Principal component analysis

PCA1 - Raw, Sea-salt corrected and with covariates Soil Water

The soil water data, whether raw or sea-salt corrected soil water, show similar patterns of loadings across the species in the PCAs (Table 2.18a). The first three components explain 67% and 70% of the raw and sea-salt corrected datasets respectively. The first principal component, PC1, shows high loadings for Ca, K, Mg, Na and PO4, elements that showed differences

between burning regimes in the ANOVA. The second principal component (PC2) shows high loadings for Al and Fe. Finally, the third component (PC3) has high loadings for SO4 and Cl. With the addition of PO4 in PC1 and Cl in

PC3, these results broadly reflect similar trends shown in Worrall and Adamson (2008). When pH, conductivity and DOC are included, the first principal component has high loading for Ca, K, Mg, Na, PO4 and

Fe suggesting a shallower soil water component. The third component has low loading for pH and conductivity and high loadings for Cl and SO4. A high

loading of DOC in the fourth principal component could explain the lack of a high loading for DOC in PC2.

Runoff Water

The runoff water dataset shows different results in the PCAs when compared to soil water data (Table 2.18b). The raw and sea-salt corrected data show similar magnitude effects though the direction of the effects is not always the same. The first three components explained 73% and 76% of the variation in the raw and sea-salt corrected datasets respectively. The first principal component (PC1) is relatively evenly weighted except for NO3 that is

associated with PC3. For PC1, Al and Fe show high positive loadings in comparison to the negative loading for other species. The second principal component shows high loadings for K, Mg, and Na. The third principal component is dominated by a large loading for NO3; Na and Cl also have

high values. Again, further analysis of the data is possible when pH, conductivity and DOC are included. The first component has positive

loading for Al, Fe and also DOC The second principal component has strong negative loading for K and Mg and low loadings for pH and conductivity. The strong loading for K and Mg, terrestrial derived species, could indicate a source deeper in the peat profile for this component. The third component

conductivity. The strong sea-salt component (Na and Cl) suggests a rainwater influence to this component.

a) Soil Water b) Runoff Water

Species PC1 PC2 PC3 PC4 PC1 PC2 PC3 PC4 a) Al 0.007 0.591 -0.017 0.394 0.326 -0.052 Ca 0.392 -0.260 0.330 -0.390 0.132 -0.311 Fe 0.192 0.559 0.106 0.353 0.377 -0.039 K 0.404 -0.001 -0.261 -0.277 0.449 -0.143 Mg 0.480 0.037 0.178 -0.307 0.493 -0.105 Na 0.471 0.221 -0.034 -0.281 0.400 0.325 Cl 0.186 -0.168 -0.600 -0.322 -0.173 0.363 NO3 0.040 0.220 0.324 -0.051 0.051 0.741 PO4 0.387 -0.325 0.086 -0.218 -0.260 -0.278 SO4 0.086 0.194 -0.553 -0.405 -0.170 -0.067 Variance Explained 0.359 0.546 0.667 0.368 0.613 0.732 b) Al 0.046 0.577 -0.040 0.338 0.400 0.107 Ca 0.415 -0.317 0.068 -0.487 0.017 0.126 Fe 0.236 0.520 0.119 0.301 0.436 0.123 K 0.378 -0.040 -0.113 -0.379 0.343 0.116 Mg 0.507 0.000 -0.009 -0.383 0.429 0.078 Na 0.471 0.237 -0.019 -0.233 0.466 -0.181 NO3 0.056 0.201 0.721 -0.033 0.049 -0.926 PO4 0.375 -0.374 0.005 -0.213 -0.290 0.202 SO4 0.061 0.240 -0.668 -0.409 -0.208 -0.080 Variance Explained 0.371 0.583 0.700 0.335 0.639 0.760 c) Al 0.041 0.560 0.046 0.175 0.345 -0.330 -0.163 0.016 Ca -0.372 -0.200 0.298 0.083 -0.356 -0.163 -0.110 -0.287 Fe -0.135 0.520 0.038 -0.301 0.313 -0.376 -0.058 0.161 K -0.361 0.096 -0.265 -0.011 -0.253 -0.457 -0.037 0.047 Mg -0.417 0.112 0.177 0.050 -0.284 -0.491 0.011 0.032 Na -0.389 0.291 -0.032 -0.105 -0.240 -0.379 0.367 0.137 Cl -0.129 -0.152 -0.602 -0.179 -0.250 0.182 0.455 0.234 NO3 -0.042 0.247 0.304 -0.331 -0.027 -0.031 0.460 -0.335 PO4 -0.355 -0.246 0.175 0.202 -0.217 0.247 -0.215 0.356 SO4 -0.048 0.191 -0.488 0.337 -0.371 0.154 -0.054 -0.306 pH -0.288 -0.033 -0.278 -0.260 -0.355 0.076 -0.154 -0.074 Conductivity -0.395 -0.163 0.066 0.204 -0.180 -0.070 -0.567 -0.158 DOC 0.023 0.247 0.000 0.673 0.217 -0.013 0.092 -0.672 Variance Explained 0.342 0.501 0.605 0.699 0.332 0.523 0.644 0.722

