Page 4 - FATTORINI2011
P. 4

TABLE 2. Extent (km2) of land cover categories and values of environmental (landscape) heterogeneity indices. N – number of
land cover categories present on an island. Land cover categories: BU – Built up areas; CU – Cultivation; CO – Coniferous forests;
BL – Broad-leaved and Mixed Forests; SC – Sclerophyllous vegetation; BR – Bare rock and sparsely vegetated areas; WA – Wet
areas. Landscape heterogeneity indices: C – Simpson dominance index; H – Shannon index; eH/N – Buzas and Gibson evenness; J –
Pielou equitability; DMg – Margalef richness index; d – Berger-Parker dominance.

Islands          N                        Landcover categories                                  Heterogeneity indices
                       BU         CU CO BL SC BR WA                                  C H eH/N J DMg d

1. Lipari        4 9.397 13.196 0.000 0.000 8.198 6.498 0.000 0.268 1.352 0.967 0.976 0.285 0.354

2. Salina        3 3.897 3.298 0.000 0.000 19.185 0.000 0.000 0.566 0.774 0.723 0.705 0.197 0.727

3. Vulcano       5 3.598 6.785 0.000 0.514 7.505 2.467 0.000 0.279 1.380 0.795 0.857 0.402 0.360

4. Stromboli     3 1.125 0.000 0.000 0.000 7.033 4.032 0.000 0.451 0.903 0.823 0.822 0.213 0.577

5.Filicudi       3 0.000 1.998 0.000 0.000 6.693 0.799 0.000 0.549 0.783 0.729 0.712 0.218 0.705

6. Alicudi       3 0.400 0.000 0.000 0.000 3.000 1.700 0.000 0.463 0.878 0.802 0.799 0.234 0.588

7. Panarea       3 0.786 0.000 0.000 0.000 1.866 0.688 0.000 0.411 0.991 0.898 0.902 0.247 0.559

8. Basiluzzo     1 0.000 0.000 0.000 0.000 0.290 0.000 0.000 1.000 0.000 1.000 0.000 0.000 1.000

9. Lisca Bianca  1 0.000 0.000 0.000 0.000 0.000 0.000 0.041 1.015 –0.007 0.993 0.000 0.000 1.000

10. Bottaro      1 0.000 0.000 0.000 0.000 0.000 0.000 0.007 1.088 –0.044 0.957 0.000 0.000 1.000

11. Scoglio Faraglione 1 0.000 0.000 0.000 0.000 0.005 0.000 0.000 1.501 –0.249 0.780 0.000 0.000 1.000

12. Pietra del Bagno 1 0.000 0.000 0.000 0.000 0.000 0.000 0.002 1.103 –0.051 0.950 0.000 0.000 1.000

13. Ustica       5 1.163 2.609 1.013 0.000 3.133 0.681 0.000 0.263 1.453 0.855 0.903 0.442 0.364

14. Levanzo      2 0.000 1.006 0.000 0.000 4.604 0.000 0.000 0.706 0.470 0.800 0.678 0.116 0.821

15. Favignana    3 0.861 13.864 0.000 0.000 4.975 0.000 0.000 0.561 0.732 0.693 0.666 0.202 0.704

16. Marettimo    3 0.000 0.728 0.000 0.000 10.226 1.107 0.000 0.731 0.528 0.565 0.481 0.213 0.848

17. Pantelleria  5 8.671 51.041 6.196 0.000 19.801 0.000 0.291 0.421 1.088 0.594 0.676 0.352 0.594

18. Linosa       3 0.854 3.117 0.000 0.000 1.369 0.000 0.000 0.432 0.956 0.867 0.871 0.233 0.584

19. Lampione     1 0.000 0.000 0.000 0.000 0.000 0.000 0.025 1.000 0.000 1.000 0.000 0.000 1.000

20. Lampedusa    4 2.898 5.741 0.857 0.000 10.704 0.000 0.000 0.384 1.107 0.756 0.798 0.303 0.530

21. Malta        7 80.380 126.750 0.694 1.488 33.156 3.272 0.261 0.391 1.089 0.425 0.560 0.483 0.515

22. Gozo         4 29.458 27.761 0.000 0.000 9.562 0.219 0.000 0.385 1.023 0.695 0.738 0.270 0.440

23. Comino       1 0.000 0.000 0.000 0.000 3.500 0.000 0.000 1.000 0.000 1.000 0.000 0.000 1.000

24. Cominotto    1 0.000 0.000 0.000 0.000 0.250 0.000 0.000 1.000 0.000 1.000 0.000 0.000 1.000

25. Filfla       1 0.000 0.000 0.000 0.000 0.060 0.000 0.000 1.000 0.000 1.000 0.000 0.000 1.000

Data analysis                                                       – Shannon index (entropy): H             Ai  ln  Ai  . H ranges
                                                                                                               A       A

Correlation of species richness with geographical and               from 0, if one land cover category dominates the landscape
environmental variables
                                                                    completely, to arbitrarily high values for landscapes with many

One of the most controversial issues in island biogeography is      categories, each with a small extent.

the importance of “habitat diversity” and area per se in deter-     – Buzas and Gibson evenness: eH/ N, where N is the number

mining species numbers (Whittaker, 1998; Fattorini, 2006a;          of categories and H is the Shannon index.

Hortal et al., 2009). To express “habitat diversity”, several       – Pielou equitability (evenness): J = H/ lnN.

authors have used the number of biotopes occurring in study         – Margalef richness index: DMg = (N – 1)/ ln(A).

areas (see Fattorini, 2006a; Tognelli & Kelt, 2004; Hortal et al.,  – Berger-Parker dominance: d = Amax/A, where Amax is the

2009). In addition to the number (N) of land cover categories       extent of the dominant land cover category.

defined above, I used selected synthetic indices of environ-        Species richness can be related to environmental variables,

mental (landscape) heterogeneity. Although Shannon’s index is       such as measures of island area, distances and environmental

typically used (e.g. Lobo & Martín-Piera, 2002; Nogués-Bravo        heterogeneity in different ways. The relationship between

& Martínez-Rica, 2004; Stefanescu et al., 2004; Maes, 2005),        number of species and area (species-area relationship, SAR) was

studies on community ecology have demonstrated that no single       here best modelled with the power function S = CAz, where S is

diversity index encompasses all the characteristics of an ideal     the number of species, A is an island’s area, and C and z are

index (Magurran, 1988; Krebs, 1999). Therefore, I also used the     fitted parameters (Martín & Goldenfeld, 2006; Fattorini, 2006b;

following indices derived from those used in studies on commu-      Dengler, 2009).

nity ecology (Legendre & Legendre, 1998; Magurran, 1988,            The power function may be fitted as lnS = lnC + z lnA using

2004; Hayek & Buzas, 2010) to express richness, dominance,          ordinary linear regression, or directly as S = CAz with a non-

evenness and relative abundance:                                    linear regression. Although the link function is identical, the two

– Simpson dominance index: C        Ai  2                          models have different distributional assumptions and are not sta-
                                     A
                                          , where Ai is the         tistically equivalent for least-square regression (see Williams et

extent of the land cover category i, and A is the total surface of  al., 2009 for details). Because there is no biological or statistical

the island. C can vary from 0 if all land cover categories have     preference for either model (Dengler, 2009), both were used in

equal extent to 1 if one category dominates the landscape com-      this study. The untransformed model was applied using a Quasi-

pletely.                                                            Newton algorithm with Statistica 6.0 software. As suggested by

662
   1   2   3   4   5   6   7   8   9