Search Results

You are looking at 1 - 10 of 53 items for :

  • "Generalized Linear Models" x
  • Refine by Access: All Content x
Clear All

The population density of three lacertid lizards (Podarcis sicula, Podarcis muralis, Lacerta viridis) was studied along several transects crossing agro-forest habitats in Mediterranean central Italy. Overall, seven transects, in three different wooded patches, were walked for lizards. Distance sampling (with uniform model design) was applied to the dataset in order to calculate population size, dispersion, and coefficient of variation at each site. In order to detect which factors may influence lizard density, a Generalized Linear Model (GLZ; multinomial distribution and cumulative log link function) was built, with environmental variables and density of predators’ variables being included in the model as covariates (scale predictor). Density of the three lizard species differed significantly among study sites, evidencing species-specific responses to local patch conditions. None of the environmental variables taken separately in the GLZ model influenced significantly the lizard densities, whereas lizards densities showed species-specific response to the considered environmental variables. The largest species (L. viridis) showed the highest density in the fragment with the most irregular shape and largest wood size, whereas the two smaller Podarcis species presented their highest population density in the site with the smallest wood patch and with a very low snake density. Density of P. sicula was negatively correlated with both the woodland area and Colubridae density, and was positively correlated with woodland shape (i.e. with circularity).

Restricted access

Old-fields (44, aged 1–15 years, from Czech Republic and Hungary) were sorted according to their soil moisture and nitrogen content into wet, mesic or dry, and nutrient poor, moderate or nutrient rich categories, resulting in 8 combinations (dry and nutrient rich fields were not present). The vegetation of old fields was sampled using phytosociological relevès. The changes in species cover data and importance of species trait categories were analysed in relation to three environmental factors, i.e., time since abandonment, soil moisture and total soil nitrogen using ordination, generalized linear models (GLM) and regression tree methods. Successional seres in the first 15 years after field abandonment were divergent. Species diversity significantly decreased with increasing site moisture and was highest in sites with moderate nitrogen content; while the relationship with time was not significant. Raunkiaer life forms and life strategies (sensu Grime) were generally the most predictive species traits considering species occurrence during the course of succession, the type of dispersal considering the different moisture status, and the ability to lateral spread considering the nutrient status of the old-fields. Most trends appeared in both parametric GLM and non-parametric regression tree analyses, several only in GLM. We consider regression trees to be a more convenient tool than GLM in cases such as ours with a rather small number of samples and robust character of data. Another advantage is that a hierarchy of species traits is taken into account. Thus, the occurrence of a species along an environmental gradient can be predicted if the species possesses a certain combination of traits.

Restricted access

Hastie, T.J. and D. Pregibon. 1992. Generalized linear models, In: J.M. Chambers and T.J. Hastie (eds.) Statistical Models in S . Wadsworth and Brooks/Cole, California, pp. 195–248. Pregibon D

Restricted access

935 McCullagh, P. and J.A. Nelder. 1983. Generalized Linear Models . Chapman and Hall, London. Nelder J

Restricted access

baseline differences between groups in mean cytokine concentrations, we then used a generalized linear model with log link and gamma family distribution, controlling for baseline level in each cytokine. For functional and laboratory measures the

Restricted access
Acta Veterinaria Hungarica
Authors:
András Gáspárdy
,
Gemma Gallagher
,
Boróka Bartha
,
Sándor Cseh
,
Sándor György Fekete
, and
Bence Somoskői

Abstract

The authors aimed to determine the plasma melatonin concentration in mares and their new-born foals in the early post-partum period. Blood samples were collected from the jugular vein of 53 mare-foal pairs within twelve hours after parturition. Plasma melatonin levels were measured by ELISA. The melatonin concentration, adjusted for the moment of parturition using a generalised linear model, was 34.58 pg mL−1 in mares. It was significantly lower (27.63 pg mL−1) in the new-born foals. However, the melatonin concentration declined differently by the end of the twelve hours, it decreased less in the offspring than in the mothers. An artificial light supplementation at the end of gestation reduced the melatonin concentration both in mares and their foals by about 10 pg mL−1, compared to the controls. An elevated melatonin production may be related to preparation of mares for parturition and ensures the chances of survival of offspring, therefore the melatonin may reach its peak at the moment of foaling regardless of its actual time. The effect of low melatonin concentration in new-born foals might be associated with the foal's health and subsequent performance. The need to monitor the melatonin concentration in the offspring justifies further studies.

