Abstract
Time series data offer a way of investigating the causes driving ecological processes as phenomena. To test for possible differences in water relations between species of different forest structural guilds at Danum (Sabah, NE Borneo), daily stem girth increments (gthi), of 18 trees across six species were regressed individually on soil moisture potential (SMP) and temperature (TEMP), accounting for temporal autocorrelation (in GLS-arima models), and compared between a wet and a dry period. The best-fitting significant variables were SMP the day before and TEMP the same day. The first resulted in a mix of positive and negative coefficients, the second largely positive ones. An adjustment for dry-period showers was applied. Interactions were stronger in dry than wet period. Negative relationships for overstorey trees can be interpreted in a reversed causal sense: fast transporting stems depleted soil water and lowered SMP. Positive relationships for understorey trees meant they took up most water at high SMP. The unexpected negative relationships for these small trees may have been due to their roots accessing deeper water supplies (if SMP was inversely related to that of the surface layer), and this was influenced by competition with larger neighbour trees. A tree-soil flux dynamics manifold may have been operating. Patterns of mean diurnal girth variation were more consistent among species, and time-series coefficients were negatively related to their maxima. Expected differences in response to SMP in the wet and dry periods did not clearly support a previous hypothesis differentiating drought and non-drought tolerant understorey guilds. Trees within species showed highly individual responses when tree size was standardized. Data on individual root systems and SMP at several depths are needed to get closer to the mechanisms that underlie the tree-soil water phenomena in these tropical forests. Neighborhood stochasticity importantly creates varying local environments experienced by individual trees.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
The main changes to version 3 were to include data analyses which tested whether (a) showers within the dry period had affected the recording of daily stem girth increment (gthi), by rewetting bark under the dendrobands (this causing possible temporary expansion), and (b) that decreases in relative humidity of the air in the forest understorey around bands in the wet period may have temporarily altered gthi in the opposite way (through possible bark shrinkage). In the first case, regression modelling of gthi on daily rainfall did show some effect at higher intensities for some but not all trees. Therefore, the regression residuals were used in place of original gthi-values. In the second case, there was no evidence of any bias having been introduced. All of the time series statistical analyses for the dry period were redone; and also so were the comparisons with the wet period, the within-day change analysis, and the relationships between fitted coefficients to SMP and TEMP and diurnal parameters. The end result was very little change to the main results and conclusions, apart for some now stronger evidence of some cases of a SMP x TEMP interaction in the dry period. A more robust analysis was nevertheless achieved to counter criticisms of the use of dendrobands under tropical conditions, with recommendation of a new data analysis protocol. Version 4 incorporates two new subsections to each of Methods and Results, corresponding to a and b and with two new appendices and an extended Discussion, improvements to the Figures in Appendix 1, moving of the text on the physical sensitivity of bands to temperature change to an appendix, further analysis of effects of nearest neighbour abundance on fitted water flux parameters (competition effects), and several minor improvements to the main text throughout. The title has been revised.
https://datadryad.org/stash/share/nUvbdXLPOaRCfk09JZJrjonG1O36r0_RvMc4CcAZVLs
