Biologia, Bratislava, 61/Suppl. 19: S261—S265, 2006Section BotanyDOI: 10.2478/s11756-006-0169-7

Effect of forest and grassland vegetation on soil hydrologyin Mátra Mountains (Hungary)

Andrea Hagyó1, Kálmán Rajkai1 & Zoltán Nagy2

1Research Institute for Soil Science and Agricultural Chemistry of Hungarian Academy of Sciences, Budapest, Hungary;tel.: +36-1-2243652, e-mail: ahagyo@rissac.hu2Szent István University, Department of Botany and Plant Physiology, Godolo, Hungary

Abstract: Water retention characteristics, rainfall, throughfall and soil water content dynamics were investigated in alow mountain area to compare a forest and a grassland. The soil water retention curve of the topsoil has similar shape inboth studied areas, however that of the deeper soil layer shows more difference. We determined the precipitation depth,duration and intensity values of rainfall events. The relationship between rainfall and throughfall depth was described inlinear regressions. Interception was calculated as the difference between rainfall and throughfall plus stemflow, assumingstemflow to be 3% of rainfall. Soil water content dynamics show a similar trend in the two vegetation types but the dryingis more intensive in the forest in the soil layers deeper than 20 cm during the growing-season.

Key words: forest, grassland, interception, throughfall, soil moisture dynamics

Introduction

Evapotranspiration (ET) and interception are the ele-ments of the water cycle of the soil-plant-atmospheresystem that are determined principally by vegetation.ET includes both soil evaporation and transpiration. In-terception loss refers to the amount of water interceptedand lost by evaporation from the canopy. There aremany studies investigating the quantitative importanceof it in different vegetation types, but most studies havebeen conducted in forests, where interception has beenfound to be a significant or even a dominant elementof ET. It can be estimated from canopy characteristicsor can be calculated from the difference of the rain-fall and throughfall plus the stemflow (Merta et al.,2006). Throughfall is the fraction of rainfall that getsthrough the plant canopy, directly through the canopyor with delayed dropping from the leaves and branches.Stemflow, the part of precipitation that reaches the for-est floor flowing down the trunks, is usually a minorcomponent of the water balance. It can reach 3–10%of the rainfall in foliated deciduous forests (Price &Carlyle-Moses, 2003). The integrated effect of veg-etation on the water cycle can be investigated by theanalysis of soil moisture dynamics (Štekauerová etal., 2006).The objectives of our study were (1) to compare

the water retention characteristics of a Vertisol in agrassland and in a forest stand, (2) to determine therelationship of rainfall measured in open grassland to

throughfall in the forest and forest canopy cover, and(3) to compare the soil water content dynamics in thetwo vegetation types.

Material and methods

The study is carried on in the Mátra Mountains, NorthernHungary (N 47◦50′37.6′′ , E 19◦43′16.5′′). Elevation is 278 ma.s.l. Mátra Mountain has a stratovulcanic structure; rhyo-lite tuff, pyroxene-andesite tuff and lava are characteristic.The experimental site is situated on a nearly flat surface.The studied soil is Haplic Vertisol according to WRB (1998).Soil texture is clay. The depth of humus layer is 35 cm. Thepotential vegetation is Quercetum petraeae-cerris.

We studied a grassland and a forest located next toeach other. The forest is dominated by Turkey oak (Quer-cus cerris) (50% cover). Small leaved lime (Tilia cordata)and field maple (Acer campestre) are present with 20–20%cover, while robinia (Robinia pseudo-acacia), northern redoak (Quercus rubra), scots pine (Pinus sylvestris) and wildpear (Pyrus pyraster) are considered as rare species (1–5%cover). The total canopy cover is 80–100%. The cover of theshrub layer is 10–50%. The studied forestry unit has an areaof about 100 by 200 m. The grassland is dominated by Fes-tuca sp., Carex cariophyllea and Poa angustifolia. The mainmass of roots is in the 0–20 cm soil layer. The native forestwas cleared and the area was converted to cultivation. Laterit had been used as a cattle pasture for about 22 years. Thegrazing was stopped in 2004. The grassland has an area of13.5 ha.

Automated soil moisture measurements have been on-going in three soil profiles in the forest (soil profiles/sites

c©2006 Institute of Botany, Slovak Academy of Sciences

S262 A. Hagyó et al.

Table 1. Measurement periods of the soil moisture probes and rain gauges at the 4 sites.

