spatial variations in throughfall in a moso bamboo forest: sampling design for the estimates of...

HYDROLOGICAL PROCESSESHydrol. Process. 24, 253–259 (2010)Published online 28 September 2009 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/hyp.7473

Spatial variations in throughfall in a Moso bamboo forest:sampling design for the estimates of stand-scale

throughfall

Yoshinori Shinohara,1* Yuka Onozawa,1 Masaaki Chiwa,2 Tomonori Kume,3 Hikaru Komatsu,1

and Kyoichi Otsuki1

1 Kasuya Research Forest, Kyushu University, Sasaguri, Fukuoka 811-2415, Japan2 Shiiba Research Forest, Kyushu University, Shiibason, Miyazaki 833-0402, Japan

3 School of Forestry and Resource Conservation, National Taiwan University, 1 Sec. 4, Roosevelt Road, Taipei 10617, Taiwan

Abstract:

We investigated the spatial and seasonal variations in throughfall (Tf) in relation to spatial and seasonal variations in canopystructure and gross rainfall (Rf) and assessed the impacts of the variations in Tf on stand-scale Tf estimates. We observedthe canopy structure expressed as the leaf area index (LAI) once a month and Tf once a week in 25 grids placed in a Mosobamboo (Phyllostachys pubescens) forest for 1 year. The mean LAI and spatial variation in LAI did have some seasonalvariations. The spatial variations in Tf reduced with increasing Rf, and the relationship between the spatial variation and theRf held throughout the year. These results indicate that the seasonal change in LAI had little impact on spatial variations inTf, and that Rf is a critical factor determining the spatial variations in Tf at the study site. We evaluated potential errors instand-scale Tf estimates on the basis of measured Tf data using Monte Carlo sampling. The results showed that the errordecreases greatly with increasing sample size when the sample size was less than ¾8, whereas it was near stable when thesample size was 8 or more, regardless of Rf. A sample size of eight results in less than 10% error for Tf estimates based onStudent’s t-value analysis and would be satisfactory for interception loss estimates when considering errors included in Rfdata. Copyright 2009 John Wiley & Sons, Ltd.

KEY WORDS bamboo forest; canopy; interception loss; Monte Carlo sampling; spatial variation; throughfall

Received 18 December 2008; Accepted 27 July 2009

INTRODUCTION

In Japan, bamboo was introduced from China and bam-boo forests were developed for timber and food produc-tion. Until recently, these bamboo forests had been man-aged, however, now these forests have begun to be aban-doned due to imports of bamboo from foreign countries(Torii, 1998, 2003). As bamboo forests are highly inva-sive, abandoned bamboo forests spread into surroundingvegetation and expand rapidly (e.g. Torii and Isagi 1997;Nishikawa et al., 2005).

This rapid expansion may alter amounts of waterresources because changes in vegetation cover can alterthe terrestrial water cycle. Among various componentsof the vegetation water cycle, evapotranspiration is oneof the most critical factors affecting the water yield(Komatsu et al., 2007). However, very few studies (Abeet al., 1984; Hattori and Abe, 1989) have examinedevapotranspiration in bamboo forests despite many stud-ies examining biomass production in bamboo forests(Singh and Singh, 1999; Shanmughavel et al., 2001;Embaye et al., 2005; Franklin, 2006; Jiang et al., 2006).Thus, characteristics of evapotranspiration in bamboo

* Correspondence to: Yoshinori Shinohara, Kasuya Research Forest,Kyushu University, 394 Tsubakuro, Sasaguri, Kasuya-gun, Fukuoka 811-2415, Japan. E-mail: [email protected]

forests are poorly understood, which prevents accurateassessment of the effects of bamboo forests’ expansionon water resources.

Rainfall interception loss, one major component ofevapotranspiration, is estimated from throughfall (Tf),stemflow (Sf) and gross rainfall (Rf) measurements.Although Sf is significant for the nutrient cycle (Leviaand Frost, 2003) and groundwater recharge (Taniguchiet al., 1996), Sf is generally less than Tf. Sf generallyaccounts for less than 10% of Rf (e.g. Marin et al., 2000;Kuraji and Tanaka, 2003; Toba and Ohta 2005; Manfroiet al., 2004), whereas Tf generally accounts for 40–90%of total annual Rf in forests (e.g. Llorens and Domingo,2007). There have been only two studies examining Tf forbamboo forests (Abe et al., 1984; Hattori and Abe, 1989)and they reported Tf accounted for 75 and 73% of totalRf in a bamboo forest, respectively. Therefore, accuratethroughfall estimates rather than stemflow estimates arecritical for accurate interception loss estimates.

Tf in forest ecosystems is affected by multiple factorssuch as canopy structure (Iwatsubo and Tsutsumi, 1967;Rowe, 1983; Park et al., 2000; Deguchi et al., 2006), andmeteorological conditions such as storm size (Staelenset al., 2006). Thus, it can vary in time and space (Lloydet al., 1988; Tanaka et al., 2005). To estimate stand-scaleTf in forest ecosystems, Tf measurements have been

Copyright 2009 John Wiley & Sons, Ltd.

