Response of water and nutrient fluxes to improvement fellings in a tropical montane forest in Ecuador

Forest Ecology and Management 257 (2009) 1292–1304

Contents lists available at ScienceDirect

Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco

Response of water and nutrient fluxes to improvement fellings in a tropical montane forest in Ecuador Wolfgang Wilcke a,*, Sven Gu¨nter b, Fabian Alt a, Christiane Geißler a, Jens Boy a, Jana Knuth a, Yvonne Oelmann a, Michael Weber b, Carlos Valarezo c, Reinhard Mosandl b a b c

Professorship of Soil Geography/Soil Science, Geographic Institute, Johannes Gutenberg University Mainz, 55099 Mainz, Germany Institute of Silviculture, Technische Universita¨t Mu¨nchen, Am Hochanger 13, 85354 Freising, Germany Universidad Nacional de Loja, Direccio´n General de Investigaciones. Ciudadela Universitaria Guillermo Falconı´, sector La Argelia, Loja, Ecuador

A R T I C L E I N F O

A B S T R A C T

Article history: Received 25 September 2008 Received in revised form 24 November 2008 Accepted 24 November 2008

Management of natural forests might be one option to reduce the high deforestation rate in Ecuador. We therefore evaluated the response of water and nutrient cycles in a natural tropical montane forest to improvement fellings with the aim of favoring economically valuable target trees which will later be harvested with additional ecosystem impacts not considered here. The study was conducted at ca. 1900–2200 m above sea level in the south Ecuadorian Andes on the east-exposed slope of the east cordillera. In June 2004, one of two paired ca. 10-ha large catchments was thinned by felling 10.2% of the initial basal area (dbh  10 cm) on 30% of the catchment. The stems remained in situ. We measured ecosystem fluxes from rainfall via throughfall and stemflow to soil solution (litter leachate, soil solution at 15 and 30 cm depth) and stream flow between May 2004 and May 2005. After the fellings, soil solutions were extracted from the gaps created by the felled trees and the forest next to the gaps. We determined aboveground water fluxes by direct measurement and soil water fluxes with a budget approach. In the solutions, we measured concentrations of NH4+-N, NO3-N, total dissolved N, PO43-P, total dissolved P, Ca, Mg, K, Na, and Cl. Fluxes were calculated as volumeweighted mean (vwm) concentrations times water fluxes. Dry deposition was estimated using Cl as inert tracer. The fellings increased concentrations of N, K, Ca, and Mg in the organic layer of the resulting gaps compared with the forest next to the gaps (vwm concentrations of N: 6.4 mg l1 in the forest next to the gap/8.7 mg l1 in the gaps, K: 9.8/11, Mg: 1.8/3.0, Ca: 3.4/5.8). Lower nutrient concentrations and fluxes in the mineral soil of the gaps than in forest next to the gaps suggested that these nutrients were taken up by ground vegetation and target trees. The paired modified and undisturbed catchments had similar water and nutrient budgets. The fellings did not have a significant impact on the water and nutrient budget at the catchment scale. ß 2008 Elsevier B.V. All rights reserved.

Keywords: Catchment Montane forest Natural forest management Nutrients Water cycle

1. Introduction Tropical montane rain forests fulfill important functions including the protection of downslope areas from flooding, prevention of soil erosion and landslides, and maintenance of a constant baseflow during dry periods (Hamilton et al., 1995; Bruijnzeel, 2000). Furthermore, particularly the north Andean montane forests are extremely rich in vascular plant species (Henderson et al., 1991). By 1991 more than 90% of the original forest cover in the north Andes had been lost (Henderson et al., 1991; Hamilton et al., 1995). The north Andean state of Ecuador

* Corresponding author. Tel.: +49 6131 3924528; fax: +49 6131 3924735. E-mail address: [email protected] (W. Wilcke). 0378-1127/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2008.11.036

suffers the highest deforestation rate in South America (FAO, 2006). Therefore, income-generating alternatives for the growing population in the mountain areas of Ecuador that maintain the functions and the biological richness of the natural forest are urgently needed (De Koning et al., 1998; Aguirre, 2007). When attempting to maintain forest functions and biodiversity and nevertheless provide economic benefit to the local population, natural forest management is considered to be an option. Compared with complete transformation of a forest into agricultural land, single-tree extraction represents a low-intensity impact (Bawa and Seidler, 1998; Chazdon, 1998; Gu¨nter and Mosandl, 2003). The technical aspects of natural forest management were developed during the last century (Lamprecht, 1986; Bru¨nig, 1996; Dawkins and Philip, 1998), but up to now it is difficult to prove the sustainability of silvicultural systems. According to FAO (2006)

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

sustainability involves seven key components: extent of forest resources, biological diversity, forest health and vitality, productive functions, protective functions, socio-economic functions, legal, policy, and institutional framework. An important criterion which affects several of the key components of sustainability of land use is that nutrient losses and gains are nearly balanced and that the used ecosystem maintains its function as regulator of water and nutrient cycles. However, in many cases the current logging standards are not compatible with these ideals of sustainability and conservation (Gu¨nter, 2001; Huth and Ditzer, 2001; Hering, 2003). Thus, sustainable management of highly diverse tropical forest ecosystems requires case-specific silvicultural solutions (Hutchinson, 1993). So called ‘‘improvement fellings’’, i.e. the removal of competitors to foster growth and regeneration of trees with high timber value are among the oldest silvicultural instruments (Dawkins and Philip, 1998). Improvement fellings reduce the crown competition and stimulate nutrient uptake and thus favor target trees that can later be harvested. This harvest will likely cause further disturbances of the forest ecosystem. In small forest gaps, such as those created by improvement fellings, nutrient concentrations in soil and soil moisture increase (Silver et al., 1996; Denslow et al., 1998; Muscolo et al., 2007; Schrumpf et al., 2007). The latter improves the nutrient supply of regrowth on the clearings and of the plants next to the gaps. From strong anthropogenic or natural forest interventions e.g., by slash and burn agricultural practices (Williams et al., 1997) or hurricanes (Silver et al., 1996; Schaefer et al., 2000) it is known that nutrient export rates increase because of reduced nutrient uptake by the damaged forest and because of enhanced nutrient release by mineralization. However, Silver et al. (1996) found for a wet subtropical forest in Puerto Rico, that nutrient pools returned to pre-disturbance conditions within a three-week period after biomass harvesting or a three-month period after gap formation by a hurricane. Ostertag (1998) reported that canopy gaps in a lowland rain forest in Costa Rica had reduced fine root length and biomass compared to undisturbed forest and this effect was more pronounced in a less fertile than in a fertile soil. Only in the less fertile soil, root growth was stimulated by fertilizer application. Ostertag (1998) therefore concluded that belowground responses to canopy opening may depend on soil fertility. A suitable method to monitor water and element cycles in forests and to assess the response of forests to stress, is the catchment approach, i.e. the measurement of input, output, and internal water and element fluxes within a well defined and usually not too large catchment (Bruijnzeel, 1990; Likens and Bormann, 1995; Matzner, 2004). Catchment budgets of water and elements indicate directions of ecosystem development in response to stress such as direct anthropogenic interferences or nutrient and pollutant inputs. Prerequisites for the catchment approach are that the catchment is water-tight and all input and output fluxes of water or any studied element are registered. However, it is difficult to determine the water-conducting properties of the subsoil and underlying rock and topographic catchment borders are not necessarily the same as hydrological ones. Catchment approaches were used at several locations in the temperate zone (e.g., Ba¨umler, 1995; Likens and Bormann, 1995; Matzner, 2004) but rarely in tropical rain forests (Bruijnzeel, 1990, 2000). Our objective was to set up water and nutrient budgets of selected chemical species of paired undisturbed and modified catchments (by improvement fellings) under natural tropical montane forest in Ecuador to assess the response of the forest to moderate thinning with the purpose of favoring potentially valuable target trees. We hypothesized that the reduced nutrient use in gaps produced by the fellings improved the nutrient supply

