Hierarchically scaled surveys were carried out on beds of the brown mussel
Perna perna (Linnaeus) on the South coast of South Africa. The object was t
o assess spatial and temporal variations in the complexity of mussel beds a
nd to investigate relationships between mussel bed complexity and mussel re
cruitment. Complexity was divided into three components: physical complexit
y; demographic complexity; associated biota. A series of variables within e
ach component were recorded at two different scales (10 and 50 cm) within n
ested quadrats on three separate occasions. The nested ANOVA design explici
tly incorporated spatial scale as levels of the ANOVA. These scales were: s
hores (areas 1 km in length separated by 25 km); transects (areas 20 m in l
ength separated by 100s of meters); 50 x 50-cm quadrats separated by meters
and 10 x 10-cm quadrats separated by cm) This approach was intended to gen
erate hypotheses concerning direct associations between recruitment and com
plexity versus co-variation due external processes. Three main questions we
re addressed: (1) At what scale does each variable of complexity exhibit gr
eatest significant variation? (2) At these scales is there similar ranking
of variables of complexity and recruitment? (3) Within this/these scales, i
s there any significant relationship between the variables measured and mus
sel recruitment? On two occasions (Nov. 97 and Mar. 98) the majority of var
iables showed greatest significant variation at the transect-scale. On a th
ird occasion (Oct. 97) most variables showed greatest significant variation
at the quadrat-scale and the site-scale. On all occasions a markedly high
percentage of the variation encountered also occurred at the smallest scale
of the study, i.e., the residual scale of the ANOVA analyses. Some similar
ity in the ranking of variables occurred at the transect scale. Within the
transect-scale, there was little indication of any relationship between var
iables of complexity and recruitment. Relationships were inconsistent eithe
r among transects or among sampling occasions Overall, the results suggest
that a high degree of variation in mussel bed complexity consistently occur
s at very small scales. High components of variance generally also occur at
one or more larger scales; however, these scales vary with season. Mussel
recruitment does not appear to be directly affected by complexity of mussel
beds. Instead it appears external factors may influence both complexity an
d recruitment independently. In addition recruitment may influence complexi
ty rather than vice versa. (C) 2001 Elsevier Science BN. All rights reserve
d.