Open access

Effective conservation of (semi-)natural grasslands requires an understanding of the factors affecting naturalness (i.e. the actual quality of a habitat or vegetation patch) and the importance of the particular factors. Both local or patch and landscape or matrix variables affect habitat quality, and the proportions of the effects need to be identified. Therefore, we performed a hypothesis generating and testing analysis with generalised linear models on three typical grassland habitat types (forest steppe meadows, Artemisia alkali steppes, and lowland wet meadows), differing in their fragmentation, ecology and history, and representing characteristic types of grassland habitats with the use of the national database of the vegetation of Hungary (MÉTA). Our results, in general, show that naturalness depends upon both intra-habitat and matrix attributes: presence or proportion of other habitat types in the surrounding landscape, threatening factors and landscape ecological attributes. Higher number of habitat types and higher proportions of (semi-)natural habitats in the landscape have significant effects: presence of other grassland types similar in ecological demands to the model habitat positively affect the naturalness, while non-characteristic, secondary or disturbed habitats and invasive alien species have negative effects. However, there are clear differences among the three habitat types, indicating that for effective conservation, good knowledge of conserved habitat types is essential. Landscape or matrix factors, both compositional and structural, affecting habitat patch quality have significant effects that cannot be overlooked. In the case of fragmented grasslands, matrix factors might be even more important than patch or local factors.

Restricted access

Alpine grasslands harbour species-rich communities of plants and invertebrates. We examined how environmental variables and anthropogenic impact shape species richness and community structure of terrestrial gastropods in alpine grasslands in the Val Müstair (Eastern Alps, Switzerland). Gastropods were sampled using a standardised method at 76 sites spanning an elevation range from 1430 m to 2770 m. A total of 4763 specimens representing 52 species were recorded. Correspondence analysis based on presence/absence data revealed that the grassland gastropod community was structured in a complex way with elevation, wetness, grazing intensity and inclination of the sites as key factors, while abundance-based analysis identified the importance of the elevation and wetness of sites. Generalized linear model showed that species richness decreased with increasing elevation and increased with increasing soil pH. The grassland gastropod communities were characterized by a high beta diversity, as indicated by the SDR-simplex analysis. Species-specific traits of gastropods showed sensitivity to the environmental characters of the sites, as shown by a fourth-corner analysis.

Restricted access

Local plant species richness and composition may vary across habitats and between plant taxonomic groups within temperate deciduous forests. Multi-taxon approach is therefore needed to provide a more detailed insight into determinants affecting vegetation structure. Fifty-four deciduous oak-dominated vegetation plots (20 m × 20 m) were sampled across central Slovakia (Štiavnické vrchy Mts) in order to study the effect of environmental (soil, light, topographic) factors on species richness and composition patterns of two main assemblages of understorey layer (herb-layer vascular plants and ground-dwelling bryophytes). The number of recorded herb-layer vascular plants and ground-dwelling bryophytes was 12–48 (mean 28) and 0–11 (mean 4) species per plot, respectively. Generalized linear model revealed that species richness of herb-layer vascular plants was driven by canopy openness, altitude, soil pH/base saturation gradient and plant-available phosphorus. Canopy openness and heat load index accompanied by soil pH/base saturation gradient determined changes of the ground-dwelling bryophyte richness. Canonical Correspondence Analysis identified soil pH/base saturation gradient, canopy openness, soil silt and topography related predictors (altitude, slope, radiation) as the main drivers of the herb-layer vascular plant compositional variability. Species composition variation of ground-dwelling bryophytes was controlled by radiation and canopy openness.

Restricted access

Abstract

The relationships between frugivorous animals and plants are of vital importance particularly in tropical forests. The way species interact and how they are organized within interaction networks could be determined by their ecological and morphological characteristics. This study evaluates the hypothesis that the topological position of species within an interaction network is determined by their degree of frugivory, body size, and abundance. Thus, we constructed the frugivory network between birds and plants in a rainforest fragment in northwestern Colombia. The position of the species within the network was calculated based on three centrality measures (degree, betweenness, and closeness), and its association with relative abundance, degree of frugivory, and body size of each bird species was evaluated by means of a generalized linear model. We found that the species that were most abundant and had the smallest body size had central positions in the interaction network. This pattern is contrary to what has been observed in pristine forests, where species with large body size are more important for network stability. Our results suggest that forest fragmentation modifies the roles of species within the network structure, in part, due to changes in the makeup of the original frugivore community. The information presented may be useful to evaluate the effects of the loss of species as a result of anthropic actions, with the aim of generating ecosystem restoration strategies.

Open access