In upland hardwood forests throughout the Eastern United States, there has been a shift in dominance from oak (Quercus
spp.) to shade-tolerant species such as maples (both Acer rubrum
and A. saccharum
), ashes (Fraxinus
spp.), and elms (Ulmus
]. This shift may be caused by “mesophication” a hypothesized feedback loop by which systematic exclusion of the natural fire regime has allowed fire-sensitive, and often shade-tolerant, species (aka “mesophytes”) to establish and create cooler, moister, and darker understory conditions that are unfavorable for oak regeneration and survival [4
]. These compositional shifts may impact the hydrologic budget at the watershed scale. For example, mesophytes have higher rates of transpiration compared to co-occurring oaks, such that an increase in mesophyte dominance increased watershed evapotranspiration losses by 29%, leading to a reduction in watershed yield by 22% [5
]. While differences in physiology between oaks and non-oaks have been shown to modify the hydrologic regime, species-specific differences in canopy structure may also contribute to changes in water inputs, thus affecting the overall water budget.
The majority of precipitation that passes through the forest canopy and contributes to net inputs of water at the forest floor is termed throughfall. Throughfall generally constitutes about 70% of bulk precipitation, with a much smaller portion, less than 5%, delivered to the forest floor along tree trunks (i.e., stemflow), and the remainder (~25%) intercepted by the forest canopy and evaporated back to the atmosphere [6
]. However, these percentages are highly variable across space and time due to storm meteorological conditions, stand conditions, and species-specific structural characteristics. For example, in deciduous forests, throughfall is lower when foliage is present because leaves provide additional surfaces for water interception, compared to when foliage is absent [9
]. Likewise, denser forest stands with more canopy layers have less throughfall than sparser stands or those with fewer canopy layers. At the individual tree level, species with glabrous leaf coatings shed water more easily than those with rough surfaces, reducing leaf interception and increasing throughfall [10
]. Other leaf characteristics, such as orientation and shape, can influence the ability of individuals to shed or intercept rainwater [12
The importance of species-level traits on the net inputs of precipitation to the forest floor has been investigated, but largely in the context of the sampling accuracy when species biodiversity is high [14
]. Other studies have focused on throughfall variability under a single tree crown [17
]. At the stand- to landscape-scale, the distribution of species has been shown to exert an influence on throughfall variability based on topographic position [19
]. Ultimately though, topographic position determined species composition, and differences in canopy structure among species were the driving cause of the differences in throughfall observed [19
Canopy and bark characteristics of mesophytes may be sufficiently different than that of co-occurring oaks to create a redistribution of water resources to the forest floor. For example, Alexander and Arthur [20
] showed that red maples have 4–12% less throughfall and +250% more stemflow than oaks in an upland oak-hickory forest in Kentucky. Siegert et al. [21
] also observed significant increases in stemflow from red maple and hickory across sites in Mississippi. Along with previous stemflow studies comparing bark morphology to stemflow production [22
], these results provide initial evidence that mesophytic species not only create understories with lower light levels, but also redirect water to soils in the immediate vicinity of their tree boles, which may further reduce the flammability of the landscape. What is less clear is whether crown architecture or leaf water storage characteristics of mesophytes differ from those of oak, and whether changes in stand-level species composition will alter the spatial and temporal input of throughfall as well. As such, this study sought to quantify the spatial and temporal variability of throughfall beneath the crowns of oak and co-occurring non-oak species in relation to canopy structural characteristics. An improved understanding of controls on the forest hydrologic cycle among species can provide hydrologists and other land managers with better tools to manage forests for water resources.
At our study site, southern red oak had the largest diameter at breast height and occupied the most basal area, but had the smallest crown area. In contrast, red maple trees had the smallest basal area but the second highest crown area, illustrating how stem size is not necessarily an indicator of crown size. As a result, the larger crowns of red maple lead to greater rainfall interception and less throughfall compared to a co-occurring oak species of the same size but with a smaller crown. Forest canopies are spatially and temporally heterogeneous for many reasons, but some inherent structure can be attributed to differences in function and ecological niches. For example, pioneer species, which are shade-intolerant, allocate resources to grow tall and fast, resulting in shallower crowns with fewer branches. In contrast, climax species, which are moderate to very shade-tolerant, grow slower and are able to photosynthesize at lower light levels, maintaining multiple layers of branches and foliage, resulting in denser and deeper crowns. In evergreen species, this tradeoff between growth strategies in relation to shade tolerance was observed with vertical branching most common among pioneer species (i.e., shade-intolerant species) and horizontal branching most common among climax species (i.e., shade-tolerant species) [32
]. As such, the structural traits of species reflect differences in functional traits associated with light acquisition and photosynthesis [33
]. Consequently, these structural traits may also impact the redistribution of rainwater by the forest canopy.