1293

of the target trees while water and nutrient budgets at the scale of the ca. 10 ha-large catchments responded to the thinning with increased nutrient export. 2. Materials and methods 2.1. Study sites Two forested microcatchments (MC2 and MC5 in Fig. 1) between the cities of Loja (48000 S 798120 W) and Zamora (48050 S 788580 W) in the province of Zamora-Chinchipe in south Ecuador were selected (Wilcke et al., 2001). Catchment MC2 was ca. 9.1 ha and MC5 ca. 12.1 ha large. The catchments are located at ca. 1900– 2200 m above sea level (a.s.l). All catchments drain into the Rio San Francisco, which flows into the Amazon basin. Weather data are recorded since April 1998 on a clear-cut area in ridge-top position at 1952 m a.s.l. between MC2 and MC5 (Fig. 1). The mean annual temperature at 1952 m a.s.l. is 15.2 8C. The coldest months are June and July, respectively, with a mean temperature of 14.4 8C; the warmest month is November with a mean temperature of 16.1 8C. The average temperature gradient between the station at 1952 m and another station at 2927 m a.s.l. is a decrease of 0.6 8C per 100 m increase in elevation (Bendix et al., 2008). The distribution of the annual precipitation is unimodal with a maximum between April and September and without a pronounced dry season (Fleischbein et al., 2005). Mean humidity was 86% (with 90% continuously from April to June 2001) and 79% in November 2000. The mean speed of the mainly easterly winds for the period between April 1998 and April 2001 was 1.5 m s1 with a maximum of 7.9 m s1. The forest at the study site is described as ‘‘bosque siempreverde montano’’, evergreen montane forest (Balslev and Øllgaard, 2002). According to the classification of Bruijnzeel and Hamilton (2000), it is a Lower Montane Forest. Lauraceae, Rubiaceae, Melastomataceae and Euphorbiaceae are the most important tree families of the area. The tallest forest is found on lower slopes and in ravines where the canopy reaches 25 m with some emergents reaching up to 35 m. The ground flora is dominated by ferns and large herbs (Homeier et al., 2002; Paulsch, 2002; Homeier, 2004). Most of the trees have some vascular epiphytes (Paulsch, 2002). The forest has a stem density of 500–1250 stems ha1 with dbh  0.1 m (diameter at breast height,) and of 1100– 3100 stems ha1 with dbh  0.05 m (Homeier, 2004). The bedrock consists of interbedding of palaeozoic phyllites, quartzites and metasandstones (the ‘‘Chiguinda unit’’ of the ‘‘Zamora series’’ in Hungerbu¨hler, 1997). Most soils are developed in surface sediments caused by landslides and possibly periglacial drift. The soils are mainly Humic Dystrudepts (USDA-NRCS, 1998) in both microcatchments. Most soils are shallow, loamy-skeletal with high mica (a layer silicate) content (Yasin, 2001). The thickness of the organic layer ranges 2–43 cm (mean: 16 cm, Wilcke et al., 2002). 2.2. Improvement fellings The improvement fellings were realized in MC5. On the experimental area of about 3.4 ha (horizontal projection) in MC5, we registered all individuals with dbh > 20 cm of the target species Tabebuia chrysantha Nichols., Cedrela sp., Podocarpus oleifolia D.Don, Nectandra membranaceae Griseb, Hyeronima asperifolia Pax & K. Hoffmann, Hyeronima moritziana Pax & K. Hoffmann, Inga acreana Harms, Clusia ducuoides Engl., and Ficus subandina Dugand. The two species T. chrysantha and Cedrela sp. can be considered as economically most valuable. We selected 137 trees with straight trunks and regular crowns as target trees to be favored by improvement fellings of 40.3 (standard error, s.e. 6.4)

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

1294

Fig. 1. Location of the study sites.

trees per ha. The most effective crown competitors of the target trees were removed in June 2004 in order to stimulate growth and natural regeneration of the target trees. The felled trees had mostly dominant in some cases codominant social positions. The average dbh of the felled trees was 29.2 cm (s.e. 1.0). In some cases only one tree was removed to favor two target trees and only in cases when both competitors were codominant we removed two competitors for one target tree. All felled stems, branches and leaves were left on the forest floor at the place where accumulated after the fellings. Before the silvicultural treatment, the stands in MC2 and MC5 had different basal areas and stem densities (Table 1). The stand at MC5 had a slightly higher canopy openness already before the Table 1 Structural properties (means and standard errors in brackets) of the stands (trees with dbh  10 cm) in the undisturbed catchment (MC2) serving as reference area and the modified catchment (MC5) in which the improvement fellings took place before and after the improvement fellings. All data refer to the slope-corrected area. Property

Canopy openness (%) Stem density (n ha1) Basal area (m2 ha1)

Undisturbed forest

Modified forest Before the fellings

After the fellings

5.0 (0.22) 642 (47) 23.9 (1.6)

6.7 (0.14) 743 (62) 35.3 (1.8)

11.9 (1.29) 703 (59) 31.7 (1.7)

experiment than MC2, which was further increased after the thinning (Table 1). The logging intensity in MC5 corresponded to the removal of 3.6 (s.e. 0.55) m2 ha1 or 10.2% of the basal area (dbh  10 cm). Due to the fellings 10.8% (s.e. 1.9) of the surface suffered collision damages by felled crowns. This area was calculated immediately after felling as product from crown length and width for all felled trees. The structural parameters for the undisturbed catchment MC2 serving as reference area were sampled on a permanent plot of 5.2 ha for trees with dbh  20 cm cm and 20 subplots with a sampling area of 0.25 ha for trees with dbh  10 cm. Structural data for the modifed catchment MC5 originate from a permanent plot with 3.4 ha for trees with dbh  20 cm and 16 subplots with a sampling area of 0.22 ha for trees with dbh  10 cm. 2.3. Solution sampling To collect incident precipitation, two gaging stations each consisting of five fixed funnel gages were installed on deforested sites between 1900 and 1950 m a.s.l. adjacent to each catchment. In the undisturbed natural forest (definition according to FAO, 2004) in catchment MC2, three transects (at 1900–1910, 1950– 1960, and at 2000–2010 m a.s.l.) were located upslope on the lower part of a 38–708 slope (Fig. 1). In the modified catchment MC5, three of sixteen 50 m  50 m plots of the whole experimental