Table 2.18. The loadings on the principal components for PCA for a) raw data, b) sea-salt corrected data and c) with pH, conductivity and

PCA2 – 10 year plots before and after managed burning

When the data from Hard Hill are combined with the ECN rainwater data, a comparison of PC1 versus PC2 shows a clear pattern of behaviour (Figure 2.11). -12 -10 -8 -6 -4 -2 0 2 4 6 -4 -2 0 2 4 6 8 10 12 14 16

Soil Water, unburnt Soil water, 10 year burn Soil water, 20 year burnt Runoff water, unburnt Runoff water, 10 year burn Runoff water, 20 year burn Rainwater

Figure 2.11. Comparison of PC1 and PC2 for sea-salt corrected data from Hard Hill and ECN precipitation data.

The majority of the data are bound by two trends; one formed from rainwater samples and the other soil water samples from 20-year burn plots. Water tables on the 20-year plots are closest to the surface on these sites so this latter trend can be interpreted as a shallower water trend. Runoff water samples occur dominantly, but not exclusively, along the rainwater trend.

Changes in water chemistry following the managed burn are able to help trace water sources contributing to soil water and runoff water. Though there is scatter in the data, soil water following managed burning shows a rotation towards more shallow water dominated trends (Figure 2.12).

y = -1.3484x - 2.7198 y = 0.9295x + 1.3666 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 Rainwater

Soil water, Pre-burn

Soil water, Post-burn

Figure 2.12. Plot of PC1 and PC2 for soil water from 10 year plots, pre- burn and post-burn

Runoff samples generally occur along the rainwater trend (Figure 2.10); however, a proportion of the pre-burn runoff water on the 10-year plots has a component associated the shallow water trend (Figure 2.13). Following the managed burn, runoff water on the 10-year plots is almost exclusively along

“Rainwater” trend “Shallow water”

y = -1.3484x - 2.7198 y = 0.9295x + 1.3666 -5 -4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 Rainwater

Runoff water, Pre-burn Runoff water, Post-burn

Figure 2.13. Plot of PC1 and PC2 for runoff water from 10 year plots, pre-burn and post-burn

Un-mixing PCA2 trends

The axes of theses trends are almost perpendicular suggesting that the behaviour of shallow soil water is independent from rainwater. By

referencing the co-ordinates of the data to these new axes, changes before and after the managed burn can be quantified. The angles between the co- ordinates and the two axes were calculated. In order to identify any

significant rotations following burning differences in the angles before and

“Shallow water” trend

changes before and after managed burning with rainwater samples lying close to the rainwater trend (Table 2.19).

Table 2.19. Average angle between co-ordinate and new axes (± standard error)

In the soil water data, a significant (p<0.05) change in the angle to soil water axis exists. This rotation towards the soil water axis can be seen by an approximate 8° change in angles between axes. Runoff samples show a significant change in the angle to the rainwater axis. This change can be seen by a rotation towards the rainwater axis with an approximately 20° rotation away from the 20-year axis.

Average angle to "Rainwater" axis Average angle to "20 year" axis Water

Type Rainwater Water Soil Runoff Water Rainwater Water Soil Runoff Water Pre-burn 21.6 (1.8) 68.0 (1.1) 36.9 (4.1) 72.9 (1.6) 26.4 (1.1) 57.4 (3.8) Post-burn 23.9 (2.0) 76.2 (1.2) 10.2 (1.7) 71.9 (2.0) 17.4 (1.3) 78.6 (1.4)

2.5 Discussion