Percepciones en la vida diaria del paciente con poliquistosis renal autosómica dominante
los 53 años, con ERC estadio II, IIIA,IIIB siendo los componentes tres varones y dos mujeres
11.6.1 Initial Data Exploration
All parameters were first tabulated with means, standard deviations and standard errors calculated. Bar charts and line graphs were produced in Excel (Microsoft, 2013) (version 15.0.4551.1005) to identify initial differences between ecotypes and treatments.
Data were then saved as CSV files for use in R statistical software version 3.1.0 (R Core Team, 2013). Histograms were created using Rcmdr version 2.0-4 (Fox & Bouchet-Valat, 2014) to help determine the distribution of each data set; ANOVA (for parametric data) and Kruskall-Wallis (for non-parametric data) was also used in this program to determine initial significant differences. Scattergraph matrices and boxplots were created using RStudio (RStudio, 2013) (version 0.98.994) to identify similarities and correlation between datasets.
11.6.2 Generalized Linear Mixed Model (GLMM)
Generalized linear mixed models (GLMM) were used to statistically predict ecotypic distributions with specific sub-sets of response variables, as this statistical technique is good for species-specific models (Guisan et al., 1999). GLMM tuition and written aids were used in the application and interpretation of the models in R Studio (RStudio, 2013) (version 0.98.994) (Field et al., 2009; Zuur et al., 2009; Winter, 2013; Smith, 2014). It was originally hoped that one model could be used (with the response variable changed for each dataset). However, due to the high number of variables and interactions, the sample size prevented a full model analysis (Smith, 2014). Partial models were therefore processed and resulting effects plots along with the Akaike Information Criterion (AIC) and P values from ANOVA comparisons values used to determine the most important factors for the final model (Winter, 2013, Smith, 2014). The standard used in all models was Cockey Down (a
calcareous loam/grazed ecotype), in calcareous loam and grazed treatment. The standard was picked to represent the [presumed] best combination for Lotus corniculatus, a calcareous loam ecotype was chosen as calcareous
pastures represent the most common habitat for the plant (Grime et al., 1992), and grazed treatment selected as this management [or cutting with clippings removed] is recommended in Natural England guidance after species-rich grassland creation (Natural England, 2010).
For the first model (Model A) it was hoped to include each ecotype, the treatment soil and management treatment. The following are the partial models used with each dataset (the bold text is the response variable which was changed to that being investigated at the time of each analysis). The most suitable combination for model A was selected for each response variable (Table 16).
Model A.1:
<-lmer(response_variable~EcotypeSeed+soil+management+(1|f.replic), data=response_variable)
This model uses the factors of ecotype seed (EcotypeSeed), soil treatment (soil) and management treatment (management) as fixed effects, it blocks them by replicate to eliminate any glasshouse effect.
Model A.2:
<-lmer(response_variable~EcotypeSeed+soil+(1|management/f.replic)data= response_variable)
This model variation differs in that it uses management treatment and replication as a nested random effect.
Model A.3:
<-lmer (grazed_biomass~seed+soil+(1|f.replic),data=grazed_biomass) Model A.3 was only used on grazed treatment clippings biomass due to this part of the experiment only involving the grazed treatment plants.
To determine significance P values of factors the Likelihood Ratio Test was conducted (Winter, 2013). The chosen (A, B and C) model was repeated,
each time with one of the factors removed (the null model). ANOVA was calculated for the comparison between the full model and null model each time. AIC and P values for each comparison indicated which factors were significant and accounted for most effect in the model (Winter, 2013).
As the largest variation in the data was seen to be the management treatment it made sense to repeat those models separately between the two
managements (therefore the datasets were split between managements ‘Model B.1 and B.2’).
Model B.1:
<-lmer (response_variable~seed+soil+(1|f.replic), data=response_variable. grazed)
Model B.2:
<-lmer (response_variable~seed+soil+(1|f.replic), data=response_variable. unmanaged)
The last model was intended to look at any interaction between ecotype with treatment and also look at donor site soil and management effects. When the dataset had already been split by management, the split was carried out again for this model (Model C.1 and C.2), when there was no management split the dataset was left intact (Model C.3).
Model C.1: <-lmer(response_variable~EcotypeSoil+soil+EcotypeMgmt+EcotypeSoil *Soil+ (1/f.replic),data=reponse_variable.grazed) Model C.2: <-lmer(response_variable~EcotypeSoil+soil+EcotypeMgmt+EcotypeSoil *Soil+ (1/f.replic),data=reponse_variable.unmanaged)
Model C.3:
<-lmer(response_variable~EcotypeSoil+soil+EcotypeMgmt+management+ +EcotypeSoil*Soil+EcotypeMgmt*management+(1/f.replic),data=reponse _variable)
As before, these models were repeated with ANOVA comparison of null (partial) models to establish P value significance of factors.
All of the above models are calculated using a Gaussian distribution. For those which needed Poisson distribution due to the nature of the data (count data or non-parametric data), ‘family = poisson’ was entered into the model equation before ‘data=’.