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

design (Gu¨nter et al., 2008) were placed randomly for detailed study (Fig. 1). Along each transect (MC2) and on each experimental plot (MC5), twenty fixed throughfall collectors were installed. Along the transects in MC2, the throughfall collectors were placed randomly along a ca. 20 m-long line. On the plots in the modified natural forest (definition according to FAO, 2004) the collectors were placed evenly spaced along the two diagonals of the experimental plot because we aimed to collect the arearepresentative throughfall of a given experimental plot. Throughfall and rainfall collectors were 2-l polyethylene sampling bottles and circular funnels with a diameter of 115 mm. Throughfall collectors were installed at a height of 0.3 m above the soil surface. To collect litter leachate, three zero-tension lysimeters were installed along each transect in the undisturbed forest and near the target tree (distance ca. 1–2 m) and in the gap created by the improvement fellings in the modified forest. The lysimeters, consisting of plastic boxes covered with a polyethylene net (0.5 mm mesh), had a surface area of 0.20 m  0.14 m and were 0.15 m high. The boxes were connected with a 1-l polyethylene sampling bottle with plastic tubes. The lysimeters were installed from a soil pit below the organic layer parallel to the surface. The organic layer was not disturbed, most roots in the organic layer remained intact. Mineral soil solution was sampled by three suction lysimeters (mullite suction cups, 1 mm  0.1 mm pore size) at each of the 0.15 and 0.3 m mineral soil depths at each sampling station with a vacuum pump. The vacuum was held permanently and adjusted to the matric potential. The solutions of all three lysimeters per depth were combined to one sample. We equipped all three transects in the undisturbed forest and the three selected experimental plots of the modified forest with such sampling stations. While in the undisturbed forest one sampling station was placed at each transect, in the modified forest we installed paired stations near the target trees (at a distance of 1–2 m) favored by improvement fellings and in the gap created by the fellings. Both, zero-tension lysimeters and suction cup lysimeters did not collect the soil solution quantitatively. Stream water samples were collected from the center of the stream above our weirs to avoid contamination by the weir material. All samples were collected weekly. Sampling in the undisturbed forest started in April 1997 (Wilcke et al., 2001) and in the modified forest in April 2004. For this study, data of a one-year period between 5 May 2004 and 4 May 2005 were considered. Between 5 May and 16 June 2004 – the phase preliminary to the improvement fellings – rainfall, throughfall, litter leachate, and mineral soil solution at the 0.15 and 0.3 m soil depths were collected near the target trees, while the instrumentation in the gaps created by the fellings was only operational on 1st July 2004 after the improvement fellings which was completed on 22 June 2004. Immediately after sampling, pH was measured with a glass electrode in unfiltered aliquots which were discarded after measurement. Another aliquot of 100 ml was filtered (ashless white ribbon paper filters, pore size, 4–7 mm, Schleicher and Schuell, Dassel, Germany), frozen at the day of sampling, and transported to Germany in frozen state. 2.4. Water and element fluxes Water fluxes by rainfall and throughfall were calculated as means of the fluxes of all individual collectors for each gaging station. For the budgets, we used the gaging stations for rainfall adjacent to each of the undisturbed and modified catchments (Fig. 1) to determine the water input into the catchments. Horizontal rain contributed < 6% to the total water input (Bendix et al., 2008) and was neglected. The altitudinal change in rainfall in the altitudinal belt covered by the study catchments which were entirely below the condensation point (Bendix et al., 2008) was

1295

considered small and also neglected. Further rainfall data were obtained from the automatic meteorologic station at 1952 m a.s.l. (Fig. 1) in high resolution (Bendix et al., 2008). Catchment-wide throughfall was derived by averaging the values of the three throughfall gaging stations in each catchment. Stemflow was neglected because it accounted for less than 2% of the total water input to the soil as measured in the undisturbed forest (Fleischbein et al., 2006). Water fluxes in soil were modeled by modifying the Soil Water Balance (SWB) model (DVWK, 1996). According to the SWB model, we determined the water fluxes out of the organic layer, the 0– 0.15 m mineral soil layer, and the 0.15–0.3 m mineral soil layer. For the organic layer fluxes were calculated as throughfall (input) minus independently determined transpiration (output) minus (or plus) change in stored water in the soil layers as calculated by the difference in water contents of the respective soil layer between two soil water content measurements with frequency domain reflectometry (FDR). For deeper soil layers, the input is the output of the overlying soil layer. We assumed direct evaporation from the soil as negligible and derived weekly transpiration rates by partitioning the annual difference between throughfall and stream flow of each catchment proportionally to the individual weeks according to the measured interception losses (i.e. rainfall-throughfall). Weekly transpiration rates were furthermore distributed among the soil layers according to the root length densities of the respective soil layer taken from Soethe et al. (2006). We assumed a linear relationship between water uptake of the vegetation and fine root abundance. We used soil water-content measurements logged by FDR probes in the lower part of the undisturbed forest at 0.1, 0.2, 0.3, and 0.4 m depths for all transects since differences in soil water content were little pronounced because of the overall wet environment of the study site (Fleischbein et al., 2006). Data gaps of soil water fluxes (because of lacking soil water contents) were substituted with the help of a regression of weekly soil water fluxes on weekly throughfall volumes. The SWB approach does not consider lateral flow occurring as response to the rare rain storm events (Goller et al., 2005). To quantify stream flow, Thompson (V-notch) weirs (908) with sediment basins were installed in the lower part of each catchment and water levels were recorded manually twice per week. At the modified catchment, the stream flow of two branches of the stream was monitored with two weirs because these two branches only merged below the experimental plots at a location which was difficult to access. The calculated stream flow of the two weirs was combined to calculate the total stream flow of the modified catchment. For the undisturbed MC2, the weir was manually calibrated with ruler, bucket, and stop watch (n = 29 manually measured streamflow-water level pairs, calibration function derived by regression of ln(stream flow) on ln(water level), r2 = 0.84), while for the modified catchment we used the water level-stream flow function published for notches with ideal 908 angles (Dyck and Peschke, 1995). Annual stream flow fluxes were calculated by assuming that the water level measured twice per week was representative for the time between one and the next measurement. This might underestimate the total stream flow because storms are likely underrepresented in our water level measurements. The rare strong storms (usually less than 5 per year, Fleischbein et al., 2006) cause overflow of our weir and cannot be registered. To assess the plausibility of our annual stream flow estimates we compared it to previously modeled five years of stream flow in MC2 with a hydrological model that was calibrated with directly measured water levels (Wilcke et al., 2008). Missing values of rainfall, throughfall, and stream flow were substituted with the help of regression functions of these fluxes on the rainfall recorded at the meteorologic station at 1952 m a.s.l. (Fig. 1).

1296

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

After export of the filtered 100-ml aliquots from Ecuador to Germany in frozen state, Ca, Mg, K, and Na concentrations were determined with flame atomic absorption spectroscopy (AAS). Furthermore, water samples were analyzed colorimetrically with a continuous flow analyzer (CFA) for concentrations of dissolved inorganic N (NH4+-N and NO3-N + NO2-N, hereafter referred to as NO3-N), total dissolved N (TDN, after UV oxidation to NO3), dissolved inorganic P (PO4-P), total dissolved P (TDP), and total dissolved Cl. Some samples had concentrations below the detection limit of the analytical methods (0.075 mg l1 for N, 0.2 mg l1 for P, 0.001 mg l1 for Ca, Mg, K, and Na). For calculation purposes, values below the detection limits were set to zero (for Ca, Mg, Na, K: <0.01%, N <1%, and P <45% of the values were below detection limit). Thus, our annual means underestimate the real concentration of chemical constituents and mean concentrations can be smaller than the detection limit.

