9.1 Manual básico de Unity 3D
9.1.3 Un vistazo rápido a la interfaz de usuario:
Before modelling, soil variables were subjected to correlation analysis and the resulting Pearson correlation coefficients and associated scatter plot correlation matrix inspected for potential correlation between variables. Based on the correlation results the following ecologically relevant soil variables were selected for inclusion in the subsequent GLMM and REML modelling procedures, these included: total organic carbon (MIR 0-10 cm), cation exchange capacity (MIR 0-10 cm), texture (0-50 cm in 10 cm increments), pH (0-50 cm in 10 cm increments), carbonate reaction (0-50 cm in 10 cm increments), electrical conductivity(0-50 cm in 10 cm increments). The data for both total organic carbon and salinity were log transformed for analysis. For both GLMM and REML procedures only those soil variables that were most significant at p<0.05 level were retained to produce the most ecologically suitable and parsimoniously acceptable model.
Only those species that occurred at more than 10 sites across the study area were modelled. For the logistic GLMM modelling using only the quantitative line intercept data restricted the number species available to be modelled. However since this modelling procedure examines factors influencing the presence of a species in the landscape it was appropriate to increase the number of available species for modelling by including the semi-quantitative belt transect data set. This resulted in the modelling of 22 species (Table 7-5).
For the subsequent REML procedure which models percentage cover of a species conditional on its presence, only the line intercept data set was used. This reduced the number of species available for REML modelling to those most commonly occurring species for which accurate quantitative percentage cover data were available. The only species available were
Eucalyptus coolabah, Muehlenbeckia florulenta, Chenopodium auricomum, Atriplex nummularia and Tecticornia indica.
The distribution of individual plant species as informed by GLMM confirmed many of the relationships described above from both indirect (SSH-NMDS) and direct (CCA) multivariate analyses. The variables pH, organic carbon content, soil texture and salinity were all important correlates of the occurrence of the 22 species that could be modelled incorporating the random effect structure of the hierarchical sampling design (GLMM, logistic regression, Table 7-5). In some instances, inclusion of all levels of the spatial hierarchy could not be achieved, but where at least one of the random effect terms could be modelled, the outputs are presented.
Table 7-5 GLMM regression results for 22 species
The table shows the soil variables that could be included in univariate models for 22 species along with the slope coefficient or sign, and probability.
Species Soil Variable
Slope
Coeff Chi Prob
Eucalyptus coolab ah ln Total Organic Carbon 0-10 cm 3.366 <0.001
Eucalyptus coolab ah ln Salinity 0-10 cm -1.22 0.002
Eucalyptus coolab ah pH 40-50 cm -0.78 0.011
Eucalyptus coolab ah Texture 20-30 cm -0.26 0.002
Atriplex nummularia ln Total Organic Carbon 0-10 cm -1.67 0.013
Atriplex nummularia pH 20-30 cm -0.96 0.045
Chenopodium auricomum Texture 30-40 cm 0.261 0.034
Muehlenb eckia florulenta pH 0-10 cm -2.02 <0.001
Tecticornia indica ln Total Organic Carbon 0-10 cm 4.889 0.019
Tecticornia indica pH 0-10 cm 4.076 0.004
Acacia salicina ln Total Organic Carbon 0-10 cm 3.705 <0.001
Acacia salicina Texture 0-10 cm -0.44 0.008
Acacia stenophylla ln Total Organic Carbon 0-10 cm 2.804 <0.001
Sporob olus mitchellii pH 0-10 cm 2.163 0.004
Enchylaena tomentosa ln Total Organic Carbon 0-10 cm 1.919 <0.001
Enchylaena tomentosa Cation Exchange Capacity;0-10cm -0.08 0.016
Atriplex angulata ln Salinity 40-50 cm 0.413 0.012
Atriplex angulata Texture 0-10 cm 0.158 0.027
Atriplex leptocarpa ln Total Organic Carbon 0-10 cm 0.962 0.043
Bauhinia gilva ln Salinity 40-50 cm -0.84 0.029
