2.5.3. Autotrophic index AI The AI is a useful means of describing changes in periphyton communities between
Ž .
sampling dates APHA, 1992 . It was calculated as follows: AI s Biomass AFDW ,mg m
y2
rchlorophyll-a, mg m
y2
Ž .
Values 200 indicate heterotrophic associations andror the presence of non-viable organic material.
2.6. RemoÕal of periphyton by grazing: stomach contents analysis Towards the end of the experiment, one fish from each cage was removed and killed
and the stomach was removed by dissection. Stomachs were left in 10 formalin solution for 24 h, then transferred to a fixed volume of filtered water, the contents
removed and thoroughly dispersed. A 1 ml sub-sample was then transferred by large-bore pipette to a Sedgwick–Rafter cell and a minimum of 150 diatom units were counted and
identified. The composition of the stomach diatom community was then compared with that of the plastic bottle periphyton community in order to evaluate grazing selectivity.
3. Results
3.1. Periphyton analysis 3.1.1. Periphyton community composition
The periphyton communities on the plastic bottles and in the fish stomachs comprised micro- and macroorganisms from both animal and plant kingdoms. Attention was
focused on the dominant diatom communities, although observations were also made on other periphytic organisms. Freshwater oligochaetes were often observed living on the
bottles and, during observation of microalgae under the microscope, zooplankton, such
Ž .
as protozoa Amoebae , Rotifera and coelenterate hydrozoa, were frequently encoun-
tered. Diatoms, identified from both fish stomach contents and substrates, are summarised
Ž in Table 1. Chlorophyceae Oocystis, Ankistrodesmis, Actinastrum, Pediastrum, Spir-
. Ž
. ogyra, Cladophora and Myxophyceae
Anabaena, Nostoc, Tolypothrix represented a relatively high proportion of the periphyton community prior to fish stocking. After
stocking, the filamentous algae became significantly reduced and proportionately more diatoms were apparent. After 4 weeks of grazing, the filamentous algae were found to be
reduced to very short colony lengths and Ankistrodesmus spp. became abundant, forming bundles in both the cage periphyton communities and in the fish stomach
contents.
Ž .
Nanoplankton - 2–3 mm diameter , diatoms, algae and, in the cages where feed Ž
. was given T1, T4 , feed particles, were abundant in the stomachs of all fish.
3.1.2. The diatom communities Diatoms were often found to be epiphytic on other, larger filamentous algae, or
periphytic, directly attached to the substrates. Epiphytic diatoms were reduced in size as
Table 2 Ž
. Changes in mean biovolume per diatom unit and the total diatom biomass biovolume per unit area s.d.
Ž .
Ž .
Ž .
before first line and 28 days after second line, italics stocking of fish O. niloticus Superscript letters are used to show which groups are significantly different from each other, based on Tukey’s
test. n.s.s not significant.
Parameter ANOVA
Treatment 1 Treatment 2
Treatment 3 Ž
. Ž
. Ž
. fish and feed
fish control
Ž .
Ž .
Ž .
Total biovolume Before
n.s. 191,440 49,800
139,340 19,340 100,420 31,760
of diatoms per unit surface area
y2
Ž .
ml cm Ž
. Ž
. Ž
. After
- 0.05
15,370 2750 7550 1770
245,960 73,360 Ž
. Ž
. Ž
. Mean biovolume
Before n.s.
4135 965 3509 1280
2706 1252 per diatom unit
3
Ž .
mm
a a
b
Ž .
Ž .
Ž .
After -
0.01 556
128 523
44 2421
455
Ž .
a result of fish grazing Table 2 . Table 1 shows differences in diatom composition between treatments and over time. Larger diatom species such as Melosira spp. or
Cycotella spp. were more represented in proportion than the smaller ones in the stomach of the fish. They were also less represented in proportion on the grazed substrates. The
mean diatom biovolume of grazed substrates was significantly smaller than those of
Ž .
ungrazed substrates Table 2 . Moreover, mean biovolume per diatom unit in the fish stomachs was 4925 2209 mm
3
, compared with values of - 2421 455 mm
3
in the Ž
. periphyton, and did not vary significantly among treatments P 0.05 .
3.1.3. Diatom community diÕersity A cluster analysis was first carried out on diatom community composition data of all
samples in order to assess species composition differences among treatments. The results are shown in Fig. 2. Four groups can be readily distinguished on the basis that they
share less than 40 similarity with each other: the diatom communities on bottles in the cages before stocking, the diatom community on the grazed bottles, the diatom commu-
nity recovered from the fish stomachs and the diatom community on ungrazed bottles 4 weeks after the beginning of the experiment. The analysis clearly demonstrates that
grazing affects periphyton community diversity. In the control cages, the periphyton community before and after stocking shows only 50 similarity. The biodiversity seems
to decrease significantly over time.
Fig. 3 shows the results of detrended correspondence analysis of the plastic bottle diatom community composition data before and after stocking. The communities can be
segregated into three distinct groups: those from the cages before stocking and after stocking and the control cages that were not stocked with fish. The graphic representa-
tion accounted for 58 of the total variance found in the diatom communities. Axis 1 is the most significant, representing 45.7 of the total variance. The high eigen value
Fig. 2. Cluster analysis showing percentage similarities of diatom species composition from samples from Ž
. bottles and stomachs of the fish O. niloticus grazing on these communities. Numbers refer to treatment and
Ž .
replicate within treatment e.g., III2 — treatment 3, cage 2 . Sample dates are given alongside community description. Groups are segregated if they show less than 40 similarities with others. The asterisk is used to
mark the cages before stocking.