Site 1 (forest) Site 2 (forest) Site 3 (forest) Site 4 (grassland)

Soil moisture 15 Apr 2005– 15 Apr 2005– 26 June 2005– 15 Apr – 15 June 2005,(SWC) 5 Apr 2006 5 Apr 2006 5 Apr 2006 2 Aug 2005 – 5 Apr 2006

Rainfall/ 27 Sept – 24 Oct 2005,27 Sept – 28 Nov 2005

27 Sept – 28 Nov 2005 27 Sept – 28 Nov 2005,Throughfall 27 March – 4 May 2006 27 March – 4 May 2006 27 March – 4 May 2006

a) b)

0-5 cm

0

10

20

30

40

50

60

70

0 1 2 3 4 5

pF (lg(cm))

Wat

er c

on

ten

t (V

/V%

)

WC-OBS grassWC-FIT grassWC-OBS forestWC-FIT forest

20-25 cm

0

10

20

30

40

50

60

70

0 1 2 3 4 5

pF (lg(cm))

Wat

er c

onte

nt

(V/V

%)

WC-OBS grassWC-FIT grassWC-OBS forestWC-FIT forest

Fig. 1. Water retention curves of the topsoil (a) and of the 20–25 cm soil layer (b) in the grassland and in the forest.

Nos 1, 2 and 3) and in one profile in the grassland (soil pro-file/site No. 4). Five capacitive soil moisture probes wereburied in each profile, between soil depths from 15 cm to 55cm with 10 cm vertical distance between the sensors. Theprobes were placed parallel to the soil surface, orienting theflat side perpendicular to the surface of the soil. One of theprobes was calibrated in the laboratory for the studied soilaccording to the user guide (Decagon Devices, Inc., 2005).The calibration equation is: SWC (V/V%) = 0.0652P (mV)− 50.683, R2 = 0.9544, where P is the output of the mois-ture probe. Soil water content (SWC) was recorded every30 minutes by the system.

Throughfall was measured with three tipping bucketrain gauges (HOBO Data Logging Rain Gauges) set up nextto the three soil profiles in the forest. Measurement resolu-tion is 0.2 mm and data were recorded at each tip. We deter-mined the canopy cover above the gauges using digital pho-tographs. The camera was placed on the rain gauges withthe lens facing upwards. The images were analysed with theAdobe Photoshop 7.0 Trial software (Adobe Systems Inc.,USA) following a protocol described in ENGELBRECHT &HERZ (2001), changing the protocol when the sky was tooclouded, in which case we set the brightness to 0. Using thismethod, we obtained the percentage of black pixels, whichwas considered to be the canopy cover.

Rainfall was measured with a tipping bucket rain gauge(Campbell ARG 100 Tipping Bucket Raingauge) with a res-olution of 0.2 mm, installed at site No. 4 in the open grass-land. Data was recorded on a 30-minute interval. The rain-fall events were differentiated as separate events if the timebetween the events exceeded 5 hours. The duration of eventswas determined as the sum of half-hour intervals when pre-cipitation was recorded. Events of only 0.2 mm precipitationdepth (25 events) were excluded from the analysis.

There were some breaks of continuity of the measure-ments because of technical failures with the data loggers and

the cables of the sensors. We summarized the time periodswhen the different measurements were going on in Table 1.

Soil texture characteristics, bulk density, humus con-tent and water retention characteristics (for three soil sam-ples from soil profiles No. 2 and 4.) were determined in thelaboratory for two soil layers (0–5cm and 20–25cm). Soilconductivity measurement with CF ec instrument (RISTO-LAINEN et al., 2006) showed that there is low spatial vari-ability within the grassland and the forest.

Results

The water retention curves of the two layers in thegrassland and in the forest are shown on Fig. 1. Thesaturated and wilting point water contents of the top-soil are similar in the grassland and in the forest. Theshape of the two curves is also comparable, but thewater content values of the grassland are about 5–7%higher along the curve. The difference can be explainedby the lower bulk density in the grassland (1.18 g/cm3

in the grassland vs. 1.26 g/cm3 in the forest). The waterretention curve of the deeper soil layer is more differentfor the two sites. Saturated water content of the grass-land is 14% lower than that of the forest. It can beresulted by the higher dry bulk density of this layer inthe grassland (1.43 vs. 1.33 g/cm3). The higher densitymay be the result of the former ploughing and cattletrampling, while tree roots loosen the soil in the forest.There is a difference of approximately 10% in the lowsuction range moisture retention between the forest andthe grassland, which is decreasing toward the highersuction range and becomes similar at wilting point. The

Effect of forest and grassland vegetation on soil hydrology S263

a) b)

0

10

20

30

-2 -5 -10 -15 15-

Precipitation depth (mm)

Fre

qu

ency

(c

ou

nt)

0

10

20

30

-5 -10 -15 -30

Rainfall duration (h)

Fre

qu

ency

(c

ou

nt)

0

10

20

30

-0.5 -1 -1.5 -2 -3 5-

Rainfall intensity (mm/h)

Fre

qu

ency

(c

ou

nt)

c)

