Enumeration

Different conventions have been evolved for the enumeration of diatoms, using either valves or frustules as the basic unit, or not distinguishing between valves and frustules. The effect that such conventions have on the final results has not been evaluated, but is likely to be small. However, it is important that the convention used be specified in advance. In the case of small diatoms, such as some Achnanthidium and Naviculoid species, it may not be possible on all occasions to distinguish with certainty between intact frustules and isolated valves. CEN, (2003) and Prygiel et al., (2002) recommended that the required number of individuals be counted, without any distinction between valves and frustules. The aim of counting diatom units is to produce semi-quantitative data for ecological conclusions. For this, it is important to know how many valves to count to get a reliable estimation of the relative species composition at a specific sampling site. The total number of valves to be counted for each sample varies according to the purpose of the analysis and according to the need to produce statistically good results. The statistical precision of percentage counting depends on the frequency of the taxon in the sample count in relation to the size of the sample count (Battarbee, 1986). In a study from South Africa by Schoeman (Schoeman, 1973) a series of experimental counts in which 200, 300, 400, 500 and 800 valves per sample were counted and their relative abundance calculated. When only 200 valves were counted, compared to when 800 valves were counted, the percentage differences of the relative abundances of individual species were often as high as 6–7%. However, the results obtained from counting 400 as opposed to 800 valves differed by only 1–2% which would suggest that counting 400 valves was satisfactory for the calculation of relative abundance of diatom species. Similarly, Battarbee (1986) demonstrated that there were marked differences in the percentages between counts of 100 and 200 valves, while there was little difference between counts of 400 and 500. For this reason he recommended that a count of 300 to 600 may be used for purposes of routine analysis. This range is supported by Prygiel et al., (2002) who, in an inter-comparison exercise, found that diatom index scores were not affected at counts of 300 and above. Hence, it is recommended that, for diatom community analysis 400 diatom valves should be counted in each sample.

Suggested rules for counting diatoms, according to CEN (2004), are summarised below:

• Counts of diatom valves on slides should be made using a compound light microscope equipped with incident light and capable of 1000 x magnification (100 x oil immersion objective in combination with a 10x eyepiece). Phase contrast optics or differential interference contrast optics (DIC) may provide better contrast.
• The eyepiece graticule or other measuring equipment must be calibrated against a stage micrometer prior to the analysis to allow for measurement of dimensions and taxonomic features.
• Either the field of view or the grid of a graticule is used as the area defining the limits of the count. All diatoms visible in the field of view (or within the grid of a graticule) are identified and counted before moving along either a horizontal or vertical traverse to the next field, or selecting a new field of view at random.
• In cases, when a diatom lies only partially inside a defined counting area, include taxa that are only partially visible at the upper, but not the lower, margin (in the case of vertical traverses) or the left, but not the right, margin (in the case of horizontal traverses).
• It is important that each subsequent traverse does not overlap with the previous one and avoiding multiple count of a single specimen. The distance that the stage is moved on each occasion must also account for any diatoms only partially visible in the field of view.
• If sample analysis is unlikely to be completed in a single session, then it is useful to record the position of each traverse to avoid overlap of traverses/counts in the subsequent session.
• Each individual specimen encountered is counted as a single unit, with no differentiation between a valve and frustules (Prygiel et al., (2002). Girdle bands (copulae) should not be enumerated as being representative of diatom taxa.
•  Occasional filaments should be recorded as the corresponding number of diatom units. If a large number of diatom units are found in filaments, a new preparation technique using a more aggressive mix of oxidizing agents is required.
• In order to eliminate the risk of including separate fragments of broken valves or frustules, valves should be counted only if approximately three quarters are present, or broken valves may be excluded altogether. The presence of many small fragments of diatoms may indicate that dead diatoms are being washed in from upstream sites.
• If many valves are obscured, then new slides should be prepared using more diluted suspensions. A diatom may not be identifiable for a number of reasons, including the presentation of a girdle view, the presence of overlying material obscuring the view.
• Some taxa are identifiable from girdle (side) views, either because the girdle view is particularly characteristic (e.g. Rhoicosphenia curvata) or because the girdle view can be assigned with confidence to a particular taxon by ‘matching’ it with corresponding valve views of taxa found in the sample. However, in cases of doubt, the analyst should record the girdle views at the lowest level to which they can be assigned with confidence (e.g. ‘unidentified Gomphonema sp.’, ‘unidentified pennate girdle view’).
• As most diatom indices presume that all taxa in a sample are identified, it is recommended that not more than five per cent of the total count should comprise unidentifiable individuals. If a diatom unit cannot be identified for any reason, photographs, digital images or detailed drawings should be made. Notes should also be taken of the shape and dimensions of the diatom unit, striae density and arrangement (at the centre and poles), shape and size of the central area, number and position of punctae and arrangement of raphe endings.