254 Y. SHINOHARA ET AL.

conducted using a large number (½40) of collectors (e.g.Johnson, 1990; Aboal et al., 1999; Marin et al., 2000;Chappell et al., 2001; Iida et al., 2005; Pypker et al.,2005), as many as 100 collectors in some cases (Konishiet al., 2006) and/or by relocating collectors for long-termmeasurement (Kumagai, 1953; Lloyd et al., 1988; Dykes,1997; Asdak et al., 1998; Marin et al., 2000; Manfroiet al., 2006). However, such measurements are laborious(Sato et al., 2002). Therefore, some studies (e.g. Gashand Morton, 1978; Gash et al., 1980; Edwards, 1982;Llorens et al., 1997) have used a small number (<10) ora modest number (15–30) of collectors for stand-scaleTf estimates. Thus, evaluations of the spatial variabilityof Tf and potential errors resulting from a limitednumber of collectors are of practical importance foreffective measurements in the estimation of stand-scaleTf. Therefore, attempts have been made to determinethe suitable number of collectors for stand-scale Tfestimates using different gauge arrangements (Holwerdaet al., 2006), Student’s t-value analysis (Kimmins, 1973;Kostelnik et al., 1989; Puckett, 1991; Rodrigo and Avila,2001; Price and Carlyle-Moses, 2003) and Monte Carlosampling (Rodrigo and Avila, 2001). However, there havebeen only a few studies examining spatial and temporalvariations in Tf in bamboo forests. These previous studiesdid not evaluate the effects of the spatial and temporalvariations on stand-scale Tf estimates.

We observed spatial variations in the leaf areaindex (LAI) and Tf in a Moso bamboo (Phyllostachyspubescens) forest for 1 year. The major objectives ofthis study were (1) to clarify the spatial and seasonalvariations in Tf in relation to LAI and Rf, and (2) to eval-uate impacts of the spatial Tf variation on stand-scale Tfestimates; that is, to determine how many collectors areneeded to account for the spatial variations in Tf.

MATERIALS AND METHODS

Site description

Experiments were carried out in an abandoned Mosobamboo (P. pubescens) forest in Munakata city, situated30 km east of Fukuoka city, Japan (33°500N, 130°310E,10 m a.s.l.). This site was an agricultural cropland untilthe 1990s. Thereafter, Moso bamboo invaded the sitefrom adjacent bamboo forests for several years. Thesite was covered by Moso bamboo in the measurementperiod. The Moso bamboo forest was ca 100 m in theeast to west direction and ca 50 m in the north tosouth direction. The southern side of the forest facedan agricultural cropland and the other sides faced broad-leaved forests. The site was tiered because of its previousagricultural cropland use.

The mean precipitation (including snowfall as well asrainfall) measured at the Munakata Meteorological Sta-tion, located 5 km south of the site, is 1697 mm. Themean annual temperature is 15Ð3 °C, and ranges from5Ð4 °C in January to 26Ð4 °C in August. The seasonal trend

in precipitation is shown in Figure 1. Most of the precip-itation is from rainfall, not snowfall. However, there aresome snowfall events in January and/or February everyyear.

An experimental plot 10 m ð 10 m in area wasestablished at the centre of the Moso bamboo forest. Theplot contained 68 bamboo plants (i.e. 6800 stems ha�1).The mean diameter at 1Ð3 m above the ground (DBH)was 11Ð3 cm (ranging 6Ð3–14Ð5 cm). The bamboo canopyheight was ca 13 m. The plot was relatively flat (<5°

slope), and understory vegetation was scarce. During theexperiments, two new bamboo plants emerged, and sixfell down. The six fallen plants were removed from theplot as soon as they were found.

Measurements

Rf was measured using two types of rain gauges: astorage-type rain gauge (funnel diameter D 210 mm),and a tipping bucket rain gauge with a resolution of0Ð5 mm (Ohtakeiki, RA-1, Tokyo, Japan). Both weresited in an open space approximately 300 m west ofthe plot. Rf collected by the storage-type gauge wasmeasured once a week, when data collection of Tfmeasurements was carried out. The tipping bucket raingauge was connected to a data logger (HOBO Event,Onset Computer, Bourne, MA, USA) and the tipping timewas recorded. Rf amounts measured by the storage-typegauge (P) were strongly correlated to those of the tippingbucket rain gauge (P0) (P D 1Ð01 ð P0 C 2Ð89, R2 D 0Ð99,n D 35). The intercept value of the relationship betweenP and P0 was relatively large owing to the scatteringof data with large P0. If we determined the interceptvalue using data with P < 10 mm, the intercept was 0Ð25mm. We used Rf data from the storage-type gauge in ouranalysis. We confirmed that our conclusions of this studywere not altered when using gross rainfall data observedusing the tipping bucket rain gauge.