Fig. 2. The frequency distribution of precipitation depth (a), rainfall duration (b) and rainfall intensity (c).

moisture content at wilting point is high in both layersat both sites because of the high clay content.The total incident precipitation during the periods

of 27 September to 28 November 2005 and 27 Marchto 4 May 2006 was 37.6 mm and 70.8 mm, respec-tively. A total of 43 precipitation events were studiedduring the two periods. Mean event incident precipi-tation input was 3.5 mm and it ranged from 0.4 mmto 22.6 mm. Individual event durations averaged 5.3 hand ranged from 0.5 h to 29.5 h. The mean rainfall in-tensity was 0.73 mm h−1 ranging from 0.09 mm h−1

to 5.05 mm h−1. The frequency of precipitation depth,duration and intensity values based on the 43 eventsare presented in Fig. 2.Canopy cover was similar in October and in May

above the same rain gauges, ranging from 78.34% to92.87%. It was highest above Rain gauge #3 and itwas the lowest above Rain gauge #1. We determinedthe cover of branches without leaves on 20 March 2006(52.20%, 34.04%, 42.27% above rain gauges 1, 2 and 3,respectively).The total throughfall in the period of 27 September

– 28 November 2005 was observed to be 63.79%, 78.2%and 73.9% of rainfall at sites 1, 2 and 3, respectively. Itwas 68.6% and 65.5% at sites 1 and 3 in the period of27 March – 4 May 2006.Considering the separate precipitation events the

mean throughfall fluxes were 46.6%, 45.05% and 40.47%varying from 8.33 to 80%, 15.56 to 80% and from 9.38 to80% in percentage of rainfall depths. The relationshipbetween incident rainfall and throughfall was approxi-mated in linear regression (equations 1–4):

TF1 = 0.6901P − 0.1539, R2 = 0.9437, N = 24, (1)

TF2 = 0.7685P − 0.1072, R2 = 0.9862, N = 8, (2)

TF3 = 0.7495P − 0.2142, R2 = 0.9732, N = 21, (3)

TFmean = 0.7293P − 0.1946, R2 = 0.9884, N = 29.(4)

TF1, TF2, and TF3 are the throughfall depths mea-sured with the three rain gauges; TFmean: mean valueof throughfall depths; P: precipitation depth.Delayed flow could be observed: delayed dropping

from leaves and branches occurred during and after rainevents. The delay between the first (start of event) andthe last tip (end of event) of recorded rain and through-fall events could be calculated with half-hour resolution.Three periods were distinguished: 29 Sept to 29 Nov2005 (1), 29 March to 10 Apr 2006 (2) and 11 Apr to4 May 2006 (3). In the autumn period the mean delayat the start of events is approximately 3 h in case of allthe TF measuring plots. The mean delays at the endof the events are 2.2 h, 2.7 h and 3.8 h at the threesites, respectively. The minimum of delay in this periodwas 1h at all sites for both the start and the end of theevents. The maximum delay ranged from 5.5 h to 6.5 hat the start and from 3 h to 7 h at the end of events.In spring the mean delay varied around 2 h at site #1(variance is 1.06 and 1.70 for start and end). Despiteat site #3, mean delay (start: 3.86 h and 2.64 h, end:4.71 h and 1.90 h) differed in the two periods.We assumed the stemflow to be 3% of rainfall

as the lower limit found in the literature (Price &Carlyle-Moses, 2003). Interception – estimated tobe the difference between rainfall and throughfall plusstemflow – was 33.21, 18.8 and 23.1% of rainfall (sites1, 2 and 3, respectively) in the period of 27 September– 28 November 2005. It was 28.4% and 31.5% in theperiod of 27 March – 4 May 2006 (sites #1 and #3).There were similar trends in soil water content dy-

namics in the forest and in the grassland (Fig. 3). The10–20 cm layer was drier than the deeper layers in all

S264 A. Hagyó et al.

Soil Profile 1

0 20 40 60 80 100 120 140 160 180

Time (days)

-55-45-35-25-15

Soil

depth

(cm

)

Soil Profile 2

0 20 40 60 80 100 120 140 160 180

Time (days)

-55-45-35-25-15

Soi

l dep

th (

cm)

Soil Profile 4

0 20 40 60

Time (days)

-55-45-35-25-15

Soi

l de

pth

(cm

)

120 140 160 180

Time (days)

-55-45-35-25-15

Soi

l dep

th (

cm)

01012162024283236404448

Soil water content (V/V%)

Fig. 3. Soil water content dynamics in Soil Profiles 1, 2 and 4, 14 April (day No. 0) – 11 October (day No. 180) 2005.

Table 2. SWC at the end of drying periods in soil profiles 1, 2, 3 and 4. F = forest, G = grassland.

