the entrance to Howard Bay at this site.


Successful application of a surrogate technology such as analysis of time series of measurements of RB in conjunction with analysis of water samples has the potential to provide a less costly and labor intensive means of determining continuous and high resolution characteristics of the variable under study that is not otherwise possible. Analyses of both RB and computed SSC, in conjunction with profiles of vertical velocity and an understanding of horizontal water current speeds measured by ADCPs in Upper Klamath Lake during 2005 contribute significant additional insights into seasonal, subseasonal, and diel patterns of AFA population dynamics in the lake. 


First posted March 16, 2011