Twenty-five grids 2 ð 2 m in area were establishedper plot and Tf measurements were made in each grid(n D 25). The storage-type Tf collectors (funnel diameterD 210 mm) were set at the centre of each grid. In a

0

100

200

300

400

500

month

Gro

ss r

ainf

all (

mm

)

1 2 3 4 5 6 7 8 109 11 12

Figure 1. Seasonal trend in precipitation (including snowfall) at theMunakata Meteorological Station; 30-year average (bar) and for 2006

(closed circle)

Copyright 2009 John Wiley & Sons, Ltd. Hydrol. Process. 24, 253–259 (2010)DOI: 10.1002/hyp

SPATIAL VARIATIONS IN THROUGHFALL IN A MOSO BAMBOO FOREST 255

grid where a bamboo plant was located at the centre, thecollector was placed next to the plant. Data were collectedonce a week from 28 December 2005 to 26 December2006. Because the measurements were conducted oncea week, Rf and Tf data in this study included 1–6rainfall events. Here, the rainfall events were classified asseparate events if the time between events exceeded 12 h.Snowfall events were included in two weekly perioddata. These data were excluded from the analysis becausesnowfall interception processes are different from rainfallinterception processes (e.g. Lundberg and Halldin, 2001).

LAI was measured at each grid once a monthusing digital non-spherical colour photographs (NikonCOOLPIX990). The photographs were taken 0Ð8 m abovethe Tf collectors on cloudy days to avoid the effectof direct sunlight, and LAI was determined from thephotographs using Gap Light Analyzer software (Frazeret al., 1999).

Sampling size for the estimates of Tf

To evaluate how many collectors are needed for stand-scale Tf estimates in the plot, we calculated the change instand-scale Tf estimation errors caused by the spatial Tfvariation with sample size (n D 2–25) as follows. First,samples (n D 2–25) were extracted from the original Tfdata for each collector (n D 25) using the Monte Carlosampling (Rodrigo and Avila, 2001) and the average ofextracted Tf samples was calculated. The extracted datumwas immediately returned to the data pool and thus aTf sample was always extracted from the original Tfdata (n D 25). The linear congruential method was usedas the algorithm for random number generation. Afterrepeating the above step 1000 times, that is, generatinga thousand average values of extracted Tf samples, theprobably density function (PDF) of the averages ofTf was determined (Kumagai et al., 2005). Finally, thecoefficient of variance (CVM) was calculated from theaverage and the standard deviation of the PDF.

RESULTS

Canopy structure

Figure 2 shows the seasonal variations in the averageLAI for the 25 grids. The annual mean of the averageLAI was 2Ð88 m2 m�2. The minimum value of theaverage LAI was 2Ð27 m2 m�2 in December and themaximum value was 3Ð91 m2 m�2 in April. According toSteel–Dwass multiple comparison tests, the average LAIsfor April and December were significantly (P < 0Ð05)different from those for the other months.

Figure 2 also shows the seasonal variations in themaximum and minimum values of the LAI for the25 grids. The difference between the maximum andminimum for the LAI was greater in April and Novemberthan the other months owing to leaf appearance in Apriland leaf fall in November.

Gross rainfall and throughfall

During the study period, the total Rf excluding thetwo weekly periods with snowfall events was 2105 mm,which was approximately 1Ð3 times the average annualprecipitation. Monthly Rf from April to July was muchhigher than that of normal years, whereas monthly Rffrom September to November was lower (Figure 1).

The total Tf was 1447 mm, which was 69% of Rf. Thisvalue was slightly smaller than that for another Mosobamboo forest examined in previous studies (Abe et al.,1984; Hattori and Abe, 1989), although the stem densityat the present site is approximately half that for theprevious studies. We found a strong relationship betweenRf and Tf for each weekly period (Tf D 0Ð80 Rf � 3Ð01,R2 D 0Ð99). We found no systematic differences in therelationship among seasons (Figure 3).

The total Tf in each grid had no clear relationshipwith the annual average LAI for each grid (R2 D 0Ð02).The PDF of the Tf ratio (Tf/Rf) using all Rf data fromthe 25 grids (n D 868) was distributed from 0Ð1 to 1Ð4,and maximized around 0Ð7 (Figure 4). This distributionpassed the normality test developed by Shapiro and Wilk(1965) (P < 0Ð05); in addition, 2% of the throughfall ratedata exceeded 1Ð0. Furthermore, among the distributionsof Tf from the 25 grids for 36 weekly periods, those for11 weekly periods passed the normality test (P < 0Ð05).

