Climate and local water availability are major evolutionary drivers of adaptive variation and plasticity in the hydraulic architecture of Tropical Montane Cloud Forest (TMCF) tree species. Between-year xylem vessel variability is key to understanding the adaptation potential of wood anatomy of trees to drought. How wood anatomical features have been influenced by the typical TMCF climate and how tree species persist in these environments remain open questions, particularly in the context of predicted extreme climate events in the future.

Here, we evaluated the effects of changes in temperature, precipitation, and evapotranspiration during drought events on ring-width and anatomical vessel traits (hydraulic diameter, vessel density, vessel grouping index, and vulnerability index) for two relict endemic and threatened oak species (Quercus delgadoana and Q. meavei) from a Mexican TMCF. The study species differed in their functional and ecological vessel anatomical traits, and their wood anatomical differences are related to specific environmental requirements. However, the Ring Width Indices (RWI) calculated for these species indicate that both have high resistance and recovery, and thus high resilience to drought events. Ring-width and vessel functional traits show differences in the between-year variability of xylem traits associated with the hydraulic efficiency of these oak species, which is crucial to understanding how they avoid drought-induced embolism and cavitation in vessel conduits. These results provide evidence for the existence of specific hydraulic systems that determine functional wood anatomy in response to climatic variation and drought in the study species. Further research assessing the wood anatomical adaptation to different climatic variables and identifying the xylem functional traits that underlie these adaptations, along with the mechanism allowing tree species persistence in these environments, is essential to gain insight into the responses of TMCF to future drought events.

Sampling techniques and chronology development

We collected wood cores from 20 trees from each oak species with a diameter at breast height (DBH; 1.3 m) ≥ 40 cm. We collected two cores per tree at breast height with an increment borer of 5 mm inner diameter (Häglof®, Langsele, Sweden) (Stokes & Smiley, 1996). We filled the holes with wood plugs treated with a mixture of 80% ethanol and 20% purified water (Thiercelin et al., 1972). The wood cores were air-dried at room temperature, glued to wooden supports, and sanded with a succession of coarse grit sandpapers (400, 600, 1000, 1200, and 2000) until the xylem cell structure was visible up to 100x magnified (Speer, 2010; Stokes & Smiley, 1996). We removed wood dust inside the lumina with a hairdryer (Rodríguez-Ramírez et al., 2018). Permission to collect the wood samples was granted directly by the Commissioner of the Medio Monte Natural Protected Area. Wood cores are stored at the Biogeography and Systematics Laboratory, Faculty of Science, Universidad Nacional Autónoma de México (UNAM), Mexico City, Mexico.

Ring widths were measured to the nearest 0.001 mm using a stereoscopic microscope (Olympus® SZ61, Olympus Corporation, Center Valley, PA, USA) with TSAP-Win v.4.67c software and a Velmex Tree Ring Measuring System (Velmex, Inc., Bloomfield, NY, USA). We statistically verified the dating accuracy through the COFECHA software (Holmes, 1983). To minimise the non-climatic variance associated with local climatic variations, and thus maximise the evidence for drought, we removed non-climatic trends from each tree-ring series using a cubic spline with a 50% response over 10-year periods (Gareca et al., 2010; Rodríguez-Ramírez et al., 2020).

We standardised raw ring-width series by applying autoregressive modelling to remove serial correlation using the ARSTAN software to calculate the mean value of detrended Ring-Width Indices (RWI; a standardised index for annual radial growth; Cook & Holmes, 1996). We evaluated the quality of the chronologies using EPS (Expressed Population Signal) as a measure of the total signal present in the chronology, considering values > 0.85 (Briffa, 1999) and the mean correlation coefficient between tree-ring series (R-bar; Wigley et al., 1984). In addition, we evaluated the signal-to-noise ratio (SNR), which expresses the strength of the recognized common signal between trees in a chronology.

Climate-growth correlations

We retrieved climatic data from the CLICOM database of the National Meteorological Service of Mexico (http://clicom-mex.cicese.mx/). We extracted the mean monthly maximum temperature (Tmax), total monthly precipitation (Prec), and total monthly evapotranspiration (EvT) data for a nearby weather station (Agua Blanca, 20.3° N, 98.4° W; 2157 m asl), which kept records from 1974 to 2016. We used Spearman's correlation coefficients (rS; because the climatic variables tend to change simultaneously; Pouteau et al., 2018) as a similarity measure to assess growth sensitivity to climate in the study species. We used the standard chronology and monthly climate data for Tmax, Prec, and EvT for a period spanning from the previous growth year (Jun (-1)) to the current growth year (Sep). We conducted this analysis with SigmaStat v.4 (Jandel Scientific, 2016).

Identification of historical drought events

To select tree rings formed during drought events for subsequent digitization (Rodríguez-Ramírez & Luna-Vega, 2020), we used specific historical drought years (1976, 1983, 1991, 1999, and 2012; directly from the Agua Blanca weather station). Droughts indicated by these data were confirmed using information from the Mexican Drought Atlas (Stahle et al., 2016). The chronologies were compared with monthly values of the Palmer Drought Severity Index (PDSI) to estimate drought intensity. The relationship of narrow rings with historical droughts allowed the identification of specific drought events (through the calculation of z-scores; Eq. 1) in the RWI for both species (1976, 1983, 1991, 1999 and 2012).