Complete time series of the various measurement stations were compared with the non-parametric Wilcoxon test for dependent data sets. Mean properties of the treatments (undisturbed vs. modified, and for soil solutions also gaps vs. forest next to gaps in the modified forest) before and after the improvement fellings were compared with one-way ANOVA followed by StudentNewman-Keuls test (if variances were homogeneous) or GamesHowell test (if variances were not homogeneous). The same tests were used to compare dissolved ion concentrations between the two catchments or three treatments (undisturbed forest, modified forst gap and modified forest next to gap) for individual months considering the individual sampling weeks as replicate concentration measures of the given month. If only two means of two groups of variables were compared, the results of the ANOVA were identical with those of a t-test. Statistical tests of differences in pH were run with H+ concentrations. Statistical analyses were conducted with SPSS 15.0 (SPSS Inc., Chicago, IL, USA). Significance was set at p < 0.05.

2.6. Calculations and statistics

3. Results and discussion

Annual element fluxes were calculated for rainfall, throughfall, and stemflow by multiplying the respective annual volumeweighted mean (VWM) concentrations with the annual water fluxes. Element fluxes with stream flow were calculated by multiplying flow-weighted mean (FWM) concentrations with the measured annual stream flow and relating the annual flux to the surface area of the catchments (undisturbed: 9.1 ha, modified: 12.1 ha). To estimate the dry deposition and quantify canopy leaching we used the model of Ulrich (1983). The total deposition (TD) of an element i was calculated with Eq. (1).

3.1. Nutrient supply of the target trees

2.5. Chemical analyses

TDi ¼ RFDi þ DDi

(1)

Here, RFD is bulk rainfall deposition measured at the three gaging stations adjacent to the study forest (including water-soluble coarse particulate deposition) and DD is dry deposition estimated with Eq. (2). This estimate of DD includes water-soluble dry particulate and gaseous deposition. DDi ¼ ðTFDCl =RFDCl ÞRFDi  RFDi

(2)

where TFDCl represents the throughfall deposition of Cl. The quotient TFDCl/RFDCl is called deposition ratio. The canopy budget (LEA) was calculated with Eq. (3). Positive values of LEA indicate leaching, negative ones uptake of an element i by the canopy. LEAi ¼ TFDi  RFDi  DDi

(3)

If deep water leakage and changes in soil moisture and groundwater storages are negligible, the catchment water budget can be described with Eq. (4). RF ¼ SF þ ET

(4)

where RF is incident precipitation, SF is stream flow, and ET is evapotranspiration (i.e. the sum of Ei, Es and Et, see below). The changes in soil moisture and groundwater storages can be neglected in our study because of the budgeting interval of a whole year. The rainfall interception losses (Ei) were derived from Eq. (5). Ei ¼ RF  TF

(5)

where TF is throughfall. Soil evaporation (Es) from the forest floor was neglected and transpiration losses (Et) were calculated with Eq. (6). Et ¼ TF  SF

(6)

One aim of the improvement fellings was to reduce nutrient uptake of competing trees and thereby increase the nutrient supply of the target trees. To test if this aim was reached we compared the chemical composition of the soil solution below the organic layer and at the 0.15 and 0.30 m mineral soil depths in the gaps created by the improvement fellings and in the forest next to the gaps in the modified catchment. The pH of litter leachate (undisturbed forest: 4.6 and modified forest: 4.7) was not significantly different and did not respond to the fellings as indicated by almost unchanged pH after the fellings (undisturbed forest: 4.5, modified forest next to gaps: 4.4, gaps: 4.6). The mineral soil solution at 0.15 m depth had the same pH (4.4) in undisturbed and modified forest and was slightly more acid in the modified forest (4.5) than in the undisturbed forest (4.7) at 0.30 m depth before the fellings. After the fellings, the pH remained unchanged in the undisturbed forest (0.15 m: 4.4/0.30 m: 4.7) and became slightly more acid in the forest next to the gaps of the modified catchment (4.2/4.2) but increased significantly in the gaps (4.8/4.9) possibly indicating chemical reduction processes caused by increased soil moisture. This is supported by the finding that at 0.15 m soil depth the mean matric potential of our observation period was 6.8 MPa in the gap, 7.9 MPa in the forest next to the gap, and 10 MPa in the undisturbed forest reflecting a gradient of soil moisture along this line (own unpublished results). Increased soil moisture in gaps compared to undisturbed forest was also reported by Muscolo et al. (2007). There was no temporal trend in the course of the concentrations of all studied nutrients in litter leachate (Fig. 2). The few extreme values are responses to dry periods particularly at the end of September 2004, in mid November 2004, and in the second half of January 2005 where weekly rainfall was 3.5–6.8 mm which is far below the mean weekly rainfall of 57 mm. Increases in ion concentrations are partly attributable to concentration effects which is supported by the simultaneous increase in the concentrations of Cl which can be considered as an inert tracer indicating concentration/dilution effects in the organic layer (Fig. 2). Concentration increases, however, can also be explained by enhanced mineralization because it is well known that C and N mineralization rates increase for a few days following the rewetting of a dry soil (Birch, 1958; Bloem et al., 1992; Cui and Caldwell, 1997; Franzluebbers et al., 1994) During the preliminary phase (05/05–16/06/2004), the litter leachates of the experimental stations in the modified forest had

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

Fig. 2. Mean course of the concentrations of (a) total dissolved N, (b) total dissolved P, (c) K, (d) Mg, and (e) Cl concentrations in the litter lysimeters between 05/05/ 2004 and 04/05/2005 of the undisturbed (MC2), the forest next to the gaps in the modified catchment (MC5) and in the gaps created by the improvement fellings (MC5 gap, n = 3).

higher volume-weighted mean (vwm) element concentrations except for PO4-P, TDP, and Na (NH4-N: 0.55 mg l1, NO3-N: 5.0, DON: 1.4, TDN: 6.9, PO4-P: 0.21, DOP: 0.03, TDP: 0.24, Cl: 2.0, K: 8.1, Na: 0.14, Mg: 1.7, Ca: 3.2) than those of the undisturbed forest (NH4-N: 0.37 mg l1, NO3-N: 1.7, DON: 1.0, TDN: 3.1, PO4-P: 0.23, DOP: 0.01, TDP: 0.24, Cl: 1.3, K: 5.9, Na: 0.28, Mg: 1.3, Ca: 2.3, Fig. 2a–d). The same was true for the one solution sample we analyzed from each of the 0.15 and 0.30 m mineral soil depths before the fellings (Fig. 3). Thus, the soil solution in the modified forest was richer in nutrients (except for P) than that in undisturbed forest. The lack of differences in P concentrations might be related with the consistently low P concentrations in soil solution frequently below the detection limit. The higher nutrient concentrations in the modified forest before the fellings than in the

