Results & Discussion
No strong relationship between CMI and detrended tree ring widths.
Figure 9. Average tree ring width for each species in each ecozone compared to average CMI for 1901 - 2019. (Google sites' compression isn't working well here, here's a link to the original figure)
Figure 9 compares the average detrended tree ring width of each species in each ecozone to the average CMI of that ecozone. As implied by the scatterplot comparing ring width and CMI in figure 8, there are very few cases where a low ring width corresponds to a similarly low CMI value. Some species share similar trends within ecozones, but their are more differences than similarities. This could be an effect of the different environmental adaptations between species (Hacke, Sperry, Wheeler, & Castro, 2006), or an effect of the distribution of sample locations shown in figure 5.
Table x. Two-sided paired t-test results of pre-drought to during-drought ring width by species and ecozone. P-values and negative mean differences (higher ring widths during drought) are highlighted in red.
Inconsistent growth responses to low CMI values.
Figure 10. Resilience indices for each species in each ecozone, with standard error.
Figure 10 shows the resistance (Rt), recovery (Rc), and resilience (Rs) index for each species in each ecozone. A resistance index above 1 implies that ring width increased during the drought event; this would mean that growth was not negatively affected by the low CMI value. A recovery index below 1 means that post-drought growth was lower than growth during the drought, indicating a continuing decline after the drought or some other factor that has a greater negative effect on growth. A resilience index above 1 indicates that post-drought growth surpassed pre-drought growth, which could mean that the negative effect of the drought did not have a lasting effect (if there was a negative effect). The expected pattern would be a Rt value below 1 that indicates the relative impact of the drought, a Rc value above 1 that shows a recovery in growth after the drought, and a Rs value below 1 that shows some lasting impact post-recovery; this was the general pattern seen in Lloret, Keeling, and Sala's study (2011). The responses represented in figure 10 show it is likely that drought was not properly identified.
Table 6. Confidence intervals and p-values from single sample t-tests for each resilience index of each species in each ecozone.
Limitations and possible improvements
The lack of consistency in ring widths over time when aggregated by ecozone indicates that another method would be more suitable for stratifying results. This can also be seen in the distribution of sample plots; many of the points are clustered and not evenly distributed throughout each ecozone. It might be more appropriate to group sample locations more clearly based on climatic conditions.
The predictor variable should have a more obvious effect on the response variable. There was no clear relationship between CMI and detrended ring widths. While there could be many reasons for this, one key reason could be the importance of timing in the relationship between moisture and tree growth. Not only were annual values used for CMI, but the CMI uses potential evapotranspiration. Since available moisture is not taken to account, actual drought timing could widely deviate due to the timing of precipitation relative to periods of high evapotranspiration. The timing of drought is also important; it would have more impact on growth during specific periods, which can depend on species (Gao, et al., 2018).
Furthermore, tree growth is a complicated process that can be affected by a variety of environmental factors, such as light (Porter, 1999), and heat (Wahid, Gelani, Ashraf, & Foolad, 2007). The effect of these factors should be considered as much as possible, as well as there own timings relative to growth. There are also a variety of factors that would be very difficult to include, but have a significant effect on annual variance; this would include factors like actual soil moisture (Martínez-Fernández, de Luis, González-Zamora, & Herrero-Jiménez, 2019), competition (Oboite & Comeau, 2020), mortality (DeSoto, et al., 2020), and other disturbances.