biochemical composition, thus rendering the experiments internally comparable; how- ever, in practice, some variation in the algal biochemical composition is to be expected.
Using monospecific diets, the growth performance of the larvae can be correlated with algal biochemical properties, without introducing differences in digestibility, cell size
and palatability which may have compromised previous studies using multispecies diets. Special emphasis was placed on algal fatty acids due to ever increasing evidence
suggesting the important role of these compounds as diet components of particular
Ž .
nutritional value Thompson et al., 1996 .
2. Materials and methods
2.1. Algal cultures, chemical and biochemical analyses The algal species diet used as food for the larvae of Mytilus edulis were PaÕloÕa
Ž lutheri obtained from Culture Centre of Algae and Protozoa, Oban, Scotland, CCAP
. Ž
. Ž
. 931r1 , Chaetoceros muelleri CCAP 1010r3 , Rhinomonas reticulata CCAP 995r2
and Skeletonema costatum, the strain of which was isolated by the first author, from a Ž
. sample taken from Syros island Greece . The algae were grown using a continuous
culture regime under a light–dark cycle of 14–10 h, at 208 18C. Two light intensities Ž
y1 y2
were employed high-light, HL s 890–950 kphotons s m
and low-light, LL s
y1 y2
. 275–337 kphotons s
m , and three nutrient conditions were used in each light
Ž .
Ž . Ž .
intensity: no nutrient limitation fr2 , phosphorus P and nitrogen N limitation. The phosphorus-limited medium and nitrogen-limited medium were prepared by adding a
quarter of the amount of the respective salt. The density of the cultures was determined twice daily and adjustments to the flow rate were made on a daily basis to keep the cell
density constant. Typically, the cultures were maintained in the selected cell density
10 throughout the feeding trials. The collector flasks were emptied regularly and algae, which were to be used either as food or for samples for cell density determination
and for the subsequent biochemical analyses, were left to stand for no more than 3 h in that vessel. The algae used in the control diet were cultured in 20 l glass round flasks at
a light intensity of 715–827 kphotons s
y1
m
y2
using a semi-continuous batch culturing system and Conway medium under constant illumination. All cultures were maintained
in uni-algal condition. Chemical and biochemical methods and techniques are described Ž
. in Leonardos and Lucas 2000a and the biochemical results are expressed in a per cell
basis since this was the unit of standardization for the feeding trials. Three samples underwent protein and carbohydrate analyses while the fatty acid profile was determined
in two samples.
2.2. Mussels spawning and feeding trials Ripe mussels, M. edulis, collected from the Conwy estuary, N. Wales, in the spring
Ž .
of every experimental year 1995–1997 , were used as broodstock for the trials. These were kept at 68 18C in flowing seawater until required for spawning. Mussels of ca. 60
mm in length were selected and epifauna and byssals threads were removed. Spawning Ž
. was induced as in Loosanoff and Davis 1963 . Fertilized eggs were incubated, unfed at
148C for 4 days and at the D-stage larval pool, the percentage of normality and Ž
Ž .
abnormality according to the developmental stages described by Bayne 1965 as well as the total number of larvae were calculated. A small sub-sample of larvae was taken
Ž from this group and photographed or videotaped see later part of the section for image
. analysis under a microscope for later calculation of the initial mean length of the larvae.
Ž .
All batches of larvae used had a small - 10 percentage of abnormal larvae. An appropriate volume was taken from the larval D-stage group and diluted to a
concentration of 30,000 larvae l
y1
. These were transferred into sufficient plastic 1-l beakers to give three replicates for each treatment. To each beaker, the respective
species and quantities of algae were added and then filled to 1 l and left at 148 18C. Each diet was tested in triplicates. Feeding densities of 50 cells ml
y1
were the same for all diets; the control diet was a mixture of P. lutheri and R. reticulata in a 4:1 ratio. A
change of the 0.2 mm filtered water and feeding of the larvae was carried out every second day with the respective algal food added to a final concentration of 50 cells
ml
y1
. For the high-light and low-light experiments of the two diatoms, S. costatum and C. muelleri, two different batches of larvae were used for each light intensity; hence,
Ž .
results are compared against two sets of controls control diet and unfed larvae . The high- and low-light experiments of R. reticulata and P. lutheri were carried out with
the same batch of larvae; therefore, only one set of controls was needed for each trial. After 2 weeks, samples were taken from the larval cultures to calculate the total
Ž .
number of live larvae as well as percentage of normal individuals Bayne, 1965 ; the samples were stored in a glass vial and fixed with few drops of Lugol’s iodine solution.
Each sample was, at a later stage, decanted into a sedimentation chamber and counted using an inverted microscope.
Growth was estimated on the basis of length increase of the larvae. To measure length, two image analysis techniques were used. The larvae were measured at the
Ž .
beginning of every experiment while at D-stage and at the end, after a 2-week period. The first technique was based on measuring photographic negatives of the animals and
was used for the S. costatum and C. muelleri feeding trials, while the second was based on processing videotaped information of the larvae and was employed in the P. lutheri
and R. reticulata experiments. The first steps for both techniques were the same: the larvae from each container were concentrated by pouring its contents into a 45-mm
nylon sieve. A small sub-sample was taken with a pipette and placed onto a slide.
In the photographic technique, the mean length of the larvae was estimated from Ž
. negatives taken with a Nikon FM2 camera; 60 normal Bayne, 1965 , clearly focused,
larvae were measured from the negative using the same microscope fitted with an eyepiece scale. The length of the larvae was calculated using a magnification factor
computed from an image of a standard graticule.
In the videotape technique, the slide with the larvae was placed on a Nikon inverted Ž
microscope with a microscope video camera Hitachi MOS Colour Video Camera, .
VK-C150ED attached. The resulting picture was recorded with a video recorder
Ž .
Panasonic, NV-J35HQ . The recorded images were later transferred to an IBM-compat- ible computer, equipped with a Miro PCTV video card. Images of the larvae in clear
Ž .
focus were grabbed with the aid of Microsoft’s Media Manager Video capture 32 and the resulting still images were saved as bitmap images on the computer’s hard disk.
Ž .
Then an image analysis software package Jandel Scientific Sigma Scan was used to measure the length of 60 larvae. This technique was compared with the photographic
technique by measuring the same slide of larvae. The mean values obtained by the two techniques did not differ by more than 5.
2.3. Statistical analysis Before applying any parametric test each data set was tested for normality and
homogeneity of variances, with a normal probability plot and with Barlett’s test, respectively. Significance was tested at the 95 level.
Logarithmic transformation of data resulted in more data sets having a normal distribution; however, in no instance were all the distributions of all data sets normal.
Even when one of the assumptions necessary for applying parametric tests was violated Ž
. in a small part of the data set typically one or two sets of larval lengths , parametric
tests were used. The experimental design was such that nested ANOVA was used in all experiments to assess whether significant differences existed between diets. This test
also reveals whether there is significant variation between the replicates of the same diet.
Ž In order to identify the groups that demonstrated significant differences 95 confidence
.
X
Ž limit , the multiple comparisons between means of T -method was used Sokal and
. Rohlf, 1987 . All tests were carried out with the aid of the statistical software package,
Minitab
w
, while preliminary calculations and graphs were made in Quattro
w
Pro. To identify any relationship between algal biochemical component and larval growth,
correlation analysis was used. This type of analysis was used instead of ordinary regression analysis because the assumptions of model I regression analysis do not hold
Ž .
true with this type of data Sokal and Rohlf, 1987 .
3. Results