1297

undisturbed forest might be related with the higher initial canopy openness of the modified compared with the undisturbed forest (Gu¨nter et al., 2008) resulting in a higher mineralization rate because of higher soil temperatures. As litterfall was not different between the modified (12.1 Mg ha1 yr1) and the undisturbed forests (11.3 Mg ha1 yr1, own unpublished data), different nutrient concentrations in the litter leachates between the modified and undisturbed forests cannot be explained by a different mass of decomposing litter. After the fellings, the concentration levels in litter leachate of the undisturbed forest and the forest next to the gaps in the modified catchment remained at a similar level as before (Fig. 2a– d), while the concentrations in the gaps created by the fellings increased for all studied chemical constituents of the litter leachate compared to the undisturbed state before the fellings (Fig. 2a–c). This increase was significant for NO3-N (mean vwm concentrations of the period 30/06/2004–04/05/2005 of three experimental stations in the forest next to the gaps: 4.6 mg l1/6.6 mg l1 in the gap), TDN (6.4/8.7), and Mg (1.8/3.0). Similar findings were reported for 10-yr old clearings in the lower forest belt at Mt. Kilimanjaro, Tanzania where Schrumpf et al. (2007) observed higher base metal (K, Mg, Ca) and N concentrations in litter leachate and mineral soil solution at 0.15 m depth in the clearing than in mature natural forest. Increased nutrient concentrations in topsoils of natural or anthropogenic treefall gaps were also observed in wet tropical forests of India (Chandrashekara and Ramakrishnan, 1994) and Costa Rica (Denslow et al., 1998). When analyzing the differences in nutrient concentrations for individual months, it turned out that only in the first month after the improvement fellings, concentrations of NO3-N, Ca, Mg, and K were significantly higher in the gaps than in the forest next to the gaps in the modified catchment and in the undisturbed forest. This was also true for K in the third month after the improvement fellings. However, with the exception of K in the third month after the improvement fellings, there were no further significant differences in nutrient concentrations of the litter leachates from the second to 10th month after the improvement fellings. Thus, there was a short response of the nutrient concentrations in the first month after the fellings which quickly disappeared. Silver et al. (1996) observed a similar quick return of nutrient pools in soil to pre-disturbance levels after biomass harvesting in a wet subtropical forest in Puerto Rico of only three weeks. The reason for increased nutrient concentrations in the gaps compared with the forest next to the gaps might be higher mineralization rates because of increased soil temperatures (Muscolo et al., 2007). Joslin and Wolfe (1993) and Lukewille and Wright (1997) reported warming-induced increases in N concentrations of soil solutions and runoff for N-rich environments. Increased nutrient concentrations might also be attributable to the fresh plant debris deposited in the gaps because of the fellings (Chandrashekara and Ramakrishnan, 1994; Denslow et al., 1998). Another reason might be reduced nutrient uptake by the vegetation after the fellings. In the western Ghats in India, Chandrashekara and Ramakrishnan (1994) attributed higher N, P, and Mg concentrations in soil one year after selection fellings partly to the disturbance of the ground vegetation in the gaps created by the fellings, particularly in the part of the gap impacted by the falling trees. Consequently, the nutrient availability in the gap next to the target trees was increased which should be favorable for the target trees if they use already present roots in the gap area or expand their roots laterally into the gap area. The increased nutrient concentrations after dry spells observed in the litter leachate can also be seen in the course of nutrient concentrations of the mineral soil solutions (Fig. 3). After the fellings, TDN (Fig. 3a and e; mainly because of increased NO3-N concentrations, not shown) and base metal concentrations (K, Ca,

1298

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

Fig. 3. Mean course of the concentrations of (a) and (e) total dissolved N, (b) and (f) total dissolved P, (c) and (g) K, and (d) and (h) Mg concentrations in soil solution at 0.15 m [(a)–(d)] and 0.30 m mineral soil depth [(e)–(h)] between 05/05/2004 and 04/05/2005 of the undisturbed forest (MC2), the forest next to the gaps in the modified catchment (MC5) and in the gaps created by the improvement fellings (MC5 gap, n = 3).

not shown), and partly Mg, Fig. 3d and e) were elevated in the soil solutions of the forest next to the gaps in the modified catchment at both depths for about six months while in the gap area nutrient concentrations in soil solution had approximately the same level throughout the one-year observation period. As the elevated nutrient concentrations were only observed in the forest next to the gaps of the modified catchment and had a similar level during the first ca. six months after the fellings, it seems unlikely that the fellings were the reason for the elevated nutrient concentrations. In the mineral soil, we observed a contrasting result to the litter leachate (Fig. 3). Both at 0.15 and 0.3 m soil depth, the a priori higher nutrient concentrations in the modified forest than in the undisturbed forest remained at the same level in the forest next to the gaps of the modified catchment after the fellings, while the mineral soil solutions in the gap became poorer in nutrients after the fellings. The concentration differences between the forest next to the gaps and the gaps were significant for NO3-N, TDN, and K at both mineral soil depths. For N, this concentration decrease in the gaps might partly be explained by enhanced denitrification because of the higher soil moisture. The lower concentrations of the base metals in the mineral soil solutions of the gaps compared with the forest next to the gaps might be the consequence of

microbial immobilization because microbial growth should be stimulated by the increased soil temperatures in the gaps. Another explanation could be that the disturbance favors nutrient uptake of plants mainly rooting in the mineral soil which are usually underrepresented in the undisturbed forest where most roots are found in the organic layer (Soethe et al., 2006). The finding of lower nutrient concentrations in the soil solution of the gaps is unexpected and in spite of the possible explanations above not fully understood. However, our results clearly indicate that the creation of the gap by our improvement fellings resulted in increased mineralization of the organic layer but nevertheless not in increased nutrient concentrations in the mineral soil solution and thus also not in increased nutrient losses by leaching. We conclude that with the elevated nutrient concentrations in the litter leachates of the gaps the aim to improve nutrient supply of the target trees has likely been reached provided that the favored trees can use the nutrients released in the gap area. 3.2. Water budget To assess the impact of the improvement fellings on the water cycle we compared all elements of the water cycle and the total water budgets of undisturbed and modified catchments.

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

During the preliminary phase (05/05–16/06/2004), the undisturbed catchment received 588 mm of rainfall and the modified catchment 655 mm. The weekly rainfall at both gaging stations was closely correlated (r = 0.99, p < 0.05). Throughfall in the undisturbed catchment during this phase was 314 mm (=54% of rainfall) and in the modified catchment 419 mm (=64% of rainfall). Again, weekly throughfall in both catchments was closely correlated (r = 0.98, p < 0.05). The higher percentage of rainfall reaching the soil in the modified catchment as throughfall can be attributed to the higher initial canopy openness in the modified than in the undisturbed catchment and the additional opening of the canopy by the fellings (Gu¨nter et al., 2008). Cumulative stream flow of the preliminary phase was 195 mm (=33% of rainfall) in the undisturbed and 307 mm (=47%) in the modified catchment. The percentage of the rainfall leaving the undisturbed catchment as stream flow was lower than on average of 1998–2002 (46%, Fleischbein et al., 2006). The higher stream flow in the modified than the undisturbed catchment coincided with the higher throughfall. Weekly stream flow correlated significantly between both catchments (r = 0.70, p < 0.05) but less closely than weekly rainfall and throughfall. Thus, the two studied catchments showed a slightly different water cycle before the natural forest management measure which needs to be considered when interpreting potential responses to the improvement fellings. The improvement fellings did not have a significant impact on weekly throughfall and stream flow in the experimental part of the modified catchment in spite of the creation of gaps (Fig. 4a and b). There was also no temporal trend of the percentage of weekly throughfall in response to the improvement fellings (66  s.d. 16% before and 69  43% after the fellings, see also Fig. 4a) as could be expected if the canopy gaps created by the fellings were reclosed in the time after the fellings. Thus, the fellings’ impact on the soil water input was so small that we were unable to detect it with our approach. Consistent with the non-detectable change in soil water input, there was no response of the stream flow to the improvement fellings (Fig. 4b). As a consequence of the higher throughfall in the modified than in the undisturbed forest, the water fluxes in soil were consistently higher between the organic layer and 0.30 m mineral soil depth (Fig. 4c and d) although the differences became increasingly smaller with increasing soil depth. There was again no detectable response of the water fluxes in soil to the fellings. As there were no significant changes in input, internal, and output water fluxes at the catchment level as a consequence of the improvement fellings, we kept to our usual hydrological year (May–April, Fleischbein et al., 2006), which allows direct comparisons with previous results from the undisturbed catchment (MC2, Wilcke et al., 2001, 2008a, b, Fleischbein et al., 2006). Table 2 summarizes the water budget of the hydrological year 05/ 05/2004–04/05/2005 together with comparison values for the

