Aristotle University Of Thessaloniki (AUTH)
Χωρικά πρότυπα βιοποικιλότητας σε πολλαπλές χωρικές κλίμακες
Abstract
dc:descriptionThe primary question of the PhD thesis refers to the identification of patterns of species diversityacross different spatial scales. The knowledge of spatial patterns of diversity is fundamental forunderstanding ecological and evolutionary processes. We focus, in particular, in estimation ofdiversity indices, in the species-area relationship (SAR) and its universality across scales and also innovel methods of community analysis and prediction (Chapter 1).Beta diversity provides a link between species diversity in a region -γ-diversity- to speciesdiversity within sites –α-diversity (chapter 2). According to Whittaker’s original definition, betadiversity is the change of community composition in relation to a gradient of environment and asimple index of it is the ratio of γ-diversity to α-diversity. Many metrics have been proposed forbeta diversity estimation each reflecting a different aspect of the phenomenon. However, many ofthem are more informative of the variation in alpha diversity among samples and the samplingeffort (number of sampling units) rather than of the species assemblage differentiation. In chapter2, we propose an alternative index of beta diversity, namely N*. We define N* as the averagesampling effort at which the total number of species occurrences shared I [=(Total number of speciesoccurrences R) – (Species richness S)] becomes equal to the species richness S. The index is estimated asthe intersection point between the average species accumulation curve (SAC) and the line y=(meanα-diversity/2)*number of sampling units. N* is not mathematically confounded to α- or γ- diversity, butit depends on the relationship between gamma and alpha and it reflects the variation in speciesoccupancies. Thus, if a number of random samples, sufficient for the reliable estimation of meanalpha diversity and the species accumulation curve, is collected, N* converges to a value that doesnot change by the inclusion ofmore samples from the same region. N* depends primarily on theproportion of new species expected in each next sample. We explore and test the properties of N*on simulated data and examine its applicability on a dataset of plant diversity from Greek Natura2000 protected areas. N* declines with nestedness in community structure, while it increases withspecies turnover. This holds also for plant diversity of Greek Natura 2000, where N* exhibitedlower values in island regions compared to continental regions reflecting the increased nestednessof island communities.The species-area relationship (SAR) represents the relationship of species richness with thescale (chapter 3). The relationship is one of the most general and studied patterns in ecology which -almost axiomatic- is referred as one of the few laws in ecology. The SAR is a powerful tool forspecies richness estimation and for projecting biodiversity loss due to habitat loss or fragmentation.The species-area curves are usually constructed by spatially sparse ecological data due to thedifficulty and cost of obtaining complete species list or performing exhaustive biodiversity surveys.Howerve, these curves are often handled as if they were SAR’s referring to contiguous area. The SAR slope depends on beta diversity patterns. It has been proposed that the value of slope is anindex of beta diversity. However, in order to use slope as beta diversity index, it requires knowingthe “true” sampling area. In chapter 3,, we approach the concept of “true” sampling area throughthe concept of “effective” area. We define effective area (Ae) as the contiguous area containing anumber of species equal to that obtained from a non-contiguous sampling (As). For our analysis,we used data from contiguous area surveys for plants (Great Britain, Mt.Holomontas), birds (GreatBritain, New York State, San Diego County and Massachusetts) and butterflies (Great Britain). Weimplemented contiguous, non-contiguous and random sampling designs which differed in samplingdensity (=sampling area/extent) for the estimation of species richness and the construction ofspecies richness curves. Based on our assumption that As≤Ae≤E, that is the effective area isbetween the sampled area and sampling extent (E) we formulated the simplest model relating Aeand As. We assumed Ae is proportional to sampled area: Ae=k As => k= Ae/As i.e. the relationshipis linear in a log-log space. Τhen, we focused our analysis on the factors affecting the value of theparameter k that we estimated directly from the data. The values of k were correlated negativelywith sampling density and positively with the average distance between samples and beta diversity.The species-non-contiguous area relationships (SASRs) overestimated the species richness for anygiven sampled area, while species-extent relationships (SAERs) underestimated the species richnessof the extent. The "effective area" can be used to correct the bias of species richness–arearelationship introduced by