Stemflow and its controlling factors in the subshrub Artemisia ordosica during two contrasting growth stages in the Mu Us sandy land of northern China

Stemflow is defined as water that drains along the exterior of plant branches and boles from rain or snowmelt (Levia et al. 2015). Stemflow plays an important ecohydrological and biogeochemical role in vegetated ecosystems, because it is a spatially localized point input of water and nutrients at the plant stem (reviewed by Levia & Frost 2003). Stemflow generally represents less than 10% of the amount of gross rainfall, and it is always considered as a minor component of forest canopy water budgets, compared with interception and throughfall (Li et al.,2008b; Levia et al. 2015). As a result stemflow receives minor attention, making it under-represented in the literature (Llorens & Domingo 2007).

Previous studies on stemflow have been conducted primarily in tropical and temperate forests, with relatively few studies having documented the stemflow characteristics for shrubs in semi-arid and arid regions (Carlyle-Moses 2004; Li et al.,2008b). Stemflow can be concentrated and stored in deeper soil layers, so it represents an important potential source of available moisture for shrub growth in desert ecosystems, where water is scarce (Tromble 1987; Li et al.,2008b, 2009). The existing studies on shrub stemflow were mainly focused on deciduous shrubs, but rarely on subshrubs, which are characterized by a woody, perennial base with annual, herbaceous shoots.

Stemflow shows great variability among and within types of vegetation (Levia & Frost 2003; Li 2011). Llorens & Domingo (2007) reported that stemflow averaged 3% of gross rainfall under Mediterranean conditions, but with a 111% coefficient of variation. Numerous previous studies have indicated that stemflow yield is influenced by both abiotic and biotic factors. The abiotic factors include rainfall (amount, intensity and duration) (e.g. Levia & Frost 2003; Li et al.,2008b; Yang et al. 2008) and wind (Van Stan et al. 2011). The main biotic factors are vegetation species, plant age, canopy height, canopy area, basal area, branch angle, leaf area index (LAI), bark roughness, biomass, presence of leaves, and epiphyte coverage (e.g. Navar & Bryan 1990; Martinez-Meza & Whitford 1996; Crockford & Richardson 2000; Levia & Frost 2003; Garcia-Estringana et al. 2010; Zhang et al. 2015). In fact, many abiotic and biotic factors are mutually interactive, such that actual stemflow yield – at the level of species and individuals – results from a complex and dynamic set of interactions that cascade as a function of space and time (Levia et al. 2015).

Canopy structure is a key factor that accounts for differences in stemflow yield (Crockford & Richardson 2000; Levia & Frost 2003; Li et al.,2008b). However, the elements of canopy structure change substantially during growing seasons, so it is difficult to obtain some canopy structure metrics during growth (e.g. total leaf count, canopy area, leaf area, and biomass) (Levia et al. 2015). Canopy structure parameters were traditionally assumed to be static constants in the hydrologic model (Zimmermann et al. 2015), and hence they were measured using destructive sampling methods in past experimental studies. Knowledge gaps therefore persist, regarding the effects of vegetation growth dynamics on stemflow production.

To address those gaps, the objectives of this study were to: (1) determine the influence of rainfall characteristics (amount and intensity) and canopy structure (basal diameter, canopy area, leaf area, LAI and individual height) on stemflow yield for the unstudied subshrub Artemisia ordosica; and (2) compare stemflow yield processes between two contrasting growth stages (vegetative growth stage versus reproductive growth stage).

The field experiment was conducted at the Ordos Sandland Ecological Research Station, Chinese Academy of Sciences, during the rainy season from May to September in 2012. The station is located at the northeastern margin of the Mu Us Sandland in the Inner Mongolia Ordos plateau (39 ° 29′ 37.6″ N, 110 ° 11′ 29.4″ E) at an elevation of approximately 1,300 m. The climate is continental semi-arid. Mean annual precipitation is 345 mm, occurring mainly between July and September, and mean annual pan evaporation is 2,535 mm. Mean annual temperature is 8 °C with a maximum monthly temperature of 22 °C (July) and a minimum of −10 °C (January). The landscape is mainly fixed dunes and mobile dunes with aeolian sandy soil (Zhang 1994). The dominant plant species in the study area is Artemisia ordosica, and prevalent associated species include Hedysarum mongolicum, Poa attenuata spp., Cleistogenes squarrosa, Oxytropis psammocharis and Cynanchum komarovii (Zhang et al. 2007).

A. ordosica, a medicinal and sand-binding subshrub of the Asteraceae, is distributed widely across northern and northwestern China (Li et al. 2008a). Mature individuals are 0.6–1.0 m tall, with numerous slender branches extending from the base of the main trunk; the above-ground part of the subshrub is approximately hemispherical from the center. Compared to deciduous shrubs, subshrubs like this species have high guard cell compliance and more potential hydraulic conductance due to their less lignified stems and leaves (Gao et al. 2013). The individuals sampled for experiments were situated in an area where grazing is prohibited, near the meteorological observation station.

An aluminum foil cylinder was fitted around the stem of each individual and secured with silicone sealant to collect stemflow (Bui & Box 1992) (Figure 1). Stemflow drainage from each sample was funneled into a plastic bottle via a flexible rubber hose, and measured with a graduated cylinder immediately after each rainfall event. The stemflow volume (mL) of each plant was divided by its canopy area to calculate the stemflow depth (mm) of each A. ordosica individual. Rainfall was measured automatically and recorded at 10 min interval with a tipping bucket rain gauge (Delta-T, Cambridge, UK) located in an open area 10 m away from the study plots. Average rainfall intensity (I, mm/h) was used to represent rain intensity. The minimum rainfall (rainfall thresholds) for stemflow generation was estimated by linear regression equations between individual stemflow and individual rainfall (Diskin 1970; Li et al. 2008a, 2008b). The stemflow percentage of gross rainfall (SF%) was calculated by dividing rainfall amount (mm) by stemflow depth (mm) based on individual rainfall events.