1299

Fig. 4. Mean course of weekly water fluxes of (a) rainfall and throughfall (b) stream flow, (c) litter leachate, (d) soil solution between 05/05/2004 and 04/05/2005 of the undisturbed (MC2) and modified catchments (MC5, n = 3 except for rainfall and stream flow: n = 1).

Table 2 Water budget of catchments under undisturbed forest (MC2) and modified forest (MC5) for the hydrological year 05/05/2004–04/05/2005 and comparison values from the literature for the same study site. MC2 (mm)

MC5 (mm)

Fleischbein et al. (2006), Wilcke et al. (2008a)a

2970 1631 n.a.b 1339 1273 1195 1133 1081 550 1889

3065 1903 n.a.a 1161 1471 1356 1259 1268 635 1797

2410–2610 1220–1580 23–25 860–1290 n.a.a n.a.a n.a.a 960–1100 290–570 1310–1580

Range (mm) Incident rainfall Throughfall Stemflow Interception loss Litter leachate Soil (0.15 m depth) Soil (0.30 m depth) Stream flow Transpiration Evapotranspiration a b

Range of the mean of five hydrological years (1998–2003) from three ca. 10-ha catchments at the same study site in Ecuador (including MC2). ‘‘n.a.’’ is not available.

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

1300

undisturbed forest (MC2, Fleischbein et al., 2006; Wilcke et al., 2008a). Stemflow was not measured in our study but was negligibly small in previous work (<2% of rainfall, Fleischbein et al., 2006). The result of the hydrological year 2004/2005 further confirms the previous findings that our study area is characterized by a high total evapotranspiration at the upper end of reported values from tropical lowland forests of 886–1606 mm yr1 (Bruijnzeel, 2000). The total evapotranspiration is mainly driven by the high interception losses while the annual transpiration is comparable to other tropical mountain forests (170–845 mm, Bruijnzeel, 2000). Differences between the two studied catchments already existed before the natural forest management measure and remained unaffected by the improvement fellings. 3.3. Nutrient budget To set up complete nutrient budgets of the two study catchments we determined all input and output fluxes which were calculated as the product of the concentrations and the above presented water fluxes, except for the dry deposition where only fluxes were estimated. Therefore, we first present the chemical properties of rainfall, throughfall and stream water. The vwm pH of rainfall was 4.6 in the undisturbed catchment and 4.5 in the modified catchment. The difference was small but statistically significant. The pH below the equilibrium value with atmospheric CO2 of 5.7 can be explained by deposition of organic acids released by the canopy of the adjacent rain forest, by local fires as e.g., used to establish and refresh pastures and by acid emissions of the forest fires in the Amazon basin (Forti and Neal, 1992; Boy et al., 2008a). The slightly lower pH in the modified than the undisturbed catchment might be explained by higher acid input into the modified catchment because of the local burning activity in the observation period and transport of the gaseous acid-forming compounds NOx and SO2 released by vegetation fires to our modified forest site while the fly ashes are deposited close to the

burning site outside our study area. The vwm concentrations of all studied chemical species in rainfall were not significantly different between the two studied catchments during the whole monitored period (Table 3). The vwm pH of throughfall in the undisturbed forest was slightly but again significantly more acid (5.6) than in the modified forest (5.7) and by one unit higher than that of the rainfall. The latter is the consequence of buffering resulting in ion leaching from the canopy particularly of base metals (Parker, 1983). Differences in the concentrations of the chemical species in throughfall were small between the two studied catchments but after the improvement fellings TDN, NH4-N, Na, and Mg concentrations were significantly higher in throughfall of the modified than the undisturbed forest (Tab. 2, Fig. 5). Although similar differences were not observed during the preliminary phase, we doubt that the significantly increased nutrient concentrations in throughfall of the modified forest compared to the undisturbed forest can be attributed to the fellings because the largest impact on the quality of throughfall is to be expected immediately after the fellings. However, Figure 5 shows that the fellings did not have a visible immediate effect. In stream water, the flow-weighted mean (fwm) pH was significantly more acid in modified (6.3 before and 6.4 after the improvement fellings) than in the undisturbed catchment (6.7 and 6.9). There was no detectable response of the pH in stream water to the fellings. The concentrations of all studied chemical constituents were not significantly different between the two studied catchments except Na, which had a significantly higher concentration in stream water of the undisturbed than the modified catchment (Table 3). There was no significant response of the nutrient concentrations in stream water to the fellings (Fig. 6). Thus, similar to the water fluxes, the natural forest management measure did not have a significant impact on the concentrations of the chemical species in throughfall and stream water. As expected from the above presented results for water fluxes and nutrient concentrations, the nutrient fluxes reflected a priori

Table 3 Volume-weighted mean of the concentrations of selected chemical constituents of the ecosystem fluxes in undisturbed forest (MC2) and in the forest next to the gaps created by the fellings (MC5) and in the gaps (MC5 gap) of the modified forest before and after the fellings. RF is rainfall, TF throughfall, LL litter leachate, SS15 soil solution at 0.15 m mineral soil depth, SS30 soil solution at 0.30 m soil depth, and SW stream water. NH4-N (mg l1)

NO3-N (mg l1)

DON (mg l1)

TDN (mg l1)

PO4-P (mg l1)

TDP (mg l1)

K (mg l1)

Na (mg l1)

Mg (mg l1)

Ca (mg l1)

Cl (mg l1)

