Page 7 - FATTORINI2011

TABLE 5. Eigenvalues, cumulative explained variance (%),
cumulative species-environment relationship (%) and species-
environment correlations for the first four constrained ordination
axes of CCA.

Axis 1 Axis 2 Axis 3 Axis 4

Land cover (raw values)

Eigenvalues 0.421 0.382 0.260 0.195

Cumulative % variance 8.6 16.4 21.7 25.7

Cumulative species-environment 25.9 49.3 65.3 77.3
relationship

Species-environment relationship 0.952 0.956 0.862 0.842

Land cover (% values)

Eigenvalues 0.378 0.304 0.256 0.159

Cumulative % variance 7.7 13.9 19.1 22.4

Cumulative species-environment 28.1.4 50.7 69.7 81.5
relationship

Species-environment relationship 0.870 0.304 0.256 0.159

CANCOR with geographical variables

With all canonical roots together, that is, without any

root removed, the overall CANCOR analysis gave F2 =
(21)

61.30, P = 0.000008 (canonical R2 = 0.87). For the Fig. 3. The relationships between circum-Sicilian islands
based on tenebrionid beetles using the Kulczynski 2 index and
second root, F2 = 24.27, P = 0.02 (canonical R2 = 0.67). UPGMA clustering.
(12)

Finally, the third root was clearly not significant

(canonical R2 = 0.18, F2 = 3.73, P = 0.82).
(5)

Total percentage of variance extracted in the X set influenced by distance to North Africa and sea depth to
Africa, while the second one was influenced by distance
(53.6%) was mainly represented by the variance extracted to Sicily. The third dimension of NMDS was represented
by the non-significant root 3. Thus, current and past isola-
by the first and second roots, while redundancy high- tion from Africa appear to be mainly responsible for the
biogeographic relationships among islands represented by
lighted the importance of the first root (Table 6). Total the first axis of the NMDS. Current isolation from Sicily
is responsible for the biogeographic relationships among
redundancy for the Y set was 57.3%. Looking at the islands represented by the second axis of the NMDS (Fig.
4).
factor structure (structure coefficients or canonical load-

ings), the first dimension extracted by NMDS was mainly

TABLE 6. Results from CANCOR with all canonical roots CANCOR with environmental variables
analysed. Variables are the same as in Table 1.

Factor Structure With all canonical roots together, CANCOR analysis
Root 1 Root 2 Root 3
using raw data of land cover gave F2 = 19.80, P = 0.53
Distributional patterns (21)
(coordinates from Non-metric
Multidimensional Scaling) –0.993 –0.044 –0.111
–0.098 –0.811 0.577
Dimension 1 –0.066 0.584 0.809
0.333 0.333 0.333
Dimension 2 0.288 0.224 0.061

Dimension 3

Variance extracted
Redundancy

Geographic characteristics 0.010 0.176 –0.202 Fig. 4. The relationships between circum-Sicilian islands
A 0.501 –0.776 –1.953 based on tenebrionid beetles using the Kulczynski 2 index and
Ds –1.748 –0.695 1.888 Non-metric Multidimensional Scaling.
Da –0.531 0.302 0.062
Di –0.733 –0.570 2.996
SDs 3.061 0.302 –4.565
SDa 0.212 0.470 –0.985
SDi 0.269 0.121 0.146
Variance extracted 0.233 0.081 0.027
Redundancy 0.865 0.671 0.183
Eigenvalues

665