            We calculated the z-score for each oak chronology as: z_t= (X_t - Xbar)/S              (1)

where z is the standardised value of the drought year, xt is the reconstructed drought ranking, Xbar is the mean of the reconstructed chronology, and s is the standard deviation of the reconstructed chronology. By plotting the z-score and the ≤ 5th percentile, we identified the historical drought years estimated to have been significantly mild or harsh (Jones & Hulme, 1996).

2.5 Effect of drought events on tree growth

We performed a Superposed Epoch Analysis (SEA; Mooney et al., 1993) to examine the effects of the five historical drought years on the RWI related to the growth of the two oaks. SEA links the RWI time series to a series of drought events. For each drought event, a 5-year window was considered, including 2 years before and 2 years after the year of the event. The 5-year windows for all events were superimposed and averaged to obtain the mean pattern of RWI associated with drought events. The mean RWI pattern for the selected years was statistically tested for significance (95% confidence interval) by running 1000 Monte Carlo simulations (Mooney et al., 1993) using random years from the RWI dataset. We performed these analyses in R using the dplr package (Bunn, 2008).

Next, to determine the effect of historical drought on RWI for each species, we assessed three sensitivity indices, namely resistance (Rt; Eq. 2), recovery (Rc; Eq. 3), and resilience (Rs; Eq. 4), based on Lloret et al. (2011), as follows:

Resistance (Rt) = Ring width_t / Ring width_t-2      (2)

Recovery (Rc)= Ring width_t+2 / Ring width_t        (3)

Resilience (Rs) = Ring width_t+2 / Ring width_t-2  (4)

where Ring width is the growth width of the annual ring in year t, Ring width-2 is the mean ring width for the 2 years preceding year t, and Ring widtht+2 is the mean ring width for the 2 years following year t (Anderegg & Meinzer, 2015). We calculated the indices at the individual tree level for each year during the 1930-2012 period. Finally, we used ANOVA and post-hoc Tukey's test to compare the means of these indicators between year type (drought vs. non-drought) and species. These analyses were performed using BoxPlotR: a web tool for creating box plots (http://shiny.chemgrid.org/boxplotr/).

2.7 Tree-ring digitalization and measurement of xylem anatomical traits

To evaluate the drought effect on vessel traits, we randomly selected 10 cores for each oak species to obtain digital wood core images for vessel traits measurements. We first prepared the wood cores using the finest grit (2500, Wetordry™). Digital wood core images were taken using a stereoscopic microscope (Leica Z16 APOA) with a depth of field of 13 to 51 μm. Digital wood core images were produced using a digital camera (Leica DFC 490) and saved in TIFF format with a 1.3 μm per pixel resolution (Rodríguez-Ramírez et al., 2018). Within each digital wood core image, we identified the area occupied by each growth ring between two wood rays (~16.5 µm wide × 54.4 µm long). We located the annual growth ring boundaries using Adobe Illustrator CC v.23.0.5 (www.adobe.com; Rodríguez-Ramírez & Luna-Vega, 2020).

For each oak species, we calculated the following vessel traits: hydraulic diameter (DH), vessel density (VD), vessel grouping index (VG), and vulnerability index (VI). We measured vessel traits in digital wood core images for the tree-rings previously identified for three different moments: before, during, and after each drought event (Rodríguez-Ramírez et al. 2020). Vessel traits (DH, VD, VG, and VI) were calculated for each digital tree ring area using ImageJ-Fiji4 (Schneider et al., 2012).

Hydraulic diameter (DH; in µm) is associated with fundamental environmental conditions and maximises climatic effects (García-González et al., 2016), where D is the vessel diameter (raised to the 4th and the 5th power), and N is the number of conduits within tree-ring containing n vessels (Souto-Herrero et al., 2017).

To estimate vessel density (VD; number per mm2), we counted the number of vessels in each selected image of each wood core and expressed this figure as 'number per mm2' in the area between two wood rays per growth ring (Rodríguez-Ramírez & Luna-Vega, 2020).

To assess the vessel grouping index (VG; the total number of vessels divided by the total number of vessels groups), we first added the number of individual vessels to the number of vessel groups (Scholz et al., 2013a).                                                          

Carlquist's vulnerability index (VI; with values ranging from 1.0 to 3.0) sensu Hoeber et al. (2014) is commonly used to reveal species adaptations to either xeric (values close to 1.0) or mesic conditions (values close to 3.0). Narrow and abundant vessels show low vulnerability, in consistency with the trade-off between vessel size and embolism resistance. VI was determined by dividing the mean vessel diameter by vessel density (Carlquist, 2020).                                                           

The differences in the values of the vessel traits between the drought years (DY) and the non-drought years (NDY) were evaluated for each study species through two-way repeated measures-ANOVA and pairwise comparisons of all pairs of species-year type combinations, with the rstatics (Kassambara, 2023), and ggplot2 (Wickham et al., 2021) R-packages.

N/A