The Impact of Hydration and Temperature on Bacterial Diversity in Arid Soil Mesocosms

Hot desert ecosystems experience rare and unpredictable rainfall events that resuscitate the arid flora and fauna. However, the effect of this sudden abundance of water on soil microbial communities is still under debate. We modeled varying rainfall amounts and temperatures in desert soil mesocosms and monitored the microbial community response over a period of 21 days. We studied two different wetting events, simulating heavy (50 mm) and light (10 mm) rain, as well as three different temperature regimes: constant 25° or 36°C, or a temperature diurnal cycle alternating between 36 and 10 °C. Amplicon sequencing of the bacterial ribosomal RNA revealed that rain intensity affects the soil bacterial community, but the effects are mitigated by temperature. The combination of water-pulse intensity with lower temperature had the greatest effect on the bacterial community. These experiments demonstrated that the soil microbial response to rain events is dependent not only on the intensity of the water pulse but also on the ambient temperature, thus emphasizing the complexity of bacterial responses to highly unpredictable environments.


Loading dataset
Replicate 1 and 2 from each dataset were loaded for a comparison between the dataset of unequal sizes. The dataset consists of soil chemical properties such as electric conductivity, pH, water content, ammonium, nitrite and nitrate as explanatory variables and the species richness as a response variable.

Null model
The null model was constructed to verify that the dataset contains true replicates that therefore don't need to be included in the model. The replicates do not explain any variance and therefore will not be considered in the following models.

Model including the entire dataset
A model including soil chemical properties such as electric conductivity, pH, water content, ammonium, nitrite and nitrate as explanatory variables and the species richness as a response variable.
Generalized linear mixed-effects model: Eveness The null hypothesis can not be rejected and therefore it's assumed that the simplified model is not statistically different from the original model and therefore it will be retained.

Simplified model assumptions
Variance inflation factors and multicolinearity ## TimeValue WC EC ## 1.520555 1.652899 1.109005 The variance inflation factors of all the variables were found satisfactory.

Q-Q plot
Final results In order to run the document, a package rmarkdown needs to be loaded and the file is prepared with the command render .

Loading dataset
Since qPCR was measured in two technical replicates, an average of the two was calculated prior to the data comparison. The electric conductivity of the soil was not included in the dataset, since too many values in that dataset were missing.

Total bacteria ribosomal count
Null model The null model was constructed to verify that the dataset contains true replicates that therefore don't need to be included in the model. The replicates do not explain any variance and therefore will not be considered in the following models.

Model including the entire dataset
A model including soil chemical properties such as electric conductivity, pH, water content, ammonia, nitrite and nitrate as explanatory variables and the species richness as a response variable. The null hypothesis can not be rejected and therefore it's assumed that the simplified model is not statistically different from the original model and therefore it will be retained. The variance inflation factors of all the variables were found satisfactory.

Q-Q plot
Final results Actinobacteria ribosomal count

Null model
The null model was constructed to verify that the dataset contains true replicates that therefore don't need to be included in the model. ## summary from lme4 is returned ## some computational error has occurred in lmerTest The replicates do not explain any variance and therefore will not be considered in the following models.

Model including the entire dataset
A model including soil chemical properties such as electric conductivity, pH, water content, ammonia, nitrite and nitrate as explanatory variables and the species richness as a response variable. The null hypothesis can not be rejected and therefore it's assumed that the simplified model is not statistically different from the original model and therefore it will be retained.

Simplified model assumptions
Variance inflation factors and multicolinearity ## N.NH4 WC ## 1.00024 1.00024 The variance inflation factors of all the variables were found satisfactory. The replicates do not explain any variance and therefore will not be considered in the following models.

Model including the entire dataset
A model including soil chemical properties such as electric conductivity, pH, water content, ammonia, nitrite and nitrate as explanatory variables and the species richness as a response variable. The null hypothesis can not be rejected and therefore it's assumed that the simplified model is not statistically different from the original model and therefore it will be retained.

Loading dataset
Replicate 1 and 2 from each dataset were loaded for a comparison between the dataset of unequal sizes. The dataset consists of soil chemical properties such as electric conductivity, pH, water content, ammonium, nitrite and nitrate as explanatory variables and the species richness as a response variable.

Null model
The null model was constructed to verify that the dataset contains true replicates that therefore don't need to be included in the model. The replicates do not explain any variance and therefore will not be considered in the following models.

Model including the entire dataset
A model including soil chemical properties such as electric conductivity, pH, water content, ammonium, nitrite and nitrate as explanatory variables and the species richness as a response variable. The null hypothesis can not be rejected and therefore it's assumed that the simplified model is not statistically different from the original model and therefore it will be retained. The variance inflation factors of all the variables were found satisfactory.

Loading dataset
Replicate 1 and 2 from each dataset were loaded for a comparison between the dataset of unequal sizes. The dataset consists of soil chemical properties such as electrical conductivity, pH, water content, ammonium, nitrite and nitrate as explanatory variables and the species richness as a response variable.

Centering of variables and generating z-scores
In order to account for different dimensions of variables, they are centered transformed into z-scores. Centering was performed as subtracting variable means from its values and scaling was achieved by diving by variable's standard deviation. The null hypothesis can be rejected and therefore the model is assumed significantly different from the full model. The environmental variable with the next highest significance is added to the model, in this cas it is the time since the start of the the analysis.
ord3 <-cca(ani.otu ~ Water.content + Ammonium.nitrogen + Time.value + Electrical.conduc tivity, data=df4) anova(ord, ord3) The null hypothesis can not be rejected and therefore it's assumed that the simplified model is not statistically different from the original model and therefore it will be retained. It should be noted, that the CCA1 eigenvalue is much higher (0.3315) then the CCA2 eigenvalue (0.0356). This can be interpreted as CCA1 explaining proportionaly ~ 9 times more variance then CCA2.

Correspondence analysis assumptions
It should be noted, that the CCA1 eigenvalue is much higher (0.3315) that the CCA2 eigenvalue (0.0356). This can be interpreted as CCA1 explaining proportionally ~ 9 times more variance that CCA2.
It should be noted, that the CCA1 eigenvalue is much higher (0.3315) that the CCA2 eigenvalue (0.0356). This can be interpreted as CCA1 explaining proportionally ~ 9 times more variance that CCA2.