The coefficients of variance of Tf (CVTf) were cal-culated using data for the 25 grids for each weekly

0

1

2

3

4

5

LA

I

7 8 9 10 11 12

month

6542 31

Figure 2. Seasonal change in average leaf area index (LAI) for 25 grids.Vertical bars represent maximum and minimum values

Gross rainfall (mm)

Thr

ough

fall

(mm

)

0 100 200 300 400

100

200

300

0

Figure 3. Relationships between gross rainfall and throughfall:December–February (squares), April–June (circles), July–August (tri-

angles), and September–November (crosses)



0 0.2 0.4 0.6 0.8 1.0 1.2 1.40

1

2

3

4

PD

F (

Tf)

Throughfall rate

Figure 4. Probability density function (PDF ) of the throughfall rate basedon data from 25 grids for all weekly periods

period, and ranged from 8Ð1 to 34Ð3% with an average of14Ð6% for the measurement period. CVTf decreased withincreasing Rf when Rf was less than ¾10 mm, whereasCVTf was nearly stable (¾10%) when Rf was ½10 mm(Figure 5), which was statistically confirmed using theF-test; the variation in CVTf for Rf < 10 mm was sta-tistically larger (P < 0Ð01) than that for Rf ½10 mm.In Figure 5a, the data is divided into three groups bythe number of rainfall events in each weekly period; thenumber of rainfall events was 1, 2, or ½3. We found nosystematic difference according to the number of rain-fall events. Furthermore, the Bartlett test and one-wayanalysis of variance indicated that the distributions of thethree groups were not significantly different from eachother (P > 0Ð05) when Rf was ½10mm.

To examine the impact of the seasonal change in LAIon the spatial variations in Tf, we differentiated the April,November and December data in Figure 5b because theLAI value and/or its spatial variations (i.e. the rangebetween the maximum and minimum LAI) in April,November and December were different from those ofthe other months. Although CVTf of the weekly periodwith the smallest Rf (3Ð1 mm) was larger than CVTf

of the other weekly periods, CVTf in April, Novemberand December was not systematically greater (or less)than CVTf in the other months. Furthermore, the Bartletttest and one-way analysis of variance indicated thatdistributions of CVTf in April, November December, andthe other months were not significantly different fromeach other (P > 0Ð05) when Rf was ½10 mm.

Sampling size for Tf estimates

Figure 6 shows the relationship between the samplesize and CVM for Tf, which was calculated using theMonte Carlo sampling technique for four weekly periodswith different Rf values, namely W.1, W.2, W.3 andW.4 (see Table I). CVM greatly decreased with increasingsample size when the sample size was less than ¾8, andit was nearly stable when the sample size was 8 or more,regardless of Rf. This suggests increasing the sample sizegreatly improves the accuracy of Tf estimates on a stand-scale when sample size is less than ¾8, whereas it doesnot when sample size is 8 or more. CVM stabilized around

0

10

20

30

40

CV

Tf (

%)

0 50 100 150 200 250 300 3500

10

20

30

40

Rainfall (mm)

(a)

(b)

Figure 5. Relationships between the gross rainfall and the coefficientof variation of throughfall (CVTf). (a) Gross rainfall in each weeklyperiod that includes one rainfall event (open circles), two rainfall events(solid circles), and more than three rainfall events (crosses). (b) Data inApril (solid circles), November (crosses), December (triangles), and other

months (open circles)

0 5 10 15 20 250

5

10

15

20

25

Sample size

CV

M (

%)

W.1W.2W.3

× W.4

Figure 6. Relationship between sample size and the coefficient of varia-tion (CVM) in throughfall for four weekly periods with different gross

rainfall amounts (see Table I)

8% for W.1, around 5% for W.2 and W.4, and around3% for W.3.

DISCUSSION AND CONCLUSIONS

To examine the spatial and seasonal variations in Tf inrelation to those of the LAI and Rf, the Tf and LAI valueswere measured in 25 grids in a Moso bamboo forestfor 1 year in this study. CVTf was strongly influencedby Rf when Rf was less than 10 mm, and was fairlyconservative when Rf was greater than 10 mm (Figure 5).This is a similar result to that obtained by Carlyle-Moses et al. (2004): CVTf values remained fairly constant



Table I. Characteristics of four weekly periods that were used for evaluating potential errors resulting from sampling size

Week Observationperiod

Gross rainfall(mm)

Number ofrainfall event

Throughfall(mm)

Throughfall rate(%)

CV(%)

W.1 3–10 October 2Ð7 3 1Ð2 44Ð4 28Ð3W.2 5–12 September 31Ð3 3 22Ð5 71Ð9 15Ð7W.3 21 February–2 March 57Ð6 3 51Ð6 89Ð6 10Ð0W.4 21 August–1 September 163Ð5 1 131Ð0 80Ð1 9Ð2

around 13Ð6% for Rf greater than 15–16 mm in anoak forest in Mexico. In addition, the average CVTf inthis study, 14Ð6%, was within the range of CVTf valuesmeasured in temperate forests: 14–22%, summarized by(Staelens et al., 2006). Furthermore, CVTf obtained inthis study was similar to values reported for broadleafforests in Japan: 11Ð8% for a mixed white oak forest(Silva and Okumura, 1996), 13Ð2% for an oriental oakforest (Otsuki, unpublished data) and 17Ð2% for a broad-leaved secondary forest (Deguchi et al., 2006).

