Page 7 - v14n2a04

Fattorini & Dapporto
Revista de Gestão Costeira Integrada / Journal of Integrated Coastal Zone Management 14(2):185-197 (2014)
three aspects were evaluated using information provided in
Aliquò et al. (2006). Geographical distribution was evaluated
with reference to the number of Italian administrative
mainland regions from which each species is known. Species
occurring in less than four regions (<20% of regions) were
considered as geographically rare. Species associated with
coastal habitats were considered as having restricted habitat
specificity. Species reported as scarce by Aliquò et al. (2006)
were considered as having scarce populations. Then, an
eight-score scale was created that reflected different types
of rarity and commonness, and each species was assigned
to a score as follow: 1: species that are not rare; 2: scarce
species (i.e. species rare for abundance); 3: species with
narrow habitat specificity; 4: restricted species (i.e. species
rare by range); 5: scarce species with narrow habitat
specificity (i.e. species rare for both habitat specificity and
abundance); 6: scarce and restricted species (i.e. species rare
for both geographical range and abundance); 7: restricted
species with narrow habitat specificity (i.e. species rare for
both habitat specificity and geographical distribution);
8: restricted and scarce species with narrow habitat
specificity (i.e. species rare for geographical distribution,
habitat specificity and abundance). This weighting scheme
assigns higher importance to geographical rarity which is
appropriate for the purpose of our study because species
with a narrow geographical distribution are more vulnerable
at a global or regional level (e.g. endemics) (Kattan, 1992).
Using this weighting scheme in the BCC calculation, species
common for all rarity measures received an α-value of 1,
whereas species rare for all three aspects received an α-value
of 8. The BCC calculated using this scheme will be referred
to as BCC2.
In a third approach, we strictly focused on the type of
distribution shown by species. For this, we divided the species
into the following categories: species endemic to single islands geographical characteristics, with LSD post hoc tests for
(SIE); species endemic to an archipelago (END); species
occurring on one or more of the study islands, but not on
the Italian mainland or major islands (Sardinia and Sicily),
which is the case of species occurring on the African on Greek
coasts (NIT); species occurring on the Italian mainland or
major islands with highly fragmented distribution (FRG);
species that are widespread on the Italian mainland (WID).
For the calculation of the BCC index, the following
arrangement of weights was used: SIE (α= 16), END (α= 8),
NIT (α= 4), FRG (α= 2), WID (α= 1). In this case, weights
followed a geometric series to take into account various
sources of incertitude (see Fattorini, 2006a). In general, it
is unlikely that species found only on islands will be later
found also on mainland areas and endemics are obviously
an important conservation target in island conservation;
moreover, single island endemics are much more valued than
species which occur on more than one island. A the other
extreme, the assignment of a species to the FRG category
may be problematic, because it is difficult to establish if a
fragmented distribution reflects a true state of affairs or is
due to the lack of knowledge, thus species with an assumed
fragmented distribution have a score only slightly superior to
that of widespread species. Finally, the intermediate weight
assigned to the NIT category is due to the fact that species
occurring on Italian islands, but not on other Italian areas,
are of high conservation value from a national perspective,
- 191 -
even if they may be largely distributed in other countries.
This criterion is similar to the IUCN procedure for regional
red list assessments (http://www.iucnredlist.org/documents/
reg_guidelines_en.pdf ). The BCC calculated using this
scheme will be referred to as BCC3.
Finally, as a forth criterion, we calculated species rarity
using the perimeter of the inhabited islands. In the BCC1
index, we calculated species rarity on the basis of the number
of occupied islands. However, the islands included in our
analysis have enormous differences in size, so a species living
exclusively in a very small island should be considered much
rarer than another living in a large island. To explore this
form of rarity, we expressed species rarity as the inverse of the
summed perimeter of the islands where the species lives. The
rationale for using island’s perimeter is due to the prevailing
habitat type, i.e. coastal areas, whose extension should be
more related with perimeter than area. Because this index is
intimately related with the BCC1, but uses island perimeter
instead of island number, we called it BCC1p.
The same rarity measures were used to calculate the
respective BCW indices, which will be refereed to as BCW1,
BCW2, BCW3, and BCW1p, respectively.
To investigate if the four BCC indices are influenced
by island geography, we correlated their values with island
area, elevation and distance to the closest mainland using
a Pearson correlation coefficient. For each BCC index, we
considered the islands included in the third quartile of the
distribution of BCC values as priority islands. Then, we
compared the various indices to assess the percentage of
“priority islands” shared by two or more indices. Finally, we
used ANOVAs to test if islands selected as priority islands
by the four indices showed significant differences in their
pairwise comparisons. The same approach was used for the
four BCW indices. Probability levels were set at 0.05 in all
tests.
3. RESULTS
Values of BCC1, BCC2, BCC3, and BCC1p are given in
Table 3. Values of BCC1 ranged from 0.004 to 0.827 (mean
± SE = 0.149 ± 0.019). Values of BCC2 ranged from 0.000
to 0.952 (mean ± SE = 0.369 ± 0.034), but only one island
(Vivara) had a value of zero. Values of BCC3 ranged from
0.000 to 0.644 (mean ± SE = 0.149 ± 0.023), with nine
islands having a value of zero. Values of BCC1p ranged from
0.0004 to 0.381 (mean ± SE = 0.020 ± 0.007).
Values of BCW1, BCW2, BCW3, and BCW1p are
also given in Table 3. Values of BCW1 ranged from 0.000
to 0.161 (mean ± SE = 0.029 ± 0.005). Values of BCW2
ranged from 0.000 to 0.225 (mean ± SE = 0.047 ± 0.006),
but only one island (Vivara) had values of zero. Values of
BCW3 ranged from 0.000 to 0.210 (mean ± SE = 0.028 ±
0.005), with nine islands having a value of zero. Values of
BCC1p ranged from 0.0002 to 0.3688 (mean ± SE = 0.0264
± 0.007).
On average, BCC2 attained higher values than BCC1
(paired t-test, t = 6.188, p<0.0001) and BCC3 (t = 14.333,
p<0.0001), whereas BCC1 and BCC3 did not show a
significant difference in their average values (t = 0.025, p =
0.980). BCC1p showed lower values than BCC1 (t = 2.299,
p<0.0001), BCC2 (t = -10.799, p < 0.0001), and BCC3 (t
= -5.967, p < 0.0001).