RF

MC2 MC5

Whole Time

0.18 0.24

0.28 0.23

0.22 0.35

0.67 0.81

0.01 0.01

0.01 0.02

0.26 0.19

0.25 0.29

0.04 0.04

0.11 0.11

0.66 0.58

TF

MC2 MC5 MC2 MC5

Before Before After After

0.24 0.26 0.22 0.32

0.31 0.14 0.69 0.75

0.67 0.60 0.57 0.56

1.2 1.0 1.5 1.6

0.07 0.08 0.20 0.20

0.09 0.09 0.20 0.21

4.3 4.3 7.4 7.6

0.17 0.16 0.27 0.54

0.25 0.23 0.47 0.54

0.38 0.31 0.66 0.72

0.97 0.94 1.9 1.7

LL

MC2 MC5 MC2 MC5 MC5 gap

Before Before After After After

0.37 0.55 0.34 0.52 0.77

1.7 5.0 3.4 4.6 6.6

1.0 1.4 1.0 1.3 1.4

3.1 6.9 4.7 6.4 8.6

0.23 0.21 0.28 0.29 0.64

0.24 0.24 0.29 0.32 0.52

5.9 8.1 6.5 9.8 11

0.28 0.14 0.39 0.39 0.46

1.3 1.7 1.5 1.8 3.0

2.2 3.2 2.7 3.4 5.8

1.2 2.0 2.2 2.5 2.9

SS15

MC2 MC5 MC2 MC5 MC5 gap

Before Before After After After

0.21 0.36 0.13 0.19 0.18

1.6 5.8 1.3 5.6 1.8

0.71 0.56 0.45 0.46 0.42

2.5 6.7 1.9 6.2 2.4

0.00 0.01 0.10 0.02 0.01

0.01 0.03 0.10 0.02 0.01

0.39 6.1 1.1 5.4 1.5

0.13 0.19 0.19 0.29 0.28

0.37 1.3 0.53 1.1 0.84

0.25 3.0 1.1 2.0 1.46

1.2 1.0 0.98 1.1 1.1

SS30

MC2 MC5 MC2 MC5 MC5 gap

Before Before After After After

0.21 0.39 0.12 0.19 0.22

0.48 6.8 0.42 6.8 2.3

0.70 0.67 0.35 0.19 0.38

1.4 7.8 0.88 6.8 2.9

0.02 0.01 0.03 0.02 0.02

0.01 0.01 0.03 0.02 0.02

0.09 3.8 0.27 4.5 1.5

0.15 0.40 0.26 0.46 0.33

0.23 1.4 0.25 0.96 0.95

0.48 2.6 0.41 2.0 2.2

0.40 1.8 0.54 1.3 1.2

SW

MC2 MC5 MC2 MC5

Before Before After After

0.04 0.03 0.09 0.07

0.14 0.04 0.11 0.16

0.13 0.09 0.12 0.12

0.31 0.14 0.29 0.35

0.01 0.01 0.02 0.01

0.04 0.01 0.03 0.02

0.28 0.15 0.24 0.27

2.7 1.1 2.4 1.3

0.30 0.07 0.35 0.13

0.37 0.06 0.43 0.13

0.37 0.43 0.59 0.54

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

1301

Fig. 5. Mean course of the concentrations of (a) total dissolved N, (b) total dissolved P, (c) K, and (d) Mg concentrations in throughfall between 05/05/2004 and 04/05/2005 of the undisturbed (MC2) and modified catchments (MC5, n = 3).

Fig. 6. Course of the concentrations of (a) total dissolved N, (b) total dissolved P, (c) K, and (d) Mg concentrations in stream water between 05/05/2004 and 04/05/2005 in the undisturbed (MC2) and modified catchments (MC5, for MC5 mean of the eastern and western weirs).

differences between the paired catchments but did not respond to the improvement fellings (see Figures 4 and 5). Therefore, we only show the annual fluxes and total budgets for the undisturbed and modified catchments for the whole hydrological year (05/05/ 2004–04/05/2005) in Table 4. As the dry deposition was estimated based on the assumption that Cl can be used as inert tracer we first had to make sure that this prerequisite was met. This was done by comparing the measured Cl concentrations with Cl concentrations expected from a pure concentration effect during the passage of Cl from throughfall through the organic layer and the mineral soil to the stream. The measured vwm Cl concentrations in the litter leachates of the whole budgeting period of one year (2.0 mg l1 in the undisturbed forest and 2.4 mg l1 in the modified forest)

agreed well with the 2.2 and 2.3 mg l1, respectively, expected as pure concentrations effect (calculated from the difference between the throughfall and litter leachate fluxes). However, in the mineral soil vwm Cl concentrations decreased with increasing depth (1.0 mg l1 at 0.15 m depth and 0.52 mg l1 at 0.3 m depth in the undisturbed forest and 2.1 and 1.4 mg l1, respectively in the modified forest) and were similarly low in stream water (0.55 mg l1 in the undisturbed forest/0.51 mg l1 in the modified forest) as in rainfall (0.66/0.58). Thus, Cl met the assumption of an inert tracer in the organic part of the soil while the mineral soil was a sink of Cl. Therefore, the results back the use of Cl as an inert tracer for our canopy budget while the decrease in Cl concentrations in the mineral soil and stream water precluded a water budgeting effort based on Cl concentrations.

1302 Table 4 Annual element fluxes in undisturbed (MC2) and modified forest (MC5, 05/05/2004–04/05/2005). RFD is rainfall deposition, DD dry deposition, LEA leaching from the canopy, TFD throughfall deposition, LL litter leachate, SS15 soil solution at 0.15 m mineral soil depth, SS30 soil solution at 0.30 m soil depth, and SW stream water. NO3-N (kg ha1 yr1)

DON (kg ha1 yr1)

MC2—undisturbed forest RFD 5.3 DD 2.4 LEA 4.1 TFD 3.6 LL 4.4 SS15 1.7 SS30 1.5 SW 0.84 Budget 6.8

8.4 3.7 2.1 10 39 17 4.8 1.2 11

6.6 2.9 0.12 9.6 13 6.0 4.7 1.3 8.2

MC5—modified forest RFD 7.4 DD 3.9 LEA 5.5 TFD 5.8 LL 7.8 LL gap 11 SS15 3.0 SS15 gap 2.9 SS30 2.8 SS30 gap 3.1 SW 0.80 Budget 11

7.0 3.7 1.1 12 69 92 76 35 85 40 1.6 9.0

11 5.7 5.7 11 20 20 6.5 6.1 3.5 5.5 1.4 15

PO4-P (kg ha1 yr1)

TDP (kg ha1 yr1)

K (kg ha1 yr1)

Na (kg ha1 yr1)

Mg (kg ha1 yr1)

Ca (kg ha1 yr1)

Cl (kg ha1 yr1)

20 8.8 5.5 23 56 24 11 3.2 25

0.23 0.10 2.5 2.8 3.4 1.0 0.33 0.22 0.11

0.23 0.10 2.6 3.0 3.6 1.0 0.35 0.38 0.04

7.6 3.4 101 112 81 12 2.7 2.7 8.3

7.3 3.2 6.5 4.0 4.7 2.1 2.7 27 16

1.3 0.6 5.2 7.0 19 6.0 2.8 3.7 1.9

3.4 1.5 5.1 10 33 12 4.8 4.5 0.4

19 8.6 0.0 28 26 12 5.8 6.0 22

25 13 9.7 28 96 122 85 44 88 48 3.7 34

0.35 0.19 2.7 3.3 4.1 8.1 0.19 0.13 0.19 0.19 0.14 0.40

0.47 0.25 2.8 3.6 4.5 6.8 0.28 0.19 0.22 0.21 0.18 0.54

5.8 3.1 122 131 139 149 75 33 55 25 3.1 5.8

8.7 4.6 4.7 8.7 5.0 5.7 3.6 3.6 5.7 4.3 16 2.6

1.3 0.7 6.9 8.9 26 40 15 13 13 13 1.5 0.6

3.5 1.8 6.6 12 50 77 30 24 26 28 1.5 3.9

18 9.4 2.6 30 35 40 15 14 18 16 6.5 21

TDN (kg ha1 yr1)

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

NH4-N (kg ha1 yr1)

