3.2 Comparison of Dry-Down Curves and Leaf Wetness Duration between Vegetation Types
After rain ceased, the period of dry-down for all crop types tended to follow a logarithmic decay (see Figure 2). Over the period from saturation to fully dry, there was a very good fit to the linearized log trends of the back-transformed data for all dry-down events (R2 ≥ 0.93), and the slopes differed greatly between vegetation types (p < 0.05).
The closed-canopy forest stayed, on average, wetter longer than any other vegetation type (see Table 1 and Figure 3, p < 0.05). Closed-forest leaf wetness dried at a relatively consistent rate after the initial 100 minutes, when the rate became less steep. Based on this observed shift in drying rate for the forest, a separate analysis of the first hour after the event was conducted. When examining the initial drying period of one hour, a linear fit was most appropriate. Closed forest had a less negative slope than all other locations, –0.52% wetness lost per minute compared to –1.22 ± 0.07% per minute. Finally, forests took ~250 minutes longer to reach the level of dryness observed in crops after only one hour (see Figure 2 and 3).
Papaya retained wetness longer at a lower level after the initial period of similar drying, instead of continuing to dry at a uniform rate. When comparing the three crops to each other individually, there is an apparent inflection point before which the non-transformed slopes of the three crops were –0.12, –0.11, and –0.10% per minute, respectively, all very steep with similarly high goodness of fit (R2= 0.93, 0.94, 0.87 for papaya, taro, and sweet potato). However, at 50 minutes, papaya begins to have a slope of –0.02% per minute, less than 1/7 of the rate of change as during the first hour (Figure 2), and LWD extends much longer than for the shorter canopy crops. Drying rates were remarkably similar between taro and sweet potato (p < 0.05, see Figure 3), which is surprising considering sweet potato is a ground-cover crop and taro had a more distinctive canopy up to 1.3 m in height.
The mean LWD in closed mature forest of 339 ± 174 minutes was 3.6 times greater than crops, which averaged 94 ± 37 minutes, and more than 2.6 times the duration of open forest (129 ± 68 min, see Figure 3). Papaya LWD behaved more similarly to open forest than to the lower crops despite being similar in height to the crops, drying almost twice as slowly (LWD = 137 ± 51 min) as taro and sweet potato (LWD = 73 ± 23 min, all with p < 0.01). The dates and time of day for each dry-down event as well as the event duration are listed in Table 1 by vegetation type.
As height increased, the slope of the dry-down curve flattened, as is expected if the higher canopy and inter-canopy space retain moisture longer after a dry-down event. We found strong evidence that height could be used to predict LWD (R2 = 0.94, see Figure 3). For short-stature vegetation around 1 m, a 1 m increase in height extends dry down rates by 18%, whereas for taller vegetation around 10 m, a 1 m increase in height only extends dry down rates by 0.17% per minute (Figure 4).

Discussion

In very wet tropical systems, the use of LWD as an additional variable to study the impact of deforestation on available water in the forms of canopy water storage capacity, intercepted rainfall, and leaf evaporation appears to have merit, as it relates to canopy properties that these other variables do not fully capture. A five-fold longer LWD in forest than crop fields is consistent with previous findings, demonstrating that LWD varies depending on position in canopy and species tested (Sentelhas et al., 2005). Although we noted significant differences in LWD between short and tall statured crops, we were unable to discern a difference between small crops using comparisons between only a few rain event dry-down cycles. However, we report a predictive relationship that can better differentiate how small changes in canopy heights, even in short-statured crops, can lead to incremental changes in LWD.
Our results also demonstrate that LWD can contribute to the larger question about the impact of land use on hydrology, particularly in very wet systems with high interception fractions. Premontane tropical forests undergo frequent wet-dry cycles, thus the fraction of the water budget affected by interception is relatively high compared to temperate regions. At this site, vegetation below the forest canopy was wet up to 85% of the time (Aparecido et al., 2016). Increased LWD is indicative not only of higher surface area of leaves intercepting precipitation, which translates to less water reaching the ground surface and more potential for water storage on the canopy, but also the aerodynamic properties that allow air in the canopy to mix, which lets leaves re-wet repeatedly throughout an event and between events that occur sequentially. LWD may therefore be a key covariant with I and E that could improve predictions of water budgets in these forests and determine the effects of deforestation on stream flow more precisely.
Tropical deforestation has been found to impact stream flow dynamics. In Mexico, for example, deforestation of cloud forest was associated with more erratic flows during the dry season (García Coll, 2002). Bruijnzeel and Scatena (2011) also concluded that conversion of lower montane cloud forests will likely considerably increase runoff locally due to low cloud water interception. The shorter LWD we observed in the crop fields show similar findings.
We further examined wetness differences between crop types by comparing RH data from sensors in the crop boundary layer. We observed the apparent conundrum that papaya humidity was lower, but it still took a long time to dry. It is reasonable that the leaf surface properties held water longer, supported by the fact that the first hour dried similarly as taro. Although leaf wetness sensors do not mimic leaf surface properties, the extended drying times in papaya suggest the leaf surfaces might trap water and release it more slowly than in taro, which has smoother and waxier leaves. Both leaves are much larger than sweet potato leaves. Furthermore, because papaya plants are more heterogeneously spaced in wide rows, the turbulent mixing ratio is higher in the space surrounding that crop. That could explain why usually papaya interspaces were less humid than taro’s or potato’s. Increased turbulent mixing within more widely spaced crops may counteract longer LWD to some extent, as seen in Bailey and Stoll (2013) and Bailey et al. (2014). Sweet potato and taro have lower roughness coefficients than the taller and structurally more complex papaya, creating a more defined boundary layer than in papaya, which acts as a pocket of humid air at the canopy (Raupach, 1994). Nonetheless, we postulate that this dampening boundary layer effect combined with greater surface area of papaya leaves could explain the net effect of longer LWD in papaya than shorter crops. Further work is needed to understand the relative importance leaf properties (size, texture), canopy properties (shape, spatial orientation), and total surface area on the water budget.
Our results demonstrate how the spatial orientation of plants in tall canopies can affect drying rates. Even though within-crop biomass differences were much smaller than biomass differences between forests and crops, LWD differed by less than an order of magnitude. Asdak et al. (1998) found that rainfall interception loss decreased with reduced canopy cover in progressively more open forest areas, which approximates our result from open canopy forest conditions. Thus, forest fragmentation is likely to have an effect on evaporation through not only decreased interception, but also faster drying rates. We found statistically indistinguishable trends in LWD between papaya and open canopy forest, suggesting that even modest canopy gaps can effectively alter the hydrologic characteristics of otherwise intact forests. This has interesting policy implications, because it suggests that agroforestry, even of thin short-lived trees like papaya, may provide similar hydrological services to that of open forests. This is an important issue to highlight, as there are many programs focusing on agroforestry in the tropics (Mercer, 2004).