Rf in each weekly period included rainfall for 1–6rainfall events. The relationship between CVTf and Rf didnot significantly differ with the number of rainfall events.This indicates that spatial and temporal variations in Tfprimarily depend on Rf amounts for each weekly periodrather than the number of Rf events. Thus, the resultsbased on data in each weekly period do not change whenusing data for each rainfall event.

In this study, Tf in each grid had no clear relation-ship with LAI for each grid, which agrees with theresults of other previous studies (Loescher et al., 2002;Carlyle-Moses et al., 2004; Ziegler et al., 2009). How-ever, the mean LAI and spatial variations in LAI hadseasonal changes that were larger than those for ever-green forests (e.g. Miyazawa et al., 2008). Deguchi et al.(2006) reported more significant spatial variations in Tfduring a growing season than in winter for a deciduousbroad-leaved forest due to more significant variations incanopy structure during the growing season. These resultsimply that seasonal changes in LAI for our site mightaffect spatial variations in Tf.

Although the mean LAI and spatial variation in theLAI had seasonal changes, the spatial variations in Tf didnot have a seasonal rhythm in accordance with the LAI(Figure 5b). This suggests that the seasonal variabilityof the canopy structure had little impact on the seasonalvariability of spatial variations in Tf at this study site,and that the effects of canopy structure on Tf estimatesare negligible. These results contrast to those from earlierreports on deciduous forests (e.g. Deguchi et al., 2006).The amplitude of the seasonal change in LAI in thebamboo forest was smaller than that of deciduous forests.This could be a reason why the seasonal variability ofspatial variations in LAI had little impact on the seasonalvariability of spatial variations in Tf.

We evaluated the potential errors resulting from sam-pling size using Monte Carlo simulations for four weeklyperiods with different Rf values. Results showed the errordecreased greatly with increasing sample size when the

sample size was less than ¾8, whereas it was nearlystable when the sample size was 8 or more, regardlessof Rf. This indicates that increasing the sample size ismore effective when the sample size is less than ¾8 thanwhen the sample size is 8 or more. A sample size ofeight gives satisfactory results when considering the errorin Tf estimates based on Student’s t-value (e.g. Kim-mins, 1973) instead of the Monte Carlo technique. Notethat the distributions of Tf for the four weekly periodspassed normality test developed by Shapiro and Wilk(1965) (P < 0Ð05). When applying the sample size ofeight, the error calculated from the Student’s t-value ata 95% confidence interval ranged between 5 and 10%for W.3, and around 10% for W.2 and W.4. The errorwas around 20% for W.1, which was higher than theerrors for the other weekly periods. However, the errorfor W.1 had less effect on Tf estimates on longer timescales (e.g. monthly and yearly time scales) because Rffor W.1 was much smaller than values for the otherthree weekly periods. Furthermore, we investigated therelationship between the sample size and the Tf estima-tion error for weekly periods with large Rf (i.e. Rf >50 mm) using Student’s t-value. Data satisfying Rf >50 mm accounted for 75% of annual Rf. We calculatedtotal Tf for the weekly periods with Rf > 50 mm. Thedistribution for total Tf passed the normality test devel-oped by Shapiro and Wilk (1965) (P < 0Ð05). A samplesize of eight resulted in approximately 7% error in Tfestimates, suggesting the appropriateness of the samplesize in Tf estimates. We assume 10% error is accept-able in stand-scale interception loss estimates for thefollowing reasons. Rf, the most important component forinterception loss estimates, usually contains 10% error(Kondo and Watanabe, 1991; Ushiyama and Matsuyama,1995; Komatsu et al., 2008). Note that when assumingan acceptable error of less than 10% for some purposes,a sample size of greater than 8 would be necessary.

A previous study by Abe et al. (1984) estimated stand-scale Tf values in a bamboo forest based on 20 collectors,which were of the same type as ours, and two largecollectors with areas of 4 m2. Abe et al. (1984) developedRf–Tf relationships based on different types of rainfallgauges. We compared Tf values calculated from theRf–Tf relationship for the 20 collectors with Tf valuescalculated from the Rf–Tf relationship for the two largecollectors. For large Rf events (i.e. Rf D 50 mm), Tfbased on the 20 collectors is almost the same as that basedon the two large collectors. This implies the accurateestimation of Tf based on the 20 collectors. These results



based on Abe et al.’s (1984) data are consistent with oursuggestion that a sample size of eight gives satisfactoryresults in stand-scale Tf estimates.

