Discussion

The particle populations sampled in O'Ne-eil Creek have been shown to be highly aggregated with floc factors (EPSD/APSD) exceeding ten.3 All of the particles evaluated in this paper for size, morphology, and settling characteristics can be shown to be aggregates as the maximum particle sizes of the disaggregated inorganic sediment (APSD) of these samples do not exceed 130 ^m, while each of the particles evaluated visually (EPS) was larger than 150 ^m. The only particles which should be considered separately in this context are the compact-organics as they in fact may not be composite particles, but could be parts of organisms or aggregates of organism parts. The organics are removed from the sediment sample before characterizing the ASPD and so their constituent size is not determined.

Visual differentiation of aggregate particles in freshwater allowed an investigation of the size, shape, and behavior of particle subpopulations identified as flocs and compact particles. The loosely packed, open matrix floc particles are observed to be larger, less dense, less rounded, and slower settling than the opaque, compact particles. They were noted to be less abundant than compact particles in all periods sampled except for the fish die-off. While the smaller, rounder, denser compact particles are carried in ambient low flow conditions (0.23 to 0.35 m s-1) their proportions are smaller in the post-fish quiescent flows (79%) than in low flows with active spawning occurring (93%). In the resuspended gravel stored sediment of active spawn and post-fish they comprise 85% and 83%, respectively (Table 4.3). It is unfortunate that the spring melt population could not be visually differentiated as the energy environment of that period was high and both the proportion of floc and compact particles would be of interest as well as their respective sizes and shapes.

In any case, the characteristics of the total population can still be useful as it is clear that the particle size of the populations change over the open water season with the smallest mean sizes observed during spring melt when local flows are deepest and fastest and the concentration of suspended sediment is higher (mean of 7.96 mg l-1) for this system. The shear stress measured in spring melt flows of this scale in O'Ne-eil Creek has been reported as being approximately 70 N m-1 27 which would act to break apart weakly bound aggregate structures resulting in smaller mean and maximum population sizes as observed. Interestingly, aggregate particles are small during the active spawn of 1995 and 2000, while the largest sizes were observed during the die-off of fish in 1996. This variability in size occurs despite the fact that measured water velocities, which reflect local shear stresses of approximately 6N m-1 at this site in O'Ne-eil Creek27 were not very different during these sample periods (Table 4.1). This implies that the activity of spawning fish resus-pending sediment could be having a similar shearing effect. Evidence to support this can be seen in the concentrations of suspended sediment during ambient flow conditions. In 2000, the background concentrations were between 2.47 and 3.76 mg l-1 after approximately 10,700 fish had entered the stream to spawn, while in 1995 the ambient concentration was 11.70 mg l-1 with 26,985 spawners in the stream. Undisturbed ambient suspended sediment concentrations are measured in O'Ne-eil Creek as < 1 mg l-1. The physical activity of these fish in digging their redds results in the reworking of the gravel bed, resuspension of the gravel stored fine sediments, and the downstream transport and deposition of these fines. The energy imparted in moving the gravels to release fine sediment, along with the action of being transported in the turbulent water column before settling again, would act to break apart weakly held structures, generating smaller and stronger and potentially more compact aggregates. In the ambient flow of active spawn in August 2000 only 7% of the total population of aggregate particles is loose floc structures. Alternately the largest aggregate sizes as well as the largest abundance of floc structures (35%) are observed in the salmon die-off period which exhibits low flow velocities with resuspended sediment concentrations of 7.22 mg l-1. While these same low shear conditions and higher resuspended particle concentrations are sampled in post-fish periods (September and October 2000) the particle aggregates do not reach the same large sizes as during die-off. Clearly some other environmental factor aside from shear velocities and suspended sediment concentrations is controlling the size of these aggregate structures. The changing size and density of the infiltration gravel bag particle populations also supports the influence of the physical role of spawning salmon on the aggregate size. Smallest mean population sizes at mid-spawn, which are also associated with high densities, suggest that the smaller compact particles that predominate have been collected in the infiltration bags during spawn. The larger less dense aggregates collected in the bags earlier in the season have presumably been modified by the fish spawning activity. While the particles stored in the gravel bed during salmon die-off are not statistically larger than at mid-spawn they are significantly less dense, which corresponds to the higher proportion of low density flocs observed in the water column (Tables 4.3 and 4.4). This could reflect the fact that the larger floc particles are less stable, breaking up when they enter the gravel matrix, or potentially being broken into smaller particles by the physical action of sieving through 2 mm mesh when the gravels are separated from the fines in the field.

