Predicting levels of UVattenuating constituents

A range of bio-optical models (described in Section 3.3) can be used to estimate UV attenuation in natural waters when the concentrations and optical properties of algal pigment and DOC are known. How can these concentrations and optical properties be predicted from regional-scale features of the environment? Predicting typical concentrations and optical properties for either algal pigments or DOC in an aquatic system will require knowledge about sources and sinks, mechanisms for changing optical properties, and how each of these responds to the environment, a subject that will be only briefly reviewed here. Only passing mention will be made here of the role of inorganic particles, which can be extremely important as UV attenuators in some aquatic systems [6,10,49]. Such systems are shallow or subject to erosion: streams and rivers with disturbed or non-vegetated watersheds, and shallow lakes or coastal systems where the high energy keeps particles suspended in the water column. One special case is the precipitation of carbonates that imparts a whitish color (observed in several lakes by Laurion et al. [62]). Another special case is the melt water from glaciers, where interactions with the underlying rock surface create small particles that give the water a milky appearance.

Phytoplankton abundance responds to nutrients, light, and grazing pressure. The primary autochthonous source of CDOM is often assumed to be photosyn-thetic organisms, but heterotrophic bacteria may play an important role by processing the relatively UV-transparent photosynthate and releasing modified compounds that absorb UVR more strongly [91]. Spatial and temporal linkage between CDOM and the microbial community appears complex and difficult to predict at present, in part because the microbial community also attenuates UVR. Predictions about microbial UV attenuation are rather difficult at present because there are so few measurements and reliability of current techniques (e.g. QFT) is still being debated. The adaptive physiological and evolutionary responses of phytoplankton to UVR are likely to differ from their adjustments of photosynthetic pigments to light and depth. Here again there is too little information to make predictions except that one should not be surprised to find UVR screening pigments in phytoplankton exposed to high UVR (perhaps when 320 nm irradiance exceeds 1 % of surface levels?).

The balance between sources and sinks for DOC will determine its concentration in natural waters and thus establish a major factor in UV attenuation. The levels and optical properties of CDOM in aquatic systems will depend in part on whether the source is within the water column (autochthonous) or coming from elsewhere (allochthonous). Allochthonous sources include watershed runoff (e.g. from soil or wetlands, especially water-saturated soils but also from man-modified surfaces which may be enriched with petroleum hydrocarbons), and wastewater discharge. In estuaries the large CDOM load typically carried by rivers is both diluted (and to a lesser extent, precipitated) as it mixes with brackish water. Allochthonous sources are affected by precipitation, evaporation, soil hydraulic residence time, and temperature, which influence production of CDOM in saturated soils and transfer, dilution, and concentration of this material in receiving waters. In a comparative study of 337 lakes from northern USA and Canada, Rasmussen et al. [102] found a strong positive correlation between CDOM and the ratio of drainage area to lake area, and a negative correlation between CDOM and average slope of the drainage landscape. CDOM in coastal regions is strongly correlated with river discharge and inversely correlated with salinity [74]. CDOM in two Australian reservoirs was highly correlated with season, increasing during rainy periods and declining during dry, sunny periods in response to microbial and photochemical degradation [103]. The two reservoirs differed in average level of CDOM, with 97% of the variability in average annual CDOM accounted for by difference in hydraulic residence time. Similarly, Arts et al. [34] suggested that long residence time in prairie lakes (Canada) causes the DOC to become more UV transparent. Climate can affect both the watershed yield of DOC to lakes and coastal oceans and hydraulic residence time by influencing snowmelt, precipitation, evaporation, and watershed soil properties [13]. A period of warm and dry years resulted in a decline in DOC of Canadian shield lakes [104]. Temporal variation in DOC and CDOM absorption in lakes of N. Michigan (USA) was correlated with ice-out date and spring-summer precipitation [119].