We examined spatial variations in Tf in a bambooforest and the effects of these variations on stand-scaleTf estimates, which have not been resolved adequatelyby previous studies. This study suggests an effectivesampling design of Tf measurements for estimating stand-scale Tf, which can be used in further studies on Tf andinterception loss in other bamboo forests. Such studiesare essential for clarifying the similarities and differencesin evapotranspiration between bamboo forests and otherforests (such as broadleaved and coniferous forests) andtherefore essential for assessing the terrestrial water cyclechange resulting from the expansion of bamboo forests.

ACKNOWLEDGEMENTS

We thank Ms. Shoko Ikezaki (Kyushu University), DrJun’ichiro Ide (Shimane University) and the Laboratoryof Ecohydrology in Kyusyu University Forest for theirhelp and support with our research. Thanks are alsodue to two anonymous reviewers for their critical andconstructive comments. This research was supportedby a research grant from Munakata city, Japan and aGrant-in-Aid for Scientific Research (#20Ð7279) from theJapanese Ministry of Education, Culture, Sports, Scienceand Technology.

REFERENCES

Abe T, Tani M, Hattori S. 1984. Measurements of throughfall andstemflow in a bamboo forest. Transactions of Kansai Branch of theJapanese Forestry Science 35: 265–268. (in Japanese).

Aboal JR, Jimenez MS, Morales D, Hernandez JM. 1999. Rainfallinterception in laurel forest in the Canary Islands. Agricultural andForest Meteorology 97: 73–86.

Asdak C, Jarvis PG, van Gardingen P, Fraser A. 1998. Rainfallinterception loss in unlogged and logged forest areas of CentralKalimantan, Indonesia. Journal of Hydrology 206: 237–244. DOI:10.1016/j.jhydrol.2004.04.007.

Carlyle-Moses DE, Laureano JSF, Price AG. 2004. Throughfall andthroughfall spatial variability in Madrean oak forest communitiesof northeastern Mexico. Journal of Hydrology 297: 124–135. DOI:10.1016/j.jhydrol.2004.04.007.

Chappell NA, Bidin K, Tych W. 2001. Modelling rainfall and canopycontrols on net-precipitation beneath selectively-logged tropical forest.Plant Ecology 153: 215–229.

Deguchi A, Hattori S, Park HT. 2006. The influence of seasonalchanges in canopy structure on interception loss: application ofthe revised Gash model. Journal of Hydrology 318: 80–102. DOI:10.1016/j.jhydrol.2005.06.005.

Dykes AP. 1997. Rainfall interception from a lowland tropical rainforestin Brunei. Journal of Hydrology 200: 260–279.

Edwards PJ. 1982. Studies of mineral cycling in a mountain rain forestin Brunei. Journal of Ecology 70: 807–827.

Embaye K, Weih M, Ledin S, Christersson L. 2005. Biomass andnutrient distribution in a highland bamboo forest in southwest Ethiopia:implications for management. Forest Ecology and Management 204:159–169. DOI: 10.1016/j.foreco.2004.07.074.

Franklin DC. 2006. Wild bamboo stands fail to compensate for a heavy1-year harvest of culm shoots. Forest Ecology and Management 237:115–118. DOI: 10.1016/j.foreco.2006.09.036.

Frazer GW, Canhan CD, Lerzman KP. 1999. Gap light Analyzer (GLA),Version2Ð0: Imaging Software to Extract Canopy Structure and GapLight Transmission Indices from True-Colour Fisheye Photographs .

Simon fraser university, British Columbia, and the Institute ofecosystem studies: Millbrook, New York; 1–36.

Gash JHC, Morton AJ. 1978. Application of the Rutter model to theestimation of the interception loss from Thetford forest. Journal ofHydrology 38: 49–58.

Gash JHC, Wright IR, Lloyd CR. 1980. Comparative estimates ofinterception loss from three coniferous forests in Great Britain. Journalof Hydrology 48: 89–105.

Hattori S, Abe T. 1989. Characteristics of interception loss in bambooforest. Suirikagaku 186: 34–53. (in Japanese).

Holwerda F, Scatena FN, Bruijnzeel LA. 2006. Throughfall in aPuerto Rican lower montane rain forest: a comparison ifsampling strategies. Journal of Hydrology 327: 592–602. DOI:10.1016/j.jhydrol.2005.12.014.

Iida S, Tanaka T, Sugita M. 2005. Change of interception process dueto the succession from Japanese red pine to evergreen oak. Journal ofHydrology 315: 154–166.

Iwatsubo G, Tsutsumi T. 1967. On the amount of plant nutrients suppliedto the ground by rainwater in adjacent open plot and forests? Bulletinof Kyoto University Forest 39: 110–124. (in Japanese with Englishsummary).

Jiang PK, Wu QF, Xu ZH, Cao ZH. 2006. Seasonal changes in soil labileorganic carbon pools within a Phyllostrachys praecox stand under highrate fertilization and winter mulch in subtropical China. Forest Ecologyand Management 236: 30–36. DOI: 10.1016/j.foreco.2006.06.010.

