As we can see in figure 1.1, the higher the area the higher the number of species. In this regard the sampling done with a different method (0.25) lies reasonably withing the expected parameters
Figure 1.1: Species richness for each treatment grouped by the sampled area
A closer look at the curve that we see for each block separated by initial habitat shows us again that the number of species at 0.25 lies closely to what we would expect for a natural progression (figure 1.2).
Figure 1.2: Curve of acumulation of species with area by block number and treatment
For all plot sizes we fitted the following general equation as a generalized mixed effect linear model:
glmer(richness ~ initial_habitat + treatment + (1 | block_no), family = poisson)
For which we get the results shown in tables 1.1 and 1.2, there we can see that generally the \(R^2\) values are higher the larger the plot size is. Also in 1.2 we can see that the initial habitat is significant at all distances, after that, when we compare to permanent exclosure, summer exclosure is significant in 4 out of the 5 distances, winter and all year grazing are significant in 3 out of the 5 distances, and mowing is never shown to be significantly different from permanent exclosure. Finally the equations of all the final models are shown in the Equations section bellow
loglik | aic | df.residual | r2.conditional | r2.marginal | rmse | plot_size |
---|---|---|---|---|---|---|
-52.369 | 114.738 | 21 | NA | 0.315 | 1.761 | 0.01 |
-106.398 | 226.796 | 37 | 0.446 | 0.378 | 2.473 | 0.10 |
-107.996 | 229.991 | 37 | 0.479 | 0.396 | 2.384 | 0.25 |
-115.746 | 245.492 | 37 | 0.501 | 0.389 | 2.843 | 1.00 |
-127.691 | 269.382 | 37 | 0.613 | 0.468 | 3.631 | 10.00 |
term | estimate | std.error | statistic | p.value | plot_size |
---|---|---|---|---|---|
(Intercept) | 1.159 | 0.237 | 4.882 | 0.000 | 0.01 |
initial_habitatRangeland | 0.594 | 0.212 | 2.808 | 0.005 | 0.01 |
(Intercept) | 1.649 | 0.166 | 9.907 | 0.000 | 0.10 |
initial_habitatRangeland | 0.471 | 0.144 | 3.281 | 0.001 | 0.10 |
treatmentSummerExclosure | 0.405 | 0.159 | 2.553 | 0.011 | 0.10 |
treatmentWinterExclosure | 0.343 | 0.161 | 2.132 | 0.033 | 0.10 |
(Intercept) | 1.676 | 0.168 | 9.999 | 0.000 | 0.25 |
initial_habitatRangeland | 0.510 | 0.149 | 3.428 | 0.001 | 0.25 |
treatmentControl | 0.326 | 0.160 | 2.041 | 0.041 | 0.25 |
treatmentSummerExclosure | 0.357 | 0.156 | 2.291 | 0.022 | 0.25 |
(Intercept) | 2.154 | 0.140 | 15.334 | 0.000 | 1.00 |
initial_habitatRangeland | 0.402 | 0.129 | 3.123 | 0.002 | 1.00 |
treatmentControl | 0.302 | 0.132 | 2.289 | 0.022 | 1.00 |
treatmentSummerExclosure | 0.304 | 0.129 | 2.357 | 0.018 | 1.00 |
treatmentWinterExclosure | 0.318 | 0.129 | 2.473 | 0.013 | 1.00 |
(Intercept) | 2.634 | 0.120 | 21.877 | 0.000 | 10.00 |
initial_habitatRangeland | 0.385 | 0.117 | 3.304 | 0.001 | 10.00 |
treatmentControl | 0.333 | 0.104 | 3.209 | 0.001 | 10.00 |
treatmentSummerExclosure | 0.273 | 0.103 | 2.648 | 0.008 | 10.00 |
