La lepre italica e la lepre europea nel Lazio: modelli di idoneità ambientale a confronto
Author(s)
Properzi, Silvia
Date Issued
March 15, 2010
Type
Doctoral Thesis
Abstract
Few examples of habitat-modelling studies of rare and endangered species exit in literature, although from a conservation perspective predicting their distribution would prove particularly useful.
In this study we try to realize a comparison among the two species of hares that are present on the territory of Lazio region, in central Italy, through the comparison of habitat suitability models.
Habitat suitability models can be generated using methods based on expert knowledge or requiring information on species presence or specie presence and absence. Analytic solutions to accommodate the lack of absence include the ecological niche factor analysis (ENFA) and use of generalized linear models (GLM) with data of absence.
Italian and European hares were surveyed by night spotlight census method and by the genetic sampling of fecal pellet. The nighttime sampling has been effected beginning from a hour after the sunset: in average the censuses are extended for 2,39 hours for night. In the two years of search 202 road transects have been crossed, in 93 of which a sighting of Lepus has been effected at least; we have explored a total of 701,9 Kms; 278 belonging individuals have been individualized to the kind Lepus, of which 113 of Italian hare, 165 of European hare. We have crossed 510 genetic sampling transects for a total of 212,6 kms: the negative transects, in which fresh pellets of Lepus are not recovered, have been 404 (159,5 kms) and those positive 106 (53,1 kms), besides 173 transects (74,1 kms, 31 positive for 16,1 kms, 142 negative for 58,0 kms) have been conducted in territory without harvest activity and the remainders to 337 transects (138,5 kms, 75 positive for 37,0 kms, 262 negative for 101,5 kms) have been conducted in non protected territory. In total we have been picked 498 fecal pellets of Lepus (270 samples and 228 replicas); we also picked 285 samples out of genetic transects (113 champions and 172 replicas), during exits in field in areas of particular interest.
The picked data have been used in two ways: the data gotten by the genetic analysis meet in the mathematical models, because thanks to the alternation of points of presence and absence they allowed their application to effect a comparison between the GLM and the ENFA.
The whole the data of presence, from genetic sampling and from spotlight censuses, are used for the validation of the models. We used in all the kind habitat models twenty-four categories of environmental descriptor were included in the analysis: topographical comprising altitude, slope and aspect, anthropogenic comprising density of road, biological as land use cover.
We transform the expert knowledge into a numerical form and then through GIS applications we produced a new information by both combining information from different sources and by spatial analysis of existing data.
The ENFA shows that the most important ecogeographical variables determining habitat suitability for the Italian hare were Broad-leaved forest, natural grasslands and the altitude; while the first factor of specialization are permanent crops, and permanent crops to in partnership temporary crops. The most important ecogeographical variables determining habitat suitability for the European hare were the altitude, natural grasslands and the absence of hunting; while the first factor of specialization points out that this kind is particularly sensitive to the Sclerophyllous vegetation and to the Artificiel surfaces.
The GLM the procedure has individualized for the Italian hare 8 ecogeographical variables particularly the broad-leaved forest (311), natural grasslands (321), the altitude, non-irrigated arable land (211), mixed forest (313), moors and heathland (322), land principally occupied by agriculture, with significant areas of natural vegetation (243) and coniferous forest (312). The variables underlined for L. europaeus have been 10: the altitude, the absence of hunting, the artificiel surfaces (1), non-irrigated arable land (211), the slops, bare rock (332), permanently irrigated land (212), moors and heathland (322), complex cultivation patterns (242), transitional woodland/shrub (324).
We evaluated the habitat suitability model accuracy by cross-validation procedure and we use the Boyce index to compare the performance of each model. We obtain different value for each model, the expert knowledge model obtain the lowest value (0.6) while ENFA 0.81 and GLM 0.73; for European hare expert knowledge model obtain the lowest value (0.4), while ENFA 0.71 and GLM 0.94
For the spatial comparison of the models we used the minimal predicted area (MPA),it is the minimal surface obtain by considering all pixels with predictions above a defined probability threshold (70 in this case). Results show that the habitat suitability maps obtain with ENFA predict a MPA value that is approximately nine times of GLM based method. These results seem to confirm the tendency of the ENFA to over-predict species distribution (Zaniewski et to 2002, Engler et to the. 2004), due to the lack of discriminate absence; while the GLM could also have produced a most restrictive models of the ENFA because of the false absences that could be been use.
All the analyses show that some variables are more important than other. Particularly, in the case of the Lepus corsicanus both the ENFA that the GLM underline that broad-leaved forest, natural grasslands and the altitude are factors that characterize the niche. The same occurs in the case of the Lepus europaeus where both the models underline a bond with the altitude, the absence of hunting and the artificial surfaces. Both the species show not to appreciate the presence of the man, but they prefer more natural areas, as underlined from the negative coefficients of marginality both for the ifi and for the artificial surfaces. Besides analyses extend to underline that the Italian Hare curtains to have a niche more narrow than European and so it might exist a spatial competition for the two species even if it should be deepened.
Results indicate a strong relationship between model performance and the kinds of species distributions being modeled. Some methods performed better, but no methods was superior in all circumstances. A suggestion is made that choice of the appropriate methods should be contingent on the goals and kinds of distributions being modeled.
In conclusion we can affirm that as it regards the two techniques of sampling, stoplight and genetic, the two methods must be use in complementary measure, above all on small areas. Some methods performed better, but no methods was superior in all circumstances. The deductive models, are goes discarded however, both because based on a subjective judgment and because it isn't reliable. the two statistic models extrapolate the same variables that mostly influence the distribution of the species, even if the ENFA extends to overestimate the suitable area in comparison to the GLM.
The analysis of the niche, has underlined a possible ecological competition among the two kinds of hares, hypothesis that needs further close examinations and study on the field adjusted. The theory, if confirmed, should have some direct effects on the management of the European Hare.
Additional information
Dottorato di ricerca in Ecologia e gestione delle risorse biologiche
File(s)![Thumbnail Image]()
Name
sproperzi_tesid.pdf
Size
3.07 MB
Format
Adobe PDF
Checksum (MD5)
5cc0db75017c1672ffcb14dc2f563a2d