Johnson RC. 1990. The interception, throughfall and stemflow in a forestin highland Scotland and the comparison with other upland forest inthe U.K. Journal of Hydrology 118: 281–287.

Kimmins JP. 1973. Some statistical aspects of sampling throughfallprecipitation in nutrient cycling studies in British Columbian coastalforests. Ecology 54: 1008–1019.

Komatsu H, Maita E, Otsuki K. 2008. A model to estimate annual forestevapotranspiration in Japan from mean annual temperature. Journal ofHydrology 348: 330–340. DOI: 10.1016/j.jhydrol.2007.10.006.

Komatsu H, Tanaka N, Kume T. 2007. Do coniferous forests evaporatemore water than broad-leaved forests in Japan? Journal of Hydrology336: 361–375. DOI: 10.1016/j.jhydrol.2007.01.009.

Kondo J, Watanabe T. 1991. A guide to study on evaporation fromcomplex land surface. Tenki 38: 19–30. (in Japanese).

Konishi S, Tani M, Kosugi Y, Takanashi S, Sahat MM, Nik AR,Niiyama K, Okuda T. 2006. Characteristics of spatial distribu-tion of throughfall in a lowland tropical rainforest, PeninsularMalaysia. Forest Ecology and Management 224: 19–25. DOI:10.1016/j.foreco.2005.12.005.

Kostelnik KM, Lynch JA, Crimm JW, Corbett ES. 1989. Sample-sizerequirements for estimation of throughfall chemistry beneath a mixedhardwood forest. Journal of Environmental Quality 18: 274–280.

Kumagai S. 1953. Sampling rainfall under crown canopy. Bulletin ofKyushu University Forest 21: 61–69. (in Japanese with Englishsummary).

Kumagai T, Nagasawa H, Mabuchi T, Osaki S, Kubota K, Kogi K,Utsumi Y, Koga S, Otsuki K. 2005. Sources if error in estimating standtranspiration using allometric relationships between stem diameterand sapwood area for Cryptomeria japonica and Chamaecyparisobtusa. Forest Ecology and Management 206: 191–195. DOI:10.1016/j.foreco.2004.10.006.

Kuraji K, Tanaka N. 2003. Rainfall interception studies in the tropicalforests. Japan Journal of Forest Society 85: 18–28. (in Japanese withEnglish summary).

Levia DF, Frost EE. 2003. A review and evaluation of stemflowliterature in the hydrologic and biogeochemical cycles of forested andagricultural ecosystems. Journal of Hydrology 274: 1–29.

Lloyd CR, Marques FA, de O. 1988. Spatial variability of throughfalland stemflow measurement in Amazonian rainforest. Agricultural andForest Meteorology 42: 63–73.

Llorens P, Domingo F. 2007. Rainfall partitioning by vegetation underMediterranean conditions. A review of studies in Europe. Journal ofHydrology 335: 37–54. DOI: 10.1016/j.jhydrol.2006.10.032.

Llorens P, Poch R, Latron J, Gallart F. 1997. Rainfall interceptionby a Pinus sylvestris forest patch overgrown in a Mediterraneanmountainous abandoned area I. Monitoring design and results downto the event scale. Journal of Hydrology 199: 331–345.

Loescher HW, Powers JS, Oberbauer SF. 2002. Spatial variationof throughfall volume in an old-growth tropical wet forest,Costa Rica. Journal of Tropical Ecology 18: 397–407. DOI:10.1017/S0266467402002274.

Lundberg A, Halldin S. 2001. Snow interception evaporation. Reviewof measurement techniques, processes and models. Theoretical andApplied Climatology 70: 117–133.



Manfroi OJ, Kuraji K, Suzuki M, Tanaka N, Kume T, Nakagawa M,Kumagai T, Nakashizuka T. 2006. Comparison of conventionallyobserved interception evaporation in a 4-ha area of the same Borneanlowland tropical forest. Journal of Hydrology 329: 329–349. DOI:10.1016/j.jhydrol.2006.02.020.

Manfroi OJ, Kuraji K, Suzuki M, Tanaka N, Suzuki M, Nakagawa M,Nakashizuka T, Chong L. 2004. The stemflow of trees in a Borneanlowland tropical forest. Hydrological Processes 18: 2455–2474. DOI:10.1002/hyp.1474.

Marin CT, Bounten W, Sevink J. 2000. Gross rainfall and its partitioninginto throughfall, stemflow and evapotranspiration of intercepted waterin four forest ecosystems in western Amazonia. Journal of Hydrology237: 40–57.

Miyazawa Y, Kikuzawa K, Otsuki K. 2008. Evaluation of leaf display ofevergreen broadleaved tree species and deciduous tree species underwarm temperate conifer plantations. Journal of Forest Research 13:59–67.