Ž .
0.489 shows this is a strong trend that can be used with confidence to describe the data.
A Kendal rank correlation was carried out between the environmental parameters Žstocking density, biomass, average weight, Secchi disc values, temperature, amount of
. feed given, time and sex-ratio and the two axes of the detrended correspondence
analysis. The higher the eigen values, the stronger the trend within the data represented
Fig. 3. Graphic representation of the detrended correspondence analysis carried out on the diatom communities established on the plastic bottles in the cages before and after stocking. Numbers refer to treatment and
Ž .
replicate within treatment e.g., III2 — treatment 3, cage 2 . The asterisk is used to mark the cages before stocking.
by the axis. Axis 1, differentiating the stocked cages from the others, was highly correlated with stocked biomass, mean fish weight and stocking density.
The periphytic diatom communities were strongly affected by the presence of fish. Although Axis 1 is also moderately correlated with the amount of feed given, it does not
significantly affect the composition of the diatom communities as the treatments cannot be readily segregated from each other. Axis 2 shows a high correlation with time,
reflecting the fact that the ungrazed diatom communities change over time. Axis 2 also shows a moderate correlation with temperature, Secchi disc value and fish sex-ratio.
However, as with supplementary feed, those effects did not significantly affect diatom community composition.
3.2. Primary productiÕity The gross carbon production of the grazed periphyton community was significantly
Ž .
lower than that protected from grazing Table 3 . The proportion of photosynthetic activity due to periphytic phytoplankton appeared to remain the same, irrespective of
whether grazed or not.
3.3. Chlorophyll-a, phaeopigments, AFDW and ash content The evolution of the chlorophyll-a, phaeopigments, AFDW and ash content over time
and in the presence or absence of grazing pressure is shown in Fig. 4. The statistically
Table 3 Ž
. Primary production rates including gross carbon production and percentage respiration s.d., ns 3 of the
Ž .
Ž periphyton community in different treatments on two occasions before first line and 28 days after second
. Ž
. line stocking of fish O. niloticus
Superscript letters indicate which groups are significantly different from each other, based on Tukey’s test. Parameter
ANOVA Treatment 1
Treatment 2 Treatment 3
Ž .
Ž .
Ž .
fish and feed fish
no fish Ž
. Ž
. Ž
. Gross carbon
Before n.s.
1.30 0.65 1.95 0.92
1.03 1.00 production
y2 y1
Ž .
mg C cm h
a a
b
Ž .
Ž .
Ž .
After P - 0.05
1.41 0.95
1.45 0.31
3.30 0.47
Ž .
Ž .
Ž .
Percentage Before
n.s. 72.0 11.9
63.5 19.5 39.9 8.5
respiration Ž
. Ž
. Ž
. After
n.s. 29.7 23.5
33.8 12.3 32.8 9.6
Ž .
significant differences in all periphyton characteristics Table 4 prior to stocking with fish were reduced after stocking in all cages but those of the controls. While AIs
remained constant, periphyton ash content decreased over the experimental period.
Ž . Ž .
Ž .
Fig. 4. Development of a chlorophyll-a and phaeophyton, b ash-free dry weight AFDW and proportion of Ž .
Ž .
ash, and c autotraphic index AI over a 28-day period during the experiment.
Table 4 Multifactor analysis of variance of five parameters assessing the periphyton qualities in the cages on three
Ž .
occasions during the experiment ns162 Ž
. The influence of three factors on each parameter was assessed: cage block distance from shore; see Fig. 1
Ž . Ž .
Ž . Ž
. b , treatment t and bottle and sampling depth d . Where letters are given, a significant relation P - 0.01
was observed; n.s.s no significant relation. Day 0
Day 14 Day 28
y2
Ž .
Chlorophyll-a mg cm b, d
t t
y2
Ž .
Phaeopigments mg cm d
t t
y2
Ž .
AFDW mg cm d
n.s. t
Ž . Ash content
t, d n.s
t Ž
. Autotrophic index defined in text
d n.s.
n.s.
Ž .
The ratio of oxygen respired to gross oxygen production Table 3 was used to assess the importance of organisms other than algae in the periphyton community. The ratio did
not vary significantly among treatments, suggesting that the communities comprised similar proportions of primary producer to primary consumer, whether grazed or not.
Therefore, the removal of animals from the periphyton community by grazing appeared to occur in the same proportion as the removal of phytoplankton.
Ž .
Periphyton chlorophyll and AFDW changed over time Table 4 . Prior to stocking, variations in biomass were the same, irrespective of block distribution, treatment or
depth, and may be explained by differences in such environmental variables as exposure, current, light or nutrients. After stocking, variation in biomass distribution fell due to
treatment effects. Periphyton characteristics showed an increasing difference in variance between treatments over time. The variance in the control cages increased whilst that in
the cages stocked with fish remained the same.
4. Discussion