Levene's test and t-test were used to analyze differences in rainfall regime and canopy structure between Stage A and B. A significant Levene's test (P < 0.05) means that equal variances were not assumed (Schultz 1985), and the results of the t-test are meaningful. Correlations between the stemflow volume and canopy structure variables were assessed using Pearson's correlation analysis. All statistical tests were performed using SPSS for Windows Version 13.0 (SPSS Inc., Chicago, USA).

A total of 45 rainfall events were recorded from 28 May to 16 September in 2012, with a total amount of 520.6 mm. Only 23 of them had an amount large enough (>1.0 mm) to enable observation of the yield of stemflow. Some erroneous data were discarded, such as measures of stemflow that exceeded the volume of the collection container because of extreme rainfall. Then, 16 typical rainfall events were chosen for analysis, accounting for 43.5% of the total amount of rainfall and 37.7% of the total number of rainfall events during the growing season. Half of the rainfall events (N = 8) occurred during the vegetative growth stage, with a range from 1.2 to 26.2 mm. Similarly, eight rainfall events occurred during the reproductive growth stage, with a range from 2.2 to 30.8 mm. The rainfall intensities (I, mm/h) ranged from 0.4 to 11.7. There is no significant difference in either the amount or intensity of rainfall between the two growth stages, based on t-test comparisons of Stage A and Stage B (Table 2).

Stemflow volume is significantly and positively correlated with rainfall amount, basal diameter, canopy area and leaf area in both growth stages (Table 3). However, there is a significant negative correlation between stemflow volume and LAI in only the reproductive stage (Table 3). This suggests that LAI has different impact on stemflow yield at different growth stages, which could be important for stemflow modeling.

The findings of this study demonstrate that rainfall amount, canopy area and leaf area are the critical factors that govern stemflow yield of the subshrub of A. ordosica (Table 3). The positive linear relationship between stemflow volume and rainfall for A. ordosica in this study agrees well with that for shrub species reported by previous studies (Price & Carlyle-Moses 2003; Li et al.,2008b; Yang et al. 2008; Janeau et al. 2015). It is interesting to note that the greater the size of the A. ordosica, the more stemflow volume it produces (Figure 6(a)), but the medium-sized individuals tend to generate more stemflow per unit canopy area (SF%, Figure 6(b)) and direct more rainfall to the base of the plant (funneling ratio, Figure 6(c)) than either small- or large-size individuals. The possible reason is that the ratio of woody to foliar biomass is greater for medium-sized individuals than that of small- or large-size individuals.

Levia et al. (2015) reported that stemflow yield was large for trees with greater ratio of woody to foliar biomass. Average funneling ratio is 75.8 for A. ordosica, meaning that the canopy of the A. ordosica subshrub can collect 75.8 times the amount of water that rainfall delivers to the plant's base as a point input resource (Janeau et al. 2015). Thus, branches and stems were contributing fully to stemflow yield because the funneling ratios were larger than 1 (Navar 1993). The ratio in the study is higher than values reported for other shrub species, such as Artemisia sphaerocephala (41.5, Yang et al. 2008), Salix psammophila (69.4, Yang et al. 2008), Tamarix ramosissima (24.8, Li et al.,2008b) and Reaumuria soongorica (53.2, Li et al.,2008b), but lower than the ratio for Caragana korshinskii (153.5, Li et al.,2008b). The stemflow percentage and funneling ratio of A. ordosica (SF% = 8.56%, F = 75.80) were comparable with values for S. psammophila (SF% = 7.6%, F = 69.4; Yang et al. 2008), another common sand-fixed shrub in China.

This study reveals that the unique growth pattern creates a significant difference in the canopy structure characteristics of the two growth stages of A. ordosica, and consequently the stemflow also differs greatly at those times of the year. Since canopy structure is dynamic and complicated for certain vegetation types, further studies are needed to investigate the effect of dynamic canopy structure on stemflow production for a wide range of vegetation species, which would be a challenging step in advancing understanding of the effect of canopy on the hydrology and biogeochemistry.

The factors that control stemflow yield for the subshrub A. ordosica at two typical growth stages were evaluated in this study. The results advance our understanding of the effects of growth dynamics on stemflow yield. The growth pattern of subshrubs in a semi-arid area differed greatly from those of trees and shrubs, which in turn had important influences on canopy structure and stemflow yield. The stemflow yield was influenced not only by rainfall events, but also by individual canopy structure characteristics. Large plants generated more stemflow volume, especially during large rainfall events; however, medium-sized plants tended to direct more rainfall to their bases per unit canopy area. Overall, the major control factors of stemflow volumes were rainfall amount and canopy area, which were not constant, but varied throughout the growth seasons. Hence, discrimination of different growth stages will be helpful for improving the accuracy of eco-hydrology model simulations to assess the water dynamics of subshrubs in semi-arid regions. The results of this work fill a knowledge gap in our understanding of stemflow production for subshrub species.

This study was financially supported by the National Science Foundation of China (Grant number: NSFC 41471018, 41390462 and 41025001), the PCSIRT (No. IRT-15R06), and projects from the State Key Laboratory of Earth Surface Processes and Resource Ecology. We also thank the editor and the anonymous reviewers for their valuable and constructive comments for the improvement of the manuscript.