The change in the particle composition and structure noted during fish die-off is associated with a temporal change in the organic composition of the aggregates. Petticrew and Arocena17 reported on these same gravel stored samples and stated that over the open water season the biofilms that cover gravel stored aggregates changed from weak, web-like structures at mid-spawn to a less porous, film-like covering in post-spawn. The stronger more extensive biofilm was associated with large aggregates while the weaker web structure existed when the aggregates were being exposed to repetitive reworking of the gravel bed (e.g., resuspension) during mid-spawn.

The sediment moving in the spring melt has the lowest mean particle size as well as the lowest fractal values, as determined from filtered samples (Figure 4.5). The sediment moving in the melt is small, dense, and rounded. In an evaluation of the filter fractals there is a significant decrease in D over the three day high flow sampling period in 1997. The suspended sediment becomes more rounded with time, indicating either a change of source21 or a modification of the particle shape with changing energy conditions. In Figure 4.5 there does not appear to be any significant differences between sediments resuspended from the gravel artificially over the season and the ambient suspended sediment from the active spawn of 2000. This would indicate that the sediments are from the same source, which we know to be the case, and experiencing similar energy conditions. This then would support the assumption that the energy imparted to the surface gravels to resuspend the stored fine sediments is similar to the work perpetrated by the fish. To corroborate this effect of physical resuspension note the timing of significant differences observed in the fractal dimensions and structure of the infiltration gravel bag sediments over the 2001 season (Figure 4.6, Table 4.4). Particles are growing more amorphous from pre-spawn through early spawn but then decrease in D values becoming rounder at the same time as becoming smaller and denser at mid-spawn when the majority of the physical reworking of the bed has been completed. As this reworking abates and die-off occurs, delivering carcass breakdown products to the stream the particle roundness decreases, density decreases, and particle size starts to increase again.

The ambient suspended sediments analyzed from the filtered populations show a significant increase in D following active spawn (Figure 4.5). This indicates that these particles are more amorphous than those of active spawn but also less rounded than the gravel stored sediment resuspended on the same dates. Intuitively this would make sense as there is less energy in the low flow water velocities which could break up the larger more amorphous particles.

In the post-spawn period the artificially resuspended (gravel stored) sediment has similar fractal values as the material comprising the ambient suspended sediment during the active spawn (August 2000) (Figure 4.5). This indicates that the material resuspended by spawning fish which remains in the water column is similar in shape to the gravel stored sediment later in the season. This is corroborated by looking at the fractal results of the infiltration gravel bags (Table 4.4, Figure 4.6). The mid-spawn and post-fish fractals indicate they are the roundest populations of particles over the season, but as well they are the densest populations. Petticrew and Arocena17 presented scanning electron microscope evidence indicating that these mid-spawn aggregates were held together with a weak-looking web of biofilm while pre- and post-spawn periods exhibited a more extensive coating of biofilm. The strength of these biofilms may relate to the physical action these particles are exposed to (low flows, active spawning) and could also play a role in regulating their size.