Decrease in SWC Soil Profile 10–20cm 20–30cm 30–40cm 40–50cm 50–60cm

28 April – 4 May 1 (F) –1.24 –7.50 –2.67 0.39 0.20Number of days = 7 2 (F) –1.04 –0.78 –3.13 –4.04 –3.00

4 (G) –3.91 –5.74 –3.59 0.13 0.20

23 May – 3 June 1 (F) –32.27 –3.85 –4.50 –1.24 0.20Number of days = 12 2 (F) –1.70 –2.48 –4.30 –5.93 –3.19

4 (G) –20.21 –3.13 –5.87 –2.28 0

29 Aug – 10 Sept 1 (F) –21.10* –18.26 –33.32 –12.13 –8.22Number of days = 13 2 (F) –9.13 –29.14 –19.56 –15.19 –25.23

3 (F) –10.50 –19.89 –23.60 –18.39 –31.624 (G) –21.26 –11.67* –2.67* –3.91* –9.71

2 – 11 Oct 1 (F) –4.22* –5.48 –16.76 –7.17 –24.25Number of days = 10 2 (F) –1.24 –23.08 –5.28 –3.13 –1.17

3 (F) –1.30 –1.17 – –0.33 –4 (G) –7.89 –3.06* –1.70* –3.33 –

* decrease in SWC is the smallest in the grassland, – missing data because of sensor failure

profiles in spring. The SWC of the layers below 20 cmare similar in all profiles until the beginning of June,then the SWC in the forest profiles decreased more thanin the grassland profile. However, layers 30–40, 40–50and 50–60 cm are always the wettest in the grassland.We studied the decrease of SWC during four drying

periods (Table 2). In the deeper layers it was smaller inthe grassland during all the periods. There was minordrying in the two deepest layers in the first period, andin the deepest layer in the second period. The smallestdecrease could be observed in the grassland profile inthe 20–30, 30–40 and 40–50 cm layers in the third andfourth period.Recorded SWC data decreased below wilting point

during drying periods, therefore data of profile 3 arenot shown. The reason can be that the sensors becameseparated over part of their length when the soil driedand shrank, and therefore false readings were receivedbelow SWC of about 10 V/V% in the Vertisol. Thusour next research aims at the sensitivity analysis of thesensors in soil drier than wilting point.

Discussion

There were differences in soil water retention character-istics between the grassland and the forest (Fig. 1). Therelationship between rainfall and throughfall can be de-scribed in linear regressions that are similar among the

Effect of forest and grassland vegetation on soil hydrology S265

three sites (equations 1–3). The mean throughfall depthof all the studied events increased with the decrease incanopy cover. In the autumn period the mean and min-imum throughfall delay are similar at the three sites;maximum delay increased with the increase in canopycover. The mean throughfall delay of the two studiedsites differed in the spring period, when foliage startedto grow. The determined interception losses are withinthe range (11–36%) given for broadleaved forests in theliterature review by Hörman et al. (1996). Soil watercontent dynamics show similar trend in the forest soilprofiles and in the grassland profile (Fig. 3). The maindifference is that the soil dried out more in the forestthan in the grassland in the layers below 20 cm. It canbe the result of the larger water use from the deepersoil layers by trees than the grass vegetation.

References

ENGELBRECHT, B.M.J. & HERZ, H.M. 2001. Evaluation of dif-ferent methods to estimate understorey light conditions intropical forests. J. Tropic. Ecol. 17: 207–224.

HÖRMAN, G., BRANDING, A., CLEMEN, T., HERBST, M. & HIN-RICHS, A. 1996. Calculation and simulation of wind controlledcanopy interception of beech forest in Northern Germany.Agric. For. Meteorol. 79: 131–148.

MERTA, M., SEIDLER, C. & FJODOROWA, T. 2006. Estimationof evaporation components in agricultural crops. Biologia,Bratislava 61(Supl. 19): S280–S283.

PRICE, A.G. & CARLYLE-MOSES, D.E. 2003. Measurement andmodelling of growing-season canopy water fluxes in a ma-ture mixed deciduous forest stand, southern Ontario, Canada.Agr. Forest Meteorol. 119: 69–85.

RISTOLAINEN, A., TÓTH, T. & FARKAS, CS. 2006. Measurementof soil electrical properties for the characterization of the con-ditions of food chain element transport in soils. Cereal Res.Comm. 34: 159–162.

ŠTEKAUEROVÁ, V., NAGY, V. & KOTOROVÁ, D. 2006. Soil waterregime of agricultural field and forest ecosystems. Biologia,Bratislava 61(Supl. 19): S300–S304.

WRB 1998. World reference base for soil resources. 1998. FAO,ISRIC and ISSS. Rome.