treatmentWinterExclosure | 0.348 | 0.101 | 3.432 | 0.001 | 10.00 |
\[ \begin{aligned} \operatorname{\widehat{richness}}_{i} &\sim \operatorname{Poisson}(\lambda_i) \\ \log(\lambda_i) &=1.16_{\alpha_{j[i]}} + 0.16_{\beta_{1}}(\operatorname{treatment}_{\operatorname{SummerExclosure}}) + 0.08_{\beta_{2}}(\operatorname{treatment}_{\operatorname{WinterExclosure}}) \\ \alpha_{j} &\sim N \left(0.59_{\gamma_{1}^{\alpha}}(\operatorname{initial\_habitat}_{\operatorname{Rangeland}}), 0 \right) \text{, for block\_no j = 1,} \dots \text{,J} \end{aligned} \]
\[ \begin{aligned} \operatorname{\widehat{richness}}_{i} &\sim \operatorname{Poisson}(\lambda_i) \\ \log(\lambda_i) &=1.65_{\alpha_{j[i]}} + 0.22_{\beta_{1}}(\operatorname{treatment}_{\operatorname{Control}}) + 0.11_{\beta_{2}}(\operatorname{treatment}_{\operatorname{Mowing}}) + 0.41_{\beta_{3}}(\operatorname{treatment}_{\operatorname{SummerExclosure}}) + 0.34_{\beta_{4}}(\operatorname{treatment}_{\operatorname{WinterExclosure}}) \\ \alpha_{j} &\sim N \left(0.47_{\gamma_{1}^{\alpha}}(\operatorname{initial\_habitat}_{\operatorname{Rangeland}}), 0.11 \right) \text{, for block\_no j = 1,} \dots \text{,J} \end{aligned} \]
\[ \begin{aligned} \operatorname{\widehat{richness}}_{i} &\sim \operatorname{Poisson}(\lambda_i) \\ \log(\lambda_i) &=1.68_{\alpha_{j[i]}} + 0.33_{\beta_{1}}(\operatorname{treatment}_{\operatorname{Control}}) + 0.12_{\beta_{2}}(\operatorname{treatment}_{\operatorname{Mowing}}) + 0.36_{\beta_{3}}(\operatorname{treatment}_{\operatorname{SummerExclosure}}) + 0.22_{\beta_{4}}(\operatorname{treatment}_{\operatorname{WinterExclosure}}) \\ \alpha_{j} &\sim N \left(0.51_{\gamma_{1}^{\alpha}}(\operatorname{initial\_habitat}_{\operatorname{Rangeland}}), 0.13 \right) \text{, for block\_no j = 1,} \dots \text{,J} \end{aligned} \]
\[ \begin{aligned} \operatorname{\widehat{richness}}_{i} &\sim \operatorname{Poisson}(\lambda_i) \\ \log(\lambda_i) &=2.15_{\alpha_{j[i]}} + 0.3_{\beta_{1}}(\operatorname{treatment}_{\operatorname{Control}}) + 0.13_{\beta_{2}}(\operatorname{treatment}_{\operatorname{Mowing}}) + 0.3_{\beta_{3}}(\operatorname{treatment}_{\operatorname{SummerExclosure}}) + 0.32_{\beta_{4}}(\operatorname{treatment}_{\operatorname{WinterExclosure}}) \\ \alpha_{j} &\sim N \left(0.4_{\gamma_{1}^{\alpha}}(\operatorname{initial\_habitat}_{\operatorname{Rangeland}}), 0.12 \right) \text{, for block\_no j = 1,} \dots \text{,J} \end{aligned} \]
\[ \begin{aligned} \operatorname{\widehat{richness}}_{i} &\sim \operatorname{Poisson}(\lambda_i) \\ \log(\lambda_i) &=2.63_{\alpha_{j[i]}} + 0.33_{\beta_{1}}(\operatorname{treatment}_{\operatorname{Control}}) + 0.13_{\beta_{2}}(\operatorname{treatment}_{\operatorname{Mowing}}) + 0.27_{\beta_{3}}(\operatorname{treatment}_{\operatorname{SummerExclosure}}) + 0.35_{\beta_{4}}(\operatorname{treatment}_{\operatorname{WinterExclosure}}) \\ \alpha_{j} &\sim N \left(0.39_{\gamma_{1}^{\alpha}}(\operatorname{initial\_habitat}_{\operatorname{Rangeland}}), 0.13 \right) \text{, for block\_no j = 1,} \dots \text{,J} \end{aligned} \]