4.4.1 Fractal Concerns

Note that the fractal results for the filtered populations (Figure 4.5) are not the same as is found in the fractal analysis of the settling chamber populations shown in Figure 4.4. In this latter data set there are no significant differences within the subpopulations by either source (ambient versus resuspended) or date but there is a statistically smaller D value for the total population of the ambient post-fish (October 2000) suspended sediment as compared to the ambient active spawn samples. This indication that the ambient suspended sediment is becoming more rounded with time is opposite to the results of the filtered samples. A comparison of fractal values for samples from the same date and source material indicates that the filtered particle populations are always lower than the settling chamber populations. This would be a result of both the larger number of particles that are counted by filters as well as the lack of inclusion of the larger particles (>400 ^m) which have the higher proportion of amorphous, less rounded flocs. Inclusion of these bigger particles, which also exhibit larger variability, increases the fractal value of the settling chamber populations as seen in Figure 4.4. Another inconsistency is that the fractal values of the gravel stored sediment (Figure 4.6) are consistently much lower than that of the resuspended sediment shown in Figure 4.4, which is meant to represent the gravel stored sediment. This could be a function of the artificially disturbed samples being mixed with low concentrations of ambient sediment or the fact that the structure of the aggregates changes with depth in the gravels. The infiltration gravel bags collect material stored to a depth of approximately 25 cm while the artificial resuspension mobilizes only the top 5 or 6 cm. In Figure 4.3 it is clear that the majority of large floc particles are associated with the artificially resuspended gravel stored samples from August and October 2000, as opposed to being abundant in the ambient suspended sediment. Aside from periods of high flow, or just following scouring events, a layer of fine grained loosely aggregated sediment is often observed to be coating the surface of the channel.28 These fine grained aggregates are more floc like than compact and are easily resuspended and moved downstream with increases in entrainment flows. These aggregates would be collected in both the artificially resuspended and the infiltration bag sediment sampling but would represent a larger portion of the population in the artificially resuspended sample as it disturbs a larger surface area and smaller depth of gravels. These surface fines would be less abundant in the gravel infiltration bag samples as they are deeper (25 cm) and have a specific surface area sampled (314 cm2). This bias of surface sampling would then result in higher fractal D values for the former (i.e., more amorphous) and lower D values, or more rounded particles in the gravel bags as observed in this comparison.

As the fractal values presented here reflect the measurement of potentially different populations (e.g., sediment populations with different upper and lower size limits as well as populations from different depths of gravel storage) care should be taken to compare results of fractal analyses between methods. Changing the upper and lower limits of the population analyzed in a fractal analysis has been found to have a significant effect on the results. For the settling chamber samples the lower limit was defined by the resolution of the image analysis technique (diameter approx. 150 /m) and the upper limit was not restricted. The regressions were always strong (r2 > 0.90) and significant, but the subpopulation sizes were not always very large (n = 5 to 182). A test was done on the largest settling chamber data set (total population, Oct 2000 resuspended, n = 315), where the sample was altered to include only the aggregate population <600, <500, <400, and <300 /m in order to determine the effect of the size limits on the fractal D values. While the D's are not significantly different as the upper size limit is reduced and the sample size becomes smaller, the 95% confidence limits increase resulting in reduced ability to distinguish statistical differences. This observation is important if one plans to use fractals for identifying source sediments or for implying processes affecting sediment structure. The results of the filter population fractals presented in Figure 4.5 were analyzed using the same method as de Boer and Stone29 who identified source differences in suspended sediment during a spring melt period. The lower limits and presumably the upper limits (as they are defined by the sampling technique) are similar to de Boer's which are detailed in his 1997 paper.21 Using this method between 1,500 and 15,000 particles can easily be counted ensuring a representative population size. In viewing this lower end of the aggregate population (7 to 400 / m) we see significant differences over a 3-day period in spring melt and a difference in the ambient suspended sediment over the season. As the data for Figure 4.4 are comprised of subpopulations of quite variable, and in some cases small sizes, these data would be considered problematic. A better method of evaluating the fractal dimensions of these samples would be to measure a large number of particles from the general population photographed in the settling chamber as opposed to using just particles that have been tracked for settling velocities. This approach was used to determine fractal values for the infiltration bag fine sediments from 2001. An excess of 1,000 particles were sized to determine the D values of the gravel stored sediment over the season. The fractal D values indicate that on all dates the particles are very rounded with the only significant differences being that the mid-spawn is rounder than the sample before it from early spawn and that the final post-fish sample is roughly the same roundness as mid-spawn with a significantly smaller D than in the period of fish die-off preceding it.

0 0

Post a comment