Introduction

The relationship between forest cover and the hydrologic cycle is complex. A growing scientific consensus considers that stream flows in temperate forests tend to have smaller peaks during flooding and sustain higher base flows during drought compared to non-forested streams, but the same dynamics do not apply in the same way in the humid tropics (Brown et al., 2005). Furthermore, potential differences in stream flows are associated with a variety of drivers, including the partitioning of evapotranspiration (Good et al., 2017). Rainfall interception is of even greater importance in the humid tropics, as these regions experience frequent wet-dry cycles, considerable canopy interception, and high evaporation of intercepted precipitation, ranging from 10-30% of gross rainfall (Bruijnzeel and Scatena, 2011; Calder et al., 1986).
Rainfall interception has been widely studied in humid tropical systems (Bruijnzeel and Scatena, 2011; Chu et al., 2014; Giambelluca and Gerold, 2011; Holwerda et al., 2011; Safeeq and Fares, 2014; Takahashi et al., 2011). In most cases, deforestation scenarios have been associated with reduced interception, increased surface runoff (Maris, 2015; Panday et al., 2015), and exacerbated streamflow peaks and low flows (Brookhuis and Hein, 2016; Bruijnzeel and Scatena, 2011; Laurance, 2007; Ty et al., 2011). Research that examines hydrological shifts accompanying vegetation changes in humid areas is directly applicable to similar regions. In fact, many reviews of such cases (Bowling et al., 2000; Brooks et al., 2012; Bruijnzeel and Proctor, 1995; Bruijnzeel, 2004; Calder, 2001; Eisenbies et al., 2007; FAO, 2005) broadly generalize the mechanisms by which forests may reduce chances of floods: through increased interception, evaporation, and reduced overland flow of water.
A study in Venezuela reported a sevenfold reduction in foliage interception between tropical montane forest and pastureland paired with higher surface runoff (Ataroff and Rada, 2000). Asdak et al. (1998) found a similar decrease with reduced canopy cover in progressively more open forest areas: closed canopy, partial canopy, and canopy gap, respectively. They proposed discontinuous canopy structure, rather than deforestation, as the cause of the reduced evaporation. Similarly, a modeling experiment in Costa Rica examined scenarios of sequentially more deforested areas, from pristine forests to extreme deforestation, and found that although water yield was not significantly altered, forest cover was inversely related to runoff peaks and low flows (Birkel et al., 2012).
Canopy interception (I) differs by orders of magnitude over very short distances in heterogeneous natural forests (Teale et al., 2014), making it difficult to capture spatial variation adequately to produce reliable estimates. I is driven by biomass, whereas canopy evaporation (E) is driven by canopy structure (in turn driven by how well air mixes in the canopy). Leaf wetness duration (LWD) is affected by both structure (Kume et al., 2006) and biomass, since higher leaf area and thus I generates humidity in the canopy that slows down E. We expect LWD is highest in canopies with high I, but the magnitude difference in LWD is much lower than differences in I.
We use a case study in a Costa Rican premontane tropical forest, a high precipitation environment, to evaluate canopy wetness properties due to conversion from mature forest to monoculture row crops. One of our objectives was to learn more about differences in disturbed forests with gaps, indicative of selective logging, which is common in this region. Further, we wanted to compare between forest and commonly grown crops in the region. This helps establish to what degree taller, more complex vegetation intercepts more moisture and stays wet longer than short-statured crops, releasing water to the atmosphere rather than it running off and increasing stream flow. Our final objective was to determine whether canopy height can be used to predict LWD.

Materials and Methods