Metal mining technical guidance: revised guidance for sample sorting, chapter 3

Basic Laboratory Subsampling Methods

Once it has been established that subsampling will result in a significant timesaving in the processing of the samples with large numbers of invertebrates (usually greater than 500-600), an appropriate method can be selected. Various methods of laboratory subsampling are available and are outlined in Wrona et al. (1982), Sebastien et al. (1988), Marchant (1989), Plafkin et al. (1989), Klemm et al. (1990), Canton (1991) and Mason (1991). Regardless of the method selected, the recommended general protocols and QA/QC documentation (Section 2.1 and 2.2) should be followed as these aspects are critical to allow national comparisons of EEM studies.

There are two basic approaches for subsampling benthic invertebrate samples. The fixed count methods and fixed fraction methods (Barbour and Gerritsen 1996). The fixed count method is commonly used for enumerating plankton samples (Charles et al. 1991) and was adapted for sorting benthic samples by Hilsenoff (1987, 1988) and refined and modified for use with the US EPA Rapid Bioassessment Protocols (RBP) by Plafkin et al. (1989). Over the development of the method the total number of organisms to be counted has been debated (e.g., 100, 200, 300) (Plafkin et al. 1989, Caton 1991, Hannaford and Resh 1995, Growns et al. 1997, Doberstein et al. 2000, Carter and Resh 2001) but the basic premise is that a predetermined number of organisms are sorted and enumerated from any given sample. The original intent for using the fixed count methods with the RBP was to facilitate a rapid return of data for analysis and interpretation, allowing the identification of sites where potential further study would be indicated (Plafkin et al. 1989, Hannaford and Resh 1995). For these objectives, the fixed count methods are comparable between sites because a consisted unit of effort is applied. However, as Cortemanch (1996) points out, a fixed effort should not be confused with sample size. It is unknown whether the 100 organisms counted came from a sample of 500 or 5000 animals. Fixed count methods would be appropriate for proportional metrics (e.g., % EPT, Cortemanch 1996). However, as none of the benthic effect endpoints are proportional they would not be reliably estimated with fixed count subsampling methods. Therefore, although there are many literature reviews discussing the merits and applicability of fixed count methods, particularly for rapid bioassessment objectives, these methods would not meet the objectives of the EEM program where quantitative methods are required.

There are several fixed fraction subsampling methods, with the simple objective of dividing up a large sample, into several portions, each of which are a random representation of the entire sample. These smaller portions are processed more efficiently and cost-effectively while producing reliable estimates of the total number and types of organisms in the sample. The three most common methods divide up the samples based on either

1. area of the sample
2. volume of the sample
3. weight of the sample

Which method is most appropriate depends on the type and amount of material in the sample. These methods are briefly reviewed below.

Area-based sieve splitting

Although there are several variations of area-based subsampling methods ranging from grided trays or frames to elaborate sieve splitting devices (Sodergren 1974, Hickley 1975, Klattenburg 1975, Rosillon 1987, Marchant 1989, Klemm 1990, Meyer 1990, Caton 1991, Mason 1991), the most commonly used in the EEM program have been based on the sieve diameter splitting method of Cuffney et al. (1993). The process is based on evenly distributing material on a tray or sieve and randomly separating a fraction (½, ¼ or 1/8) by planar area. Specifically, the sample is placed on a sieve marked with six equally spaced diameters. The sieve is immersed in water and agitated to produce an even distribution of material after which the water is drained. A dice is rolled to randomly select which of the six diameter markings to use for splitting the sample and then a ruler is used to divide the sieve into halves by aligning it with the marks and pressing the ruler edge into the sample. A scraper and wash bottle are used to help separate the sample into halves and remove the selected portions. Once the sample has been split, a coin toss is used to randomly select which half of the sample to sort. DO NOT discard the other half of the sample as it may be needed for QA/QC determination, or if the one half of the sample does not contain enough animals to fulfill the minimum 300 animal criteria it will need to be sorted. All unsorted and sorted fractions are to be retained until taxonomy and sorting efficiency are confirmed and the data are reviewed by the regional EEM coordinator. Very large samples may need to be split repeatedly to obtain suitable subsample fractions, but for each split, the corresponding half will need to be identified and saved for QA/QC purposes until all data is verified and accepted. An appropriate fraction(s) is sorted and the total number of organisms in the original sample is calculated by the inverse of the fraction (Section 3.4).

A new variation of this area-based subsampling method with a modified apparatus was recently developed by researchers at National Water Research Institute (NWRI) (Environment Canada 2002). Somewhat similar to the apparatus' and process described by Meyer (1990) and Mason (1991), it is capable of handling large samples typically collected in EEM programs and quantitatively divides samples into 4 quarters. The apparatus consists of an agitation rod and a 60-cm length of PVC pipe (8" dia.) with an attached (removable) sieve, pre-divided into quarters by an inset frame. The tube and attached sieve are immersed in standard 20L- bucket water to a depth of 40-50 cm. The sample is poured into the tube and agitated with the rod for a minimum of 30 seconds. The rod is removed, ensuring that no organisms adhere, and the entire apparatus is quickly lifted straight up from the water. As the water drains from the sieve, debris and invertebrates settle evenly over the sieve. The sieve is then detached from the tube and any material, such as filamentous algae, that straddles 2 quartiles are placed in the quarter to which the majority of the material resides. A prepared PVC template is placed over the sieve and one quarter of the sample is easily removed using either a spatula or wash bottle alone. Further quarters can similarly be removed from the sieve and stored appropriately for sorting or archiving. Tests for random splitting of samples have produced satisfactory results.

Volume-based sample splitting

The Imhoff cone method detailed by Wrona et al. (1982), randomly distributes fine organic material and invertebrates in a l liter volume. The recommended protocol includes initially fractionating the sample into course and fine, producing a homogenous fine fraction with large numbers of organisms. This fraction is then placed in the cone apparatus and the volume brought to the l litre mark with water. A regulated air supply with an air stone siliconed into the base of the cone is used to gently agitate the sample for 2-5 min ensuring a random distribution of material. Note that inorganic material should have been removed previously in the elutriation step as it will congregate on the bottom of the cone reducing the effectiveness of the air stone. Approximately 50% of the sample can be removed from the cone with 55ml test tubes while continuously agitating the sample. The test-tube subsamples are sorted sequentially until the minimum 300 animals are found. As with the other subsampling methods, if this minimum number is reached part way through a subsample sort, the subsample must be sorted in it's entirety so that a known fraction is sorted. The portion of the sample remaining in the cone is retained along with contents of any unsorted test tubes as the unsorted fraction for potential further sorting for subsampling accuracy calculations. This method of subsampling is simple and effective for a wide range of benthic sample types. However, the technique may not be compatible with samples containing large amounts of filamentous algae.

Weight-based sample splitting

This method as documented by Sebastien et al. (1988) is a simple, reliable alternative for subsampling benthic invertebrate samples, especially those where macroinvertebrates are entangled in filamentous algae or bryophytes. The technique involves thoroughly mixing the sample in a 2-litre beaker to ensure random distribution and subsequently pouring and distributing the material evenly onto a pre-weighed sieve. The moist sample (and sieve) is weighed to the nearest 0.1g to obtain a total weight for the sample. Material is removed from the sieve at random in appropriate fractions (e.g., approximately 10% or 25% of material) and subsample weight recorded to determine the actual subsample fraction. Once the sample has been split, the first fraction to sort is selected randomly. Consecutive subsamples are sorted until the minimum of 300 animals is recovered and the particular fraction is sorted in it's entirety. The other fractions are retained for potential QA/QC determination.