non– contiguous sampling designs.The universality of the shape of the species–area relationship (SAR) is subject of scientificdebate (chapter 4). The fundamental question is whether SAR reflects biological processes or is just astatistical/mathematical relationship. Recently, Storch et al. (2012) argued that species–arearelationships collapse into a single curve at continental scales after the axes are rescaled adjustingthe area-axis being to the mean range size of species and the species-axis to the species richness ofan area equal to mean range size. At continental scales, species ranges are contiguous, and the grainof analysis approaches the mean range size of species; therefore, the majority of species are oftenonly present in one or a few sampling units, leading to the linear scaling of species richness.Consequently, space have no effect on species distributions. At this point, species turnover reachesa maximum value. Thus, the effect of variation on species turnover is minimal, and differences inrange sizes formulate the shape of the SAR curve. In chapter 4, we test the universality of therescaled SAR across scales smaller than the continental level. We also test an alternative rescalingapproach, by using the total extent of the study area and total species richness. At the scalesanalysed here, we found that the proposed rescaling by Storch et al. adjusts for differences in theextent of the study area and for variation in gamma diversity. Consequently, while the rescaledcurves were closely aligned, they did not converge into a single curve. The rescaling of the two SARaxes, either by using the mean species range and corresponding species richness or by using thetotal extent of the study area and total species number, resulted in the differences being smoothedbut the curves did not collapse into a single curve. The observed remaining variation is independent of scale and biogeographical region, and is correlated with beta diversity, thus, it reflects thedifferentiation of species composition in space.The comprehension of community assembly rules is fundamental for the understanding ofdiversity patterns (chapter 5). In chapter 5, we explore whether network analysis can provideadditional information of community assembly comparative to classical methods of ordination andclassification. For our analysis, we used abundance (cover) data of plant diversity of the beechforests of North and Central Greece. The extent of coocurrence for every possible combination ofspecies was estimated by the phi coefficient fidelity index. In the network, two species are connected ifthe value of phi coefficient is greater than a defined threshold value. Setting different thresholdvalues resulted in networks that differed in the edge density. In the networks, we estimatedcentrality measures, communities and maximal complete subgraphs (cliques). We compared theresults of network analysis with the results of the classical analysis of communities. Cliques ofnetwork analysis are more informative of species occurrence patterns compared to classicalanalysis. Furthermore, we found that species’ centrality degree depends on species’ occurrencefrequency and abundance, but also on species’ co-occurrence pattern. Therefore, the degree isinformative of community structure, indicating whether the community diverges from randompattern. Communities of network analysis provide a rough picture of species grouping comparedto classical analysis.The observed diversity patterns are directly related to the patterns of species distributionsand consequently to environmental factors that affect them (chapter 6). However, our knowledge ofspecies’ distributions is incomplete and therefore, we have to use predictive models. The predictivemodels of species’ distributions are empirical models which correlate species data withenvironmental or/and spatial data. Species-level modelling is the most common approach, butalternative a community-level modelling can be implemented. In chapter 6, we implemented ManyGeneralized Linear Model analysis to construct a community-level model. For our analysis, we usedabundance (cover) data of plant diversity of the beech forests of Rodopi Mountains. Our modelgave a relatively good estimation of species’ occurrences and abundances. Furthermore, thedifferentiation in species composition in space (overlap of species’ distributions) is similar with theobserved community
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2014
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Lazarina, Maria
- Λαζαρίνα, Μαρία
Subjects
dc:subject × 19- β-ποικιλότητα
- Χωρική αντικατάσταση ειδών
- Εγκιβωτισμός
- Σχέση αριθμού ειδών-έκτασης
- Καθολικότητα
- Ισοδύναμη συνεχής έκταση
- Δικτυακή ανάλυση
- β-diversity
- Spatial species turnover
- Nestedness
- Natura 2000
- Species-area relationship
- Universality
- Effective area
- Network analysis
- Φυσικές Επιστήμες
- Γεωεπιστήμες και Επιστήμες Περιβάλλοντος
- Natural Sciences
- Earth and Related Environmental Sciences
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/37302