The habitats of the mid-shelf were surveyed to the
West (tow 21), North West (tows 1, 2 & 3) and East (tows 9, 10 & 11)
of Lord Howe, and South (tow 17) and West (tow 19) of Balls Pyramid.
They formed a distinct habitat type characterised by dense algal
communities wherever hard substrata occurred (Figs. 6 & 7). The
spread of points in the 2 dimensional biplot of Figure 5 indicated
that the benthos was more variable in the deeper slope and shelf
break areas. Tow site 13 to the South East of Lord Howe Island was
distinct, with predominantly rubble, rock and stone supporting the
attachment of prolific whip gorgonian gardens not found elsewhere.
Deeper water tows to the North (tow 4), East (tow 12) and South East
(tows 5 & 8) of Lord Howe, and South East of Balls Pyramid were
similar with some hard bedrock algal communities and silty sediments
supporting gorgonians.
Pencil urchins, Phyllacanthus sp., were quite
common in some areas regardless of the hardness of the seafloor, but
were most abundant on the bedrock on the northern margin of Balls
Pyramid (tow 14) (Appendices 2 & 6). The algal cover was diminished
or denuded over reasonably extensive areas there. Across the trench
and on the South East margin of Lord Howe Rise (tow 13), the urchins
were among the gorgonians. On the North West of the Island (tow 4),
the urchins were encountered predominantly over the sandy bottom and
were presumably moving across or capturing brown algae drifting
along it.
Slightly less than half of the coral colonies
identified from the towed video (75 colonies) were recorded from the
reef south of Balls Pyramid and adjacent to South East Rock (tow
17). The reef here was bounded by low profile sand waves in 40m+,
onto rocky, then stony ground at 35m where the consolidated bedrock
was exposed. Algae were dominant over the broken and consolidated
bottoms and coral colonies were also established in these areas.
Figure 6. Multivariate 2 dimensional biplot of towed
underwater video data of benthic classes and their associated
substrata. The primary split was related to the amount of silt (open
circle) and secondarily by the amount of bedrock/reef (closed circle
and square). The closed circle sites were predominantly bedrock with
algal communities.
Figure 7. Depth profile of the tow path and
distribution of the recorded substrata and benthic classes on tow 1,
NW of Lord Howe Island, crossing the fossil reef.
|
Interpretation of multivariate analyses: biplots
In this report both the measures of abundance of the
various substrata and benthic classes and fish species from multiple
sites are analysed. Multivariate analyses are required to examine
the multidimensional relationships among these categories or fish
species. The following is provided to facilitate understanding and
interpretation of these analyses.
-
Biplots graphically display the multivariate
relationships of the rows (individual ‘sites’ or ‘samples’) and
columns (class/taxa) of a data matrix on a single two-dimensional
plot. The term ‘biplot’ therefore indicates that both sets of
relationships (i.e. among sites and among class/taxa) are displayed
in a single plot.
-
The biplots represent each site by a point on the
plot with different symbols used to identify their membership of a
group as determined by the multivariate analysis.
-
The class/taxa are represented on the biplot as
vectors (i.e. lines). The vectors are labelled with the class/taxa
names.
-
Of the total variation in abundance of all class/taxa,
the percentage explained by each dimension of the biplot is shown in
the lower left corner of the plot. The first dimension explains the
greatest percentage of the total class/taxa variance and is shown on
the x‑axis. The second dimension explains the next largest
percentage of the total class/taxa variance and is shown on the
y‑axis.
Ordination methods that represent high‑dimensional
data in low‑dimensional space are used for the analysis of the
multivariate substrate and benthic classes and, fish species data.
These unconstrained ordination analyses reduce the multivariate data
to a set of uncorrelated derived variables which are linear
combinations of the original variables, and which have been
calculated to account for the maximum amount of variability in the
data. In this report the biplots represent the first two dimensions
(or derived variables) from these analyses thus displaying the most
informative 2‑dimensional view of a multidimensional distribution.
The biplots show the relationships of the original variables (i.e.
class/taxa) to each other and indicates their role in explaining the
observed spatial (i.e. among sites) patterns. This is achieved by
super-imposing vectors for the original variables (i.e. class/taxa)
over the plot of points, which represent the spatial (i.e. among
sites) patterns.
In the biplots for fish abundance in this report the
species vectors generally form an arc defining the gradient
(direction) of greatest abundance. The length of a vector
approximates the variability (standard deviation) of the associated
species. Thus short vectors mean that the species is consistent in
abundance among sites and a long vector means that the species is
highly variable among reefs. If a reef has a high abundance of a
particular species, the site point and species vector are far away
from the origin and in the same direction. If a site has a low
abundance of a particular species, the site point and species vector
are in opposite directions and far apart. Site points close to the
origin represent sites that have typical abundances of all species.
Sites that are close together on the biplot have similar proportions
of most species. The angle between two vectors represents the
correlation between the two species that the vectors represent. Thus
if the angle between them is small (0°) the species are highly
correlated, if large (180°) the species are negatively correlated
and if at right angles (90°) the species are uncorrelated. For ease
of interpretation of the biplots only the vectors of those species
that correlated highly with the derived dimensions of variability
were shown on the plots. Therefore only a small proportion of the
species included in each analysis are displayed on the plots.
|
Multivariate analysis of the benthic and substrata
data identified 2 primary groups based on the extent of algae on a
site/tow (Fig. 6 and refer to Interpretation of multivariate
analyses: biplots). Those with high algal cover (closed circles)
contrasted with tows of low or no algal cover, which were more
accurately described as communities associated with unconsolidated
substrata (sand and sand waves), semi-consolidated substrata (rock
and stone) and the steeper-graded silty bottoms. Tow 16 was somewhat
unique in having extensive deep-water bedrock supporting sparse
algae as well as gorgonians. Tow 13, on the shelf break SE of Lord
Howe Island, was unique as it included all benthic classes recorded,
excluding barren seafloor.
Areas surveyed in the Commonwealth Sanctuary Zone to
the east of Lord Howe Island were not dissimilar to that elsewhere
in the Marine Park. The hard seafloor, algal-dominated community
(tow 10) was also found in the adjacent proposed State Sanctuary
Zone (tow 9), outside the northern boundary (tow 11), NW of the
Island (tows 1, 2 & 3), west of the Island (tow 21) and south and
west of Balls Pyramid (tows 17 & 19). The slope and shelf break
communities in the north of the same portion of the Commonwealth
Sanctuary Zone (tow 12) occurred SW of the Island (tows 4 & 8) and
NE of Balls Pyramid (tow 16).
