Spatial Variation
INTRODUCTION
In the last issue of The Water Line we discussed temporal variation in lake water quality and how volunteers can reduce the influence of these variations on LMVP’s data. In this issue we will review spatial variation, and explain how the LMVP deals with differences in water quality across a lake surface.
LARGE SCALE VARIATION
When flipping through the 2008 LMVP data report you will notice that not all lakes have the same water quality characteristics. Differences among lakes can be attributed to variations of lake shape and depth (morphology), watershed size and slope (hydrology), and land use within the watershed. Large, deep lakes in unaltered watersheds have the lowest concentrations of nutrients, algae and sediments, as well as the clearest water. Shallow lakes in rich soils with large watersheds dominated by agriculture have high concentrations of nutrients, algae and sediments, and very murky (turbid) water.
Figure 1. Top-down view of longitudinal variation in reservoir water quality
LONGITUDINAL VARIATION
Not only do water quality conditions differ depending on the region of the state, conditions can vary within a single lake. Sites located in tributary arms or up-lake (A in the illustrations) tend to have higher nutrient and suspended sediment levels than sites located in the main lake channel or near the dam (B in the illustrations). The reason for this variation is that the tributary/up-lake sites tend to be located closer to pollution sources, with water quality reflecting these inputs. As water moves down-lake, processes such as dilution and sedimentation lead to a decrease in nutrient and suspended sediment concentrations.
To deal with these variations, the LMVP simply sets-up multiple sites on the larger lakes in the program. We try to space the sites out wisely, monitoring enough sites to describe water quality throughout the lake while avoiding monitoring redundant sites. The differences in water quality observed at either end of a small lake are negligible compared to the differences observed on a large lake. Using the additional equipment to monitor another lake is better than collecting samples from sites on the same lake that replicate information.
Figure 2. Cross-section of longitudinal variation in reservoir water quality
SMALL SCALE VARIATION
When we start looking at variation within a smaller area we find that differences do occur, but they tend to be smaller than differences observed among tributary arms or from opposite ends of the lake. During the first LMVP sample season volunteers collected three samples from along the dam instead of the one sample that is currently taken. Results from that first year indicated that the differences among the three sites tended to be quite small (around 15%), so the program moved to just one site at the dam. For our purposes, it is better to collect data from more lakes than collect redundant data on a single lake.
While the difference is likely negligible, chlorophyll concentrations (our measure of the ‘standing crop’ of algae) might even vary from one side of the boat to the other. Causes for variation at this scale may include patchy algal populations, activity by the grazer community (e.g. zooplankton, zebra mussels, etc) or wind-driven water circulation (see Langmiur Circulation: Windrows and Scumlines, next page).
To address this scale of spatial variation, LMVP volunteers are asked to ‘composite’ sample. For our purposes, that means they grab three separate water samples from their lake site (from around the boat) and combine those samples in a bucket before filling their sample bottle. This effectively ‘averages’ the water in the immediate area.
VERY SMALL SCALE VARIATION
Variation associated with space is present even within the sample bottle. In the laboratory we take multiple subsamples (10 mL in size) for nutrient analyses and pipette them into test tubes. Inevitably there are differences among these tubes. Usually the difference among the individual tubes is less than 5%, meaning we feel quite comfortable with the average value that we generate. When the difference among the tubes is greater than 5%, we will repeat the analysis until we have an average value that we can feel confident in. While some differences observed among the test tubes can probably be attributed to human error in the laboratory, variation caused by particles within the sample bottle is a much larger issue. A single daphnia (a.k.a. water flea, a genus of zooplankton) may contain up to 0.2 µg of phosphorus and 1.5 µg of nitrogen. Even though these estimates are at the high end, this is still a very small quantity of nutrients. However, when a single daphnia is put into a 10 mL tube (1/100 of a liter), it could alter the final estimation of lake nutrient concentrations by as much as 20 µg/L for phosphorus or 150 µg/L for nitrogen!
Like algae, zooplankton are part of the total lake nutrients and we don’t want to exclude them from our samples. However, the particulate nature of zooplankton, algae and sediments highlight the necessity of thoroughly shaking the sample bottle multiple times during processing.
A chlorophyll filter with daphnia (a type of zooplankton) on it. A single daphnia in a 10mL test tube can increase the measured phosphorus concentration by as much as 20 micrograms per liter!
Shake those bottles!
ENSURING QUALITY DATA
The LMVP office at the University of Missouri addresses variability mathematically. In our data report we display seasonal values as geometric means, a way of describing the central tendency of the data while minimizing the influence of extreme values. This mathematical technique is commonly used to remove much of the ‘noise’ associated with water quality data’s inherent variability.
A few simple steps by volunteers can reduce the confounding effects of temporal (see last newsletter) variation and spatial variation. Sampling regularly is the best approach for addressing temporal variation. Sampling in the same spot and compositing the sample helps address spatial variation.