Cressa cretica Carbonate 10-20 cm 1.475 <0.001
Cyperus gymnocaulos Cation Exchange Capacity 0-10 cm -0.12 0.017
Cyperus gymnocaulos Potassium Exchange 0-10 cm -3.34 0.017
Einadia nutans pH 0-10 cm -1.45 <0.001
Einadia nutans Texture 0-10 cm -0.17 0.005
Eremophila b ignoniiflora ln Total Organic Carbon 0-10 cm 2.099 <0.001
Maireana coronata ln Salinity 0-10 cm -0.72 0.025
Marsilea drummondii pH 40-50 cm 1.756 0.019
Marsilea drummondii Texture 0-10 cm 0.521 0.007
Osteocarpum acropterum ln Salinity4 0-50 cm 0.297 0.027
Sclerolaena diacantha ln Total Organic Carbon 0-10 cm -2.01 <0.001
Sclerolaena diacantha ln Salinity 40-50 cm -0.3 0.053
Sclerolaena intricata Cation Exchange Capacity 0-10 cm -0.05 0.014
Sclerolaena intricata ln Total Organic Carbon 0-10 cm -0.68 0.034
Plant species typical of riparian woodland communities, Eucalyptus coolabah, Acacia salicina, A. stenophylla, Eremophila bignoniiflora, Enchylaena tomentosa and Einadia nutans
were significantly correlated with high levels of organic carbon, while species typical of floodplain swamps, Chenopodium auricomum and Marsilea drummondii were significantly correlated with heavy textured soils (positive response to texture in Table 6-5). The occurrence of Muehlenbeckia florulenta was tightly correlated with higher levels of acidity (negative response to pH), while another prominent shrub in the region, Atriplex nummularia, was associated with soils having lower organic carbon and higher acidity. Another member of the Chenopodiaceae, typical of sandier portions of the outer floodplain (and lower slopes of fringing dunes), Sclerolaena diacantha, was significantly correlated with lower levels of organic carbon and lighter textured soils, as expected. The most abundant species of
Sclerolaena on the region‟s floodplains, S. intricata, was also significantly correlated with lower levels of organic carbon and cation exchange capacity. Species associated with saline soils included the dominant samphire, Tecticornia indica, and other chenopods such as Atriplex angulata and Osteocarpum acropterum. Curiously, another small chenopod, Maireana coronata, was negatively associated with salinity. It grew more frequently in lower-salinity soils than the region‟s average, in upper parts of the Cooper floodplain. Another character tree species in the region, Bauhinia gilva, restricted to the upper half of the study area, was also negatively associated with salinity. This species is usually observed growing on small sandy mounds alongside the river channels, tending to avoid the lower parts of the floodplain. Species significantly correlated with soils of higher alkalinity included the dominant grass on some lake beds, Sporobolus mitchellii, Marsilea drummondii (nardoo), Tecticornia indica (samphire) and a plant frequently found in rain-fed claypans, Solanum oligacanthum.
Graphical illustrations for a number of the relationships for the species and soils variables in Table 7-5 are contained in Appendix 21.
Rather than examine each of the models listed in Table 7-5 in greater detail, in the cause of economy of presentation a suitable sub-set will be considered. For this purpose the first five species listed in Table 7-5 will be considered. These five species appropriately represent the typical mesic to xeric gradient across the floodplain from channel to dunefield. This transition is schematically presented in Figure 7-28, showing five of the six sites located in sub-region J of the study area. Generally, as illustrated, both alkalinity and salinity increase across the floodplain from channel to dunefield. The actual disposition of all six sites in sub-region J is shown in Figure 7-29, juxtaposed with an aerial photograph of the area. Associated with each site label is the associated soil classification group to which it belongs (eg J1-1 soil group 5, J1- 2 soil group 1).
Figure 7-29 Site distribution in sub-region J
Aerial photograph shows the close proximity of the surrounding dunefields of the Strzelecki Desert with the floodplain