10
T. B. REYNOLDSON
(Plate la). It is clear that no two lakes are identical in all these qualities
and it was necessary to depart from this ideal to some extent otherwise
too few lakes would have been available. However, if a moderately
sheltered shore and shallow water less than 1 ft (30.5 cm) deep were
absent or the stones were all very small ( <3 in, G7.5 cm), or very large
( 2 8 in, 2 2 0 cm), the lake was omitted from the quantitative work.
Such lakes left out for these reasons were included in qualitative sampling
if otherwise they provided a suitable habitat for triclads.
The sampling method had to be simple, not requiring any previous
visit or any complex bulky apparatus, applicable to limited variation
in size of stone and able to provide an adequate number of triclads.
After experimenting with various methods including trays, baiting and
timed hand collections, the last was found to be most reliable. It
enabled a wide area of shore to be examined which helped to even out
local concentrations of triclads. The time of collecting could also be
adjusted to population size so providing 50-100 specimens for analysis.
A basically similar method has been used successfully in other investigations, e.g. by Macan (1950) for molluscs and by Mann (1955) for
leeches. When triclads were scarce, collecting was continued up to a
maximum of 1 hour and if none were found they were recorded as
absent. For comparative purposes the data were converted into number
collected per hour. Since much practice in this sort of collecting had
been experienced prior to routine sampling, it is unlikely that the data
were influenced by an improving rate of collecting-a common danger.
However, the method suffered from some drawbacks. For example, as
the actual population increases, relatively more time is spent transferring triclads from stone to jar. This is akin to the artificial predatorprey situation analysed by Rolling (1959) and a similar relationship was
found between the real population and the number collected per hour,
as Holling found between predation rate and prey number (Reynoldson,
1958a, Table 2). The divergence between an estimated triclad population and the real one became large when the collecting rate exceeded
400 per hour. However, the majority of the populations (90%) were
below this value, most considerably below, so that the error from this
source is not extensive. Other drawbacks arise from the fact that usually
only one sample was taken from each lake thus no confidence limits can
be calculated. Reproducibility was tested by taking three sets of samples
from each of eight lakes in Anglesey during spring, summer and winter
and examining rank correlation. It might be added that this was a
stringent test since, apart from one lake of low calcium and total
mineral content, the rest had fairly high values in the range where experience had indicated already that there would be no marked effect on
the triclad population. The spring and summer samples showed a rank
T. B. REYNOLDSON
(Plate la). It is clear that no two lakes are identical in all these qualities
and it was necessary to depart from this ideal to some extent otherwise
too few lakes would have been available. However, if a moderately
sheltered shore and shallow water less than 1 ft (30.5 cm) deep were
absent or the stones were all very small ( <3 in, G7.5 cm), or very large
( 2 8 in, 2 2 0 cm), the lake was omitted from the quantitative work.
Such lakes left out for these reasons were included in qualitative sampling
if otherwise they provided a suitable habitat for triclads.
The sampling method had to be simple, not requiring any previous
visit or any complex bulky apparatus, applicable to limited variation
in size of stone and able to provide an adequate number of triclads.
After experimenting with various methods including trays, baiting and
timed hand collections, the last was found to be most reliable. It
enabled a wide area of shore to be examined which helped to even out
local concentrations of triclads. The time of collecting could also be
adjusted to population size so providing 50-100 specimens for analysis.
A basically similar method has been used successfully in other investigations, e.g. by Macan (1950) for molluscs and by Mann (1955) for
leeches. When triclads were scarce, collecting was continued up to a
maximum of 1 hour and if none were found they were recorded as
absent. For comparative purposes the data were converted into number
collected per hour. Since much practice in this sort of collecting had
been experienced prior to routine sampling, it is unlikely that the data
were influenced by an improving rate of collecting-a common danger.
However, the method suffered from some drawbacks. For example, as
the actual population increases, relatively more time is spent transferring triclads from stone to jar. This is akin to the artificial predatorprey situation analysed by Rolling (1959) and a similar relationship was
found between the real population and the number collected per hour,
as Holling found between predation rate and prey number (Reynoldson,
1958a, Table 2). The divergence between an estimated triclad population and the real one became large when the collecting rate exceeded
400 per hour. However, the majority of the populations (90%) were
below this value, most considerably below, so that the error from this
source is not extensive. Other drawbacks arise from the fact that usually
only one sample was taken from each lake thus no confidence limits can
be calculated. Reproducibility was tested by taking three sets of samples
from each of eight lakes in Anglesey during spring, summer and winter
and examining rank correlation. It might be added that this was a
stringent test since, apart from one lake of low calcium and total
mineral content, the rest had fairly high values in the range where experience had indicated already that there would be no marked effect on
the triclad population. The spring and summer samples showed a rank