Sample Representivity

The problems of obtaining representative samples do not end with the choice of substratum. Many other factors can also affect the diatom assemblages. With the interest of water quality estimation in mind the observed diatom assemblages can be adversely affected by accumulation of dead cells, grazing, seasonal and temporal variation, flood events, current velocity and many other changes in the physical environment.

1.4.3.1 The Accumulation of Dead Cells

By their very nature diatoms are persistent within a system after their death. The siliceous valves of dead diatoms can remain entangled in the mucilage complex of the live community (Owen et al. 1979) or valves washed in from other, potentially very different, areas may become trapped (Pryfogle & Lowe 1979, Battarbee & Flower 1984). The inclusion of these dead cells in studies often goes unnoticed due to the use of oxidative preparatory techniques which remove all the organic matter (Battarbee 1986). The consequences of including the dead component could be very misleading. When using the diatom community to assess water quality, one is only interested in the living component at a given time. Thus dead cells from potentially different conditions, could bias a water quality estimation. Pryfogle and Lowe (1979) observed epilithic communities to have between 2-77% dead cells, with no apparent pattern in the variability. When compared to the live component alone the inclusion of dead cells led to an over-estimation of species diversity. Similarly the results of Owen et a l (1979) showed approximately the same range of dead cells on glass slides. All the slides had equal exposure times and almost all of them had over 10% dead cells. This is contrary to the results of Patrick et al. (1954) who reported few dead cells on glass slides. It is unclear how these differences arise although a shorter exposure time was used in the latter study, therefore less dead cells would have accumulated.

The error is further increased if the dead cells are allochthonous, i.e. from an outside source. Lotie systems offer many secondary sources from which diatoms could originate, e.g. soil, impoundments, tributaries and even fossil diatoms from the erosion of bed-rock. Although in-washed diatoms have been investigated as a source of error in palaeolimnological studies (e.g. Battarbee & Flower 1984), there is very little information regarding their impact on the interpretation of modem diatom communities.

The assessment of live cells within a sample can be made quite easily by the use of staining techniques. The stain acts on the protoplast of living cells only, thus all dead cells will remain colourless (Owen et al. 1979). The stained samples can then be conventionally mounted in high index mountants for identification and counting. However, such methods can heavily stain some organelles and therefore compound the taxonomic problems. With

the preparation of conventionally cleaned slides for reference, however, this problem should be possible to overcome (Owen et a l 1979). The publication of live-diatom keys, which rely on features of the living cell, rather than the silica cell walls, also aid in identification of living material (e.g. Cox 1996).

1.4.3.2 The Effects of Grazing

Another impact on the diatom assemblages, which is less easily quantified, is the effect of grazing by invertebrates and other organisms. Diatoms, as one of the most abundant aquatic primary producers, form a major part of the diet of a wide range of grazers from amoebae to young fish (Round 1981). It would therefore be expected that an inverse relationship should exist between the numbers of diatoms and grazing intensity. This was demonstrated by Douglas (1958) with respect to the effect of the caddis larvae Agapetus fuscipes on A. minutissima populations. Similarly the total removal of grazers from a

system has been show to facilitate algal blooms. Eichenberger and Schlatter (1978) added insecticides to outdoor channels and observed large blooms of filamentous algae followed by a cyanobacterial bloom. These blooms were ended with the réintroduction of grazers. It can therefore be assumed that algal numbers are, at least in part, controlled by grazing in most aquatic systems (Allan 1995).

It is not just simply loss from the diatom community that can cause problems in sampling. Not only do grazers reduce total standing crop of algae but they can also alter the floristic composition by selective grazing (Allan 1995). These effects are difficult to demonstrate but Hill and Knight (1988) found a seven-fold over-representation of one diatom species in the gut contents of the caddis Neophylax. Heavy grazing by this species could result in a biased sample being collected which could have serious implications for a diatom-based water quality assessment.

Even without selective grazing the removal of diatoms allows for recolonisation to take place thus disturbing community structure. With other microhabitat variations, e.g. light and nutrients, the response of different species will vary, further upsetting the community stability due to shifts in competitive forces (Burton et a l 1994). In contrast, the tube building activities of some chironomids have been observed to increase diatom biovolume

(Pringle 1985). The sand grains used to build the tubes were found to support large numbers of diatoms with an estimated 12-fold increase in diatom biovolume compared to adjacent substrata without tubes. There was also found to be an increase in species numbers due to greater habitat diversity.

When water quality is the primary issue under investigation, the interactions between grazers and diatoms become extremely complex. Grazing pressure is very variable due to the environmental preferences of the grazers. This is further complicated by the sensitivity of many invertebrates to water quality (Wright et al. 1989). Thus in areas of very poor water quality, grazing is likely to be negligible, whereas, in areas of high quality, where the water is well oxygenated, grazing could have a considerable affect on the observed diatom assemblage.

1.4.3.3 Seasonality

The variation in both diatom productivity and species composition over seasonal cycles has been well documented (Douglas 1958, Moore 1976, Marker & Casey 1982, 1983, Cox 1990a, 1990b). Marker and Casey (1982), working with artificial streams, reported diatom cell numbers of 4.2 x 10^ cells mm'^ during the month of May with numbers falling off to less than 20,000 cells mm'^ for the rest of the year. In natural systems this spring bloom is also often seen but with a less marked decline in the summer months (Moore 1976). There are of course many factors involved in the seasonality of a stream, including flow patterns, nutrient availability, water temperature, day length and light intensity. The observed seasonality can therefore be seen to be extremely variable between different systems {cf.

Douglas 1958 and Marker & Casey 1982). It is usual to see shifts in species composition as well as productivity. Cox (1990a) showed seasonal shifts in small Navicula spp. from epilithic stream samples. Similarly some species have been reported as being typical of certain times of the year e.g. M. circulare is known as a vernal species; only occurring at other times of year under low temperature conditions (Cox 1993).

Seasonal changes in the diatom flora have been well studied from an ecological viewpoint and would appear to be very variable both within and between different river systems (Douglas 1958, Low 1972). Despite this there seems to be little available information on

the effect of such variation on a diatom-based monitoring programme. Using two diatom indices over the period of one year, Kelly et a l (1995) found upland sites gave more consistent results than lowland sites, but the variation was not however considered to be significant. For future monitoring studies it would appear necessary to keep sampling dates and conditions consistent where possible and also to gain more information on the seasonal changes in the diatom community to establish how these seasonal shifts in species composition might affect a diatom-based assessment of trophic status.

1.4.3.4 The Physical Environment

The diatom assemblage at a site is also determined by the physical conditions. Only species which can firmly attach such as Achnanthes spp. and Cocconeis spp. are found at high current velocities, whereas motile and loosely attached taxa are more common at low current velocities, e.g. Melosira varians, Navicula spp., Gomphonema spp. and Cymbella

spp. (Patrick 1977). Such morphological adaptations are a necessary response to the hydrological environment (Rott 1991). Within a site, changes in the current velocity have been found to explain the majority of observed species variation (Passy 2001). For the purposes of water quality assessment this means that two sites of identical water quality but very different current velocities may have very different diatom species present. H omer et at. (1990) observed different diatom taxa at the same current velocities when phosphoms concentrations were changed. It is possible, therefore, that providing biological information is gained from enough sites with different current velocities, over a range of phosphorus concentrations, the importance of velocity will be minimised. Effectively there can be different groups of indicator species to denote the same water quality range for both high and low current velocity sites.