Nishikawa R, Murakami T, Yoshida S, Mitsuda Y, Nagashima K,Mizoue N. 2005. Characteristic of temporal range shifts of bamboostands according to adjacent landcover type. Japan Journal of ForestSociety 87: 402–409. (in Japanese with English summary).

Park HT, Hattori S, Kang HM. 2000. Seasonal and inter-plot variationsof stemflow, throughfall and interception loss in two deciduous broad-leaved forests. Journal of Japan Society of Hydrology and Waterresource 13: 17–30.

Pypker TG, Bond BJ, Link TE, Marks D, Unsworth MH. 2005. Theimportance of canopy structure in controlling the interception ofrainfall: examples from a young and an old-growth Douglas-fir forest.Agricultural and Forest Meteorology 130: 113–129.

Price AG, Carlyle-Moses DE. 2003. Measurement and modeling ofgrowing-season canopy water fluxes in a mature mixed deciduousforest stand, southern Ontario, Canada. Agricultural and ForestMeteorology 119: 69–85. DOI: 10.1016/S0168-1923(03)00117-5.

Puckett LJ. 1991. Spatial variability and collector requirementsfor sampling throughfall volume and chemistry under a mixed-hardwood canopy. Canadian Journal of Forest Research 21:1581–1588.

Rodrigo A, Avila A. 2001. Influence of sampling size in the estimationof mean throughfall in two Mediterranean holm, oak forests. Journalof Hydrology 243: 216–227.

Rowe LK. 1983. Rainfall interception by an evergreen beech forest,Nelson, New Zealand. Journal of Hydrology 66: 143–158.

Sato Y, Otsuki K, Ogawa S. 2002. Estimation of annual canopyinterception by lthocarpus edulis Nakai. Bulletin of Kyushu UniversityForest 83: 15–29. (in Japanese with English summary).

Shapiro SS, Wilk MB. 1965. An analysis of variance test for normality(complete samples). Biometrika 52: 591–611.

Shanmughavel P, Francis K. 2001. Bioproductivity and nutrient cyclingin bamboo and acacia plantation forests. Bioresource Technology 80:45–48.

Silva IC, Okumura T. 1996. Throughfall, stemflow, and interception lossin mixed white oak forest (Quericus serrata Thunb.). Journal of ForestResearch 1: 123–129.

Singh AN, Singh JS. 1999. Biomass, net primary production and impactof bamboo plantation on soil redevelopment in a dry tropical region.Forest Ecology and management 119: 195–207.

Staelens J, Schrijvrt AD, Verheyen K, Verhoest NEC. 2006. Spatialvariability and temporal stability of throughfall water under a dominantbeech (Fagus sylvatica L.) tree in relationship to canopy cover. Journalof Hydrology 330: 651–662. DOI: 10.1016/j.jhydrol.2006.04.032.

Tanaka N, Kuraji K, Suzuki M, Ohta T. 2005. Spatial distributionof throughfall beneath canopies of understory trees in a MatureChamaecyparis obtuse stand. Bulletin of Tokyo University Forest 113:133–154. (in Japanese with English summary).

Taniguchi M, Tsujimura M, Tanaka T. 1996. Significance of stemflow inground-water recharge 1: evaluation of the stemflow contribution torecharge using a mass balance approach. Hydrological Processes 10:71–80.

Toba T, Ohta T. 2005. An observational study of the factors thatinfluence interception loss in boreal and temperature forests. Journalof Hydrology 313: 208–220. DOI: 10.1016/j.jhydrol.2005.03.003.

Torii A. 1998. Estimation of range expansion rate of bamboo stands usingaerial photograph—case study on Mt. Hachiman, Shiga prefecture, andMt. Otoko, Kyoto prefecture, Japan. Japanese Journal of Ecology 48:37–47. (in Japanese with English summary).

Torii A. 2003. Bamboo forests as invaders to surrounded secondaryforests. Journal of Japanese Society Revegetation Technology 28:412–416. (in Japanese).

Torii A, Isagi Y. 1997. Extension of bamboo forests in southern part ofKyoto prefecture. Japanese Journal of Ecology 47: 31–41. (in Japanesewith English summary).

Ushiyama M, Matsuyama H. 1995. A trial manufacture of simpleraingauge and comparison with the traditional observation. Journalof Japan Society of Hydrology and Water resource 8: 492–498. (inJapanese with English summary).

Ziegler AD, Giambelluca TW, Nullet MA, Sutherland RA, Tantasarin C,Vogler JB, Negishi JN. 2009. Throughfall in an evergreen-dominatedforest stand in northern Thailand: comparison of mobile and stationarymethods. Agricultural and Forest Meteorology 149: 373–384. DOI:10.1016/j.agrformet.2008.09.002.


spatial variations in throughfall in a moso bamboo forest: sampling design for the estimates of...

Documents