The following table (Table 16) outlines which models were used for each response variable.
Table 16. Models used with each response variable. Refers to model descriptions in 'Generalized Linear Mixed Models' and ‘Spatial Autocorrelation’ in Chapter 11.
Response Variable GLMM (lme4) models used Spatial
Autocorrelation GLMM (nlme) models used
Main Stem Lengths (Harvest) A.1, B.2, B.3, C.1, C.2 D.1, D.2 Main Stem Lengths (Soil+Mgmt) A.1, B.2, B.3, C.1, C.2 N/A
Leaflets per main stem (Harvest) A.2, B.2, B.3, C.1, C.2 D.1, D.2 Stems per plant (Harvest) A.2, B.2, B.3, C.1, C.2 D.1, D.2
Hirsuteness A.1, C.3 D.3
Time taken for seed pod formation A.1, B.1, B.2, C.1, C.2 D.1, D.2 Seed pod number A.1, B.1, B.2, C.1, C.2 D.1, D.2 Seed pods sampled A.1, B.1, B.2, C.1, C.2 N/A
Mean seeds per pod A.1, B.1, B.2, C.1, C.2 D.1, D.2 Grazed treatment clippings dry biomass A.3, C.3 D.3 Harvest dry biomass (vegetation) A.2, C.3 D.3 Harvest relative moisture content (%) of
biomass (vegetation)
A.1, B.2, B.3, C.1, C.2 D.1, D.2
Hydrogen Cyanide (HCN) A.1, C.3 D.3
Nitrogen A.1, B.2, B.3, C.1, C.2 D.1, D.2
Total flower number A.2, C.3 D.3
11.6.3 Spatial Autocorrelation
The mixed modelling package NLME (Non-linear Mixed Effects) in RStudio (2013) (version 0.98.994) was used to look at the spatial autocorrelation between ecotype sites. Additional columns of OS Grid Northing and Easting data were entered into each dataset. The model was split between
management treatments again when relevant. Model D.1:
<-lme(response_variable~EcotypeSeed+Soil+Management,data= response_ variable.unmanaged, random = ~1|f.replic)
Model D.2:
<-lme(response_variable~EcotypeSeed+Soil+Management,data=na.omit (response_variable. grazed),random= ~1|f.replic)
These two models split the data between management treatment, Model D.2 also has the addition of ‘na.omit’ to allow data with NA (zero) values to be used by ignoring them.
Model D.3:
<-lme(response_variable~EcotypeSeed+Soil+Management,data= response_ variable, random = ~1|f.replic)
This model was used when the management split was not needed.
A semi-variogram was then created for each, to show the effect which spatial proximity between ecotype sites had on the response variable, these model graphs were built and interpreted with tuition and written aids (Neilson and Wendroth, 2003; Winter, 2013; Smith, 2014). To establish whether spatial distance between sites was a significant factor in the models, an ANOVA test was completed each time, to compare the model containing the spatial factor to the model without. The AIC and P values were then used to see which model showed the best fit and if there was significant difference with the
spatial factor. Table 16 indicates which models were used for each response variable.
Spatial Results
The variograms generated can be found in Appendix XI, along with AIC
numbers and ANOVA P values between the null model and the spatial model. As the ANOVA AIC numbers and P values ruled out any significant effect on the models from spatial distance between ecotype sites they have been removed from the results write-up. The pattern shown in each variogram also did not indicate any spatial autocorrelation was influencing these parameters apart from for main stem length. Therefore a small paragraph has been kept in this sub-chapter.
12 RESULTS (STUDIES 1 & 2)
Please note that within this chapter, calcareous sand is referred to as ‘sand’ and cut with aftermath grazing is referred to as ‘cut’. ‘Grazing treatment’ refers to simulated grazing from cutting by hand.
Once the seedlings had been potted on, the experimental design detailed in chapter 11 was implemented and parameters monitored for use in identifying plant fitness and/or changes in relation to herbivore nutrition in relation to treatments. This chapter displays the growing and harvest results of the main experiment of this thesis.
Table 17 lists abbreviations used in results for ecotypes and treatments and should be used in conjuction with subsequent tables and figures.
Table 17. Key to ecotype references used in results
Ref. Ecotype Name Ecotype Management Ecotype Soil
cd Cockey Down Grazed Calcareous loam
ss Southstoke Cut [with aftermath grazing] Calcareous loam wb Woodborough Cut [with aftermath grazing] Calcareous loam ff Folly Farm Cut [with aftermath grazing] Neutral loam
hh Hellenge Hill Grazed Neutral loam
sp Salisbury Plain Unmanaged Neutral loam
bd Berrow Dunes Unmanaged [Calcareous] sand
ww Woolacombe Warren Unmanaged [Calcareous] sand
dw Dawlish Warren Grazed [Calcareous] sand
Ref Treatments
C Calcareous loam treatment N Neutral loam treatment S [Calcareous] sand treatment G Grazed treatment