At regional and larger scales several end members and other patterns of UV attenuation have emerged. The clearest lakes and ocean regions are those most isolated from terrestrial sources of CDOM and nutrients. A barren watershed containing little or no vegetation surrounds Crater Lake and Lake Vanda, and the permanently unmixed bottom waters likely serve as a trap for nutrients present in sinking biomass. The open ocean may appear to be equally isolated, but the connection with a deep-circulating source of CDOM and nutrients may provide a low baseline level of UV attenuating substances that increases in the presence of deep mixing or upwelling. High elevation lakes tend to have lower levels of DOC [20] along with less vegetation in the watershed [62]. Mountain lakes of the Alps and Pyrenees [62] show an expected decrease in watershed vegetation with elevation, but when meadows and forest are compared at the same elevation the meadow-dominated watersheds have higher DOC levels. Lakes in contact with wetlands tend to have high levels of DOC (reviewed in [24]),but even higher levels occur in arid closed basin lakes [34].

Variations in DOC quality (DOC - specific attenuation and absorption) warrant further discussion. These variations can be divided into two related categories: changes in specific attenuation correlated with DOC concentrations, and changes in specific attenuation correlated with lake and watershed characteristics. The immediate basis for the power relationships between Kd32o and [DOC] (Table 3) comes directly from the power relationship between ad32o and [DOC] (Table 2) for the data of Morris et al. [60]. But why should DOC quality change together with DOC concentration over the scale of lakes in this study? What determines the exponent in the power model relating specific absorption and DOC concentration?

Land cover in lake watersheds, climate, and hydraulic residence time influence the spatial pattern of DOC quality. Data from Laurion et al. [62] show that lakes with high-elevation watersheds had lower DOC concentrations than those at lower elevations in the same region. Lakes at similar elevations had lower DOC-specific attenuation if the watershed was rocky compared to watersheds with meadow or forested land cover. These data also showed that, for lake watersheds with similar land cover, DOC-specific attenuation was elevated when catchment area was more than 50 times larger than lake area, a characteristic associated with short residence times. For all the lake surveys in Table 2 and the majority in Table 3 the power equation exponent is greater than 1. These cover mountain lakes and wet areas at mid to high latitudes. The exceptions, where DOC-specific absorption varies inversely with DOC concentration, include ponds and lakes in the arid prairies of central Canada.

Temporal variation in DOC quality provides clues to explain the spatial variation. The decline of DOC-specific absorption in stratified surface waters [22, 23] and lake versus feeder stream [90] are attributed to cumulative photo-bleaching of the DOC pool. While photobleached DOC is in some cases subject to enhanced microbial utilization [105], the old and previously bleached DOC of saline prairie lakes is metabolized very slowly [90]. It is self-evident that photobleached DOC will not be dominant in the DOC pool while the rate of influx of non-bleached DOC is high. If hydraulic residence time is short then even low rates of DOC influx or production will be adequate to prevent accumulation of photobleached DOC.

3.4.1 A conceptual model for UV-DOC relationships

In most comparative studies to date the concentration and optical qualities of

DOC are the best predictors of UV attenuation. A conceptual model is proposed here to predict spatial and temporal patterns of UVR attenuation, including the relationship of KdUV to [DOC], and acdom to [DOC], for all non-turbid aquatic systems. The DOC concentration will tend to be high in aquatic systems where influx or production is high relative to the rate of dilution or flushing by direct precipitation, snowmelt, or other source of low-DOC water. The [DOC] will also be high when a system with long hydraulic residence time experiences evaporation rates that exceed the rates of DOC degradation (microbial and photochemical). DOC concentration will be low in aquatic systems where DOC influx or production is low relative to rates of degradation or flushing with low-DOC water. DOC optical quality reflects the biotic source and the extent of photobleaching. For systems with short hydraulic residence times the DOC-specific absorption will be higher for sources from higher plant and lower for sources from the microbial community. Systems with long hydraulic residence times will tend to have low DOC-specific absorption when the long term rate of influx or production of DOC is slow compared to the rate of photobleaching by sunlight. DOC-specific absorption will become low as well in water that is seasonally isolated by density differences in a thin surface layer exposed to prolonged sunlight. Combining the processes influencing DOC concentration and specific absorption results in two hypotheses explaining the power relationships between Kd and [DOC] in equation (15). When DOC is not concentrated by water loss, the power exponent tends to exceed 1.0 because the systems with low [DOC] have experienced more cumulative photobleaching. In arid or frozen regions where cumulative water loss occurs in systems with long hydraulic residence time, the power exponent tends to be less than 1.0 because systems with high [DOC] have experienced more cumulative photobleaching.

0 0

Post a comment