METHODS OF SAMPLING THE BENTHOS
249
intertidal bivalve niollusc Tellinn tenuis (Fig. 41) in which spacing-out
of individuals may have been related to the foraging activities of the
long siphons on thc surface of the sand. This type of distribution is
probably the exception among burrowing animals.
The subject of dispersion, or microdistribution, is discussed by
Greig-Smith (1957, chapter 3). The pattern, as he points out, is not
an absolute characteristic, but is dependent on the size and sometimes
the shape of the samplc area. It is therefore misleading to write of
populations as random, aggregated, or evenly distributed unless the
are8, and scale of sampling are cnvisaged. To take a simple example
we may consider the sampling of any regularly-arranged objects,
such as apple trees in an orchard. If the sampling area is very small
and confined to the orchard, no departure from a random distribution
will be detected (cp. Greig-Smith, 1957, p. 53). Increasing the sampling
area, still within the orchard, will show that the trees are very evenly
distributed, the coefficient of dispersion being less than unity. Sampling
on a larger scale will show a very significant aggregation of treesinto orchards, with few or no apple trees in the intervening are@. . 4
still larger scale of Rampling might show that, as between counties or
districts, the numbers of apple trees showed a random dispersion.
It should be noted that we are here concerned only with the total
numbers per sample, sincc in benthos sampling the exact position of
individuals is not known. Where the position of individuals can be
ascertained, other techniques such as the ‘‘ distance to nearest neighbour” may be used to further nnalyse the pattern of distribution
(Holme, 1950; Clark and Evans, 1954).
A study of small-scale distribution of animals on the sea floor has
been made by Holme (1953). Using a scoop-sampler taking two
adjacent samples simultaneously (Fig. 40) i t was possible to analyse
the Variance as between the pairs of samples in each haul and as
between successive hauls, with the ship drifting. The samples, which
exhibited Rome tendency towards aggrcgation in most species, showed
almost as much variation between the pairs of samples m between
successive hauls. (It is unlikely that the same result would have been
obtained on any other inshore ground in the Plymouth area where
extreme patchiness of the fauna and deposits is the rule (Ford, 1023)).
I n a useful paper, Marlier (1953), having noted that the number
of samples is nevcr sufficient to satisfy the statistician, goes on to
calculate, for a given population, what is thc minimum number of
samples necessary to show a significant difference in density as between
two populations. For a population of chironomids in a lake, Marlier
estimated the minimum number of samples (X) to be collected at two
249
intertidal bivalve niollusc Tellinn tenuis (Fig. 41) in which spacing-out
of individuals may have been related to the foraging activities of the
long siphons on thc surface of the sand. This type of distribution is
probably the exception among burrowing animals.
The subject of dispersion, or microdistribution, is discussed by
Greig-Smith (1957, chapter 3). The pattern, as he points out, is not
an absolute characteristic, but is dependent on the size and sometimes
the shape of the samplc area. It is therefore misleading to write of
populations as random, aggregated, or evenly distributed unless the
are8, and scale of sampling are cnvisaged. To take a simple example
we may consider the sampling of any regularly-arranged objects,
such as apple trees in an orchard. If the sampling area is very small
and confined to the orchard, no departure from a random distribution
will be detected (cp. Greig-Smith, 1957, p. 53). Increasing the sampling
area, still within the orchard, will show that the trees are very evenly
distributed, the coefficient of dispersion being less than unity. Sampling
on a larger scale will show a very significant aggregation of treesinto orchards, with few or no apple trees in the intervening are@. . 4
still larger scale of Rampling might show that, as between counties or
districts, the numbers of apple trees showed a random dispersion.
It should be noted that we are here concerned only with the total
numbers per sample, sincc in benthos sampling the exact position of
individuals is not known. Where the position of individuals can be
ascertained, other techniques such as the ‘‘ distance to nearest neighbour” may be used to further nnalyse the pattern of distribution
(Holme, 1950; Clark and Evans, 1954).
A study of small-scale distribution of animals on the sea floor has
been made by Holme (1953). Using a scoop-sampler taking two
adjacent samples simultaneously (Fig. 40) i t was possible to analyse
the Variance as between the pairs of samples in each haul and as
between successive hauls, with the ship drifting. The samples, which
exhibited Rome tendency towards aggrcgation in most species, showed
almost as much variation between the pairs of samples m between
successive hauls. (It is unlikely that the same result would have been
obtained on any other inshore ground in the Plymouth area where
extreme patchiness of the fauna and deposits is the rule (Ford, 1023)).
I n a useful paper, Marlier (1953), having noted that the number
of samples is nevcr sufficient to satisfy the statistician, goes on to
calculate, for a given population, what is thc minimum number of
samples necessary to show a significant difference in density as between
two populations. For a population of chironomids in a lake, Marlier
estimated the minimum number of samples (X) to be collected at two