With regards to canopy water partitioning, the architecture of tree crowns has direct impacts on the number of layers to intercept rainwater. In tropical forest ecosystems where species diversity is highest, differences in DBH and crown size were observed and were most pronounced among midstory trees [35
]. At our study site in a temperate forest, the crown to basal area ratio was 2.5- to 25-times greater for red maple overstory trees compared to those of white oak and southern red oak, respectively. In the midstory, red maple and winged elm had crown to basal area ratios up to 11 times greater than either oak species. Thus, both canopy layers displayed differences in structure among species, but these traits were stronger in the overstory.
In this study, the rank of standardized throughfall was different between gauges located at the bole of the tree trunk compared to the mid-crown point underneath overstory trees (Figure 5
). Although there were no consistent trends between gauge location or the interaction with species (p
> 0.050). However, overall, the variability of throughfall at this study was within the range (<50%) of CV observed in other systems [15
]. Compared to these studies, we also observed greater variability (i.e., the range of boxplots in Figure 5
and CVs in Figure 3
and Figure 4
) during the leaf-on canopy phase compared to the leaf-off canopy phase, especially for the mid-crown gauges. More variability when foliage is present would be expected, as the number of intercepting surfaces increases with new foliage, whereby the spatial complexity of branching architecture combined with that of foliage distribution leads to even higher throughfall variability.
Keim et al. [38
] considered the spatial scale over which throughfall collectors were correlated, and found that during the leaf-on canopy phase, the structure of individual tree crowns resulted in correlation between gauges, but in the leafless canopy, that correlation weakened significantly [38
]. When foliage is present, the architecture of the tree crown may be more pronounced, leading to stronger spatial patterns. Differences between species in leaf surface characteristics may also influence retention of water by the tree crown. Not only do oaks have smaller crowns, but they have waxier leaf surfaces that help shed intercepted rainfall [10
], both characteristics leading to greater throughfall than non-oaks. Species that are more shade-tolerant also tend to have larger leaves to capitalize on light interception and photosynthesis [40
], which would also increase rainfall interception and decrease throughfall.
Compounding these interactions is also the myriad of storm conditions (i.e., rainfall intensity, wind speed and intensity, synoptic seasonality [41
]), adding to the complexity of an already spatially complex process. In this study, throughfall was collected over 4-week sampling intervals, so any fine-scale temporal influence that species have on throughfall partitioning was not evaluated. However, it is likely that more frequent sampling intervals would help elucidate differences between species. For example, the timing of leaf life cycles in a deciduous forest from leaf emergence to full canopy occupancy to senescence, termed phenoseasons [44
], determine the quantity of radiation making it to the forest floor. At the same time, the quantity of rainfall will also be impacted by phenoseasons. Growth strategies and physiology determine leaf life span [45
], so it is reasonable to hypothesize that these attributes differ between species and will impact throughfall. Additionally, coatings on leaf surfaces change throughout the growing season, such that leaf hydrophobicity decreases as leaves age, increasing interception storage capacity, thus reducing throughfall [47
]. Had this study been able to sample at shorter-time intervals, differences in phenoseasons beyond leaf-on vs. leaf-off canopy phase may have been expressed more strongly and led to periods during the year when throughfall was even more different among species.
The forest hydrologic budget is a balance of inputs via throughfall and stemflow (i.e., net precipitation) and outputs via interception, evaporation, transpiration (i.e., evapotranspiration) and streamflow. This study considered how structural traits vary among species to influence throughfall specifically. Here, we have presented preliminary evidence that suggests throughfall may be reduced during mesophication. However, net precipitation also includes stemflow and the smoother and thinner bark of mesophytes has been shown to substantially increase stemflow production compared to co-occurring oaks [20
]. Evidence also suggests that mesophytes transpire more water than co-occurring oaks [5
]. The impacts of mesophication on forest hydrologic cycles is still unresolved but these few studies provide evidence that watershed yield is decreasing as a result of changing species composition.
Mesophication and changing species composition and structure may have implications beyond forest hydrology. Throughfall is enriched with nutrients that accumulate in forest canopies during antecedent dry periods which are subsequently washed off, and also leached from plant surfaces during rainfall [48
]. Species-level differences in throughfall have been observed across a range of forest functional types [51
], and recent work has linked canopy-derived nutrient inputs with soil microbiological communities [54
]. For example, presence of canopy epiphytes increased nutrient fluxes in throughfall and led to differences in abundance of ammonia-oxidizing soil bacteria, compared to soils underneath trees where epiphytes were not present [54