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

Total deposition (i.e. rainfall + dry deposition) was similar for both catchments. The modified catchment, however, received a significantly higher N deposition than the undisturbed catchment. This is in line with the more acid pH of the rainfall at the modified catchment and further supports the assumption that the modified catchment was more exposed to acid inputs as nitric acid emitted by local fires. Nitric acid has been shown to be the major acid deposited at our study site as result of Amazonian forest fires (Boy et al., 2008a). The total deposition of N ranged at the upper end of the records at the same study site for 1998–2003, that of P at the lower end and those of the base metals well within the range reported by Wilcke et al. (2008b). While in both studied catchments the base metals (except Na) and P were leached on balance from the canopy as consequence of H+ buffering, N species (except DON in the undisturbed forest) and Na were retained in the canopy suggesting that these compounds were taken up by the plants (trees and epiphytes) and phyllosphere organisms or immobilized in the soil-like accumulations of the canopy (Ulrich, 1983; Nadkarni et al., 2004; Clark et al., 2005). The fluxes of all studied nutrients at all depths (litter leachate and mineral soil solutions at the 0.15 and 0.30 m depths) were consistently higher in the modified than in the undisturbed forest (Table 4). However, this was not attributable to the management measure because nutrient concentrations were already elevated before the fellings and did not respond to the fellings (see Figures 2 and 3). Differences in nutrient fluxes were driven by higher water fluxes in soil of the modified than of the undisturbed forest because of the higher throughfall as a consequence of the more open canopy in the modified forest. As for the nutrient concentrations, there were systematic differences in nutrient fluxes between the forest next to the gaps and the gaps created by the fellings. The nutrient fluxes with litter leachate were consistently higher in the gaps while the reverse was true for the mineral soil solutions at both depths. Differences were significant for NO3-N, TDN, and K (mineral soil only). Thus, the higher release of nutrients by the enhanced mineralization was leached into the mineral soil where it was retained because of the possible reasons outlined above. Export fluxes of N, K, and Cl were similar in both studied catchments, while export fluxes of P, and the other base metals (Ca, Mg, Na) were lower in the modified than in the undisturbed catchments (Table 4). Thus, the higher fluxes of all studied chemical constituents in soil solutions of the modified catchment did not result in elevated element losses. This is different to responses of tropical catchments which are affected by slash and burn practices (Williams et al., 1997) or strongly damaged by a hurricane like Hugo in 1989 in the Caribbean (Schaefer et al., 2000). These severe disturbances usually result in strongly increased nutrient export from forested catchments. The net budgets of N, K, Ca, and Cl in both catchments and of P and Mg in the modified catchment were positive indicating that these elements were retained in the vegetation and/or soil. It has, however, to be considered that the N budget is incomplete because we did not measure gaseous N fluxes. In contrast, Na showed negative net budgets in both catchments. As the most mobile of the studied cations, Na is likely strongly leached in the acid soils of the study area. The same was true for Mg in the undisturbed forest. The slight net loss of TDP in undisturbed forest can be attributed to P release in the subsoil because of weathering. Boy et al. (2008b) observed a decoupled upper P cycle between the vegetation and the upper soil and a second P cycle in the deep soil which is not reached by plant roots. However, the reliability of our P results is limited by the fact that many P concentrations were near or below the detection limit. We conclude that our forest management did not have an impact on the nutrient fluxes in the modified forest at the catchment scale.

1303

4. Conclusions The improvement fellings resulted in increased nutrient concentrations in litter leachates of the gaps near the target trees, while nutrient concentrations in the litter leachates next to the gaps remained unchanged after the fellings. Thus, nutrient supply in the organic layer, where most plant roots are located was locally improved and should favor the target trees. However, decreased N concentrations in the mineral soil of the gap area compared with the undisturbed area may also be explained by N losses via denitrification. As nutrient export with stream flow was even lower in the modified than in the undisturbed catchment there are no indications of enhanced nutrient leaching after the fellings. The fellings did not have a significant impact on water fluxes, nutrient concentrations in all ecosystem solutions and the resulting nutrient fluxes except for the soil solutions in the gaps created by the fellings. We therefore conclude that the removal of 10.2% of the basal area for dbh  10 cm on ca. 30% of the catchment area can be considered as sustainable with respect to water and nutrient cycles at the catchment scale. However, the later logging and removal of the target trees will produce further disturbances of the ecosystem which require a thorough analysis of effects on water and element cycles before the sustainability of the suggested forest management can be fully judged. Acknowledgements We thank Nature and Culture International (NCI) in Loja, Ecuador for access to the forest and the facilities at the research station and for much other support and the Instituto Ecuatoriano Forestal de Areas Naturales y Vida Silvestre (INEFAN) for the permission to conduct this study. We are indebted to the German Research Foundation (DFG) for funding this project (DFG 402/2 TP B10A, WI 1601/5-2) and to the Excellence Cluster ‘‘Geocycles’’ of the State of Rheinland-Pfalz, Germany for contributing to our laboratory infrastructure. References Aguirre, N., 2007. Silvicultural contributions to the reforestation with native species in the tropical mountain rainforest region of south Ecuador. Dissertation. Lehrstuhl fu¨r Waldbau und Forsteinrichtung, Wissenschaftszentrum Weihenstephan fu¨r Erna¨hrung, Landnutzung und Umwelt, TU Mu¨nchen, Germany. Balslev, H., Øllgaard, B., 2002. Mapa de vegetacio´n del sur de Ecuador. In: Aguirre, M.Z., Madsen, J.E., Cotton, E., Balslev, H. (Eds.), Bota´nica Austroecuatoriana. Estudios sobre los recursos vegetales en las provincias de El Oro, Loja y ZamoraChinchipe. Ediciones Abya-Yala, Quito, Ecuador, pp. 51–64. Ba¨umler, R., 1995. Dynamik gelo¨ster Stoffe in verschiedenen Kompartimenten kleiner Wassereinzugsgebiete in der Flyschzone der Bayerischen Alpen—Auswirkungen eines geregelten forstlichen Eingriffs. Bayreuther Bodenkundliche Berichte 40, University of Bayreuth, Germany. Bawa, K.S., Seidler, R., 1998. Natural forest management and conservation of biodiversity in tropical forests. Conserv. Biol. 12, 46–55. Bendix, J., Rollenbeck, R., Richter, M., Fabian, P., Emck, P., 2008. Climate. In: Beck, E., Bendix, J., Kottke, I., Makeschin, F., Mosandl, R. (Eds.), Gradients in a tropical mountain ecosystem of Ecuador. Ecological Studies 198. Springer, Berlin, Germany, pp. 63–73. Birch, H.F., 1958. The effect of soil drying on humus decomposition and nitrogen availability. Plant Soil 10, 9–31. Bloem, J., Deruiter, P.C., Koopman, G.J., Lebbink, G., Brussaard, L., 1992. Microbial numbers and activity in dried and rewetted arable soil under integrated and conventional management. Soil Biol. Biochem. 24, 655–665. Boy, J., Rollenbeck, R., Valarezo, C., Wilcke, W., 2008a. Amazonian biomass burningderived acid and nutrient deposition in the north Andean montane forest of Ecuador. Glob. Biogeochem. Cycle. 22, GB1027, doi:10.1029/2007GB002960. Boy, J., Valarezo, C., Wilcke, W., 2008b. Water flow paths in soil control element exports in an Andean tropical montane forest. Eur. J. Soil Sci. 59, 1209–1227. Bruijnzeel, L.A., 1990. Hydrology of moist tropical forests and effects of conversion: a state of knowledge review. Unesco, Division of Water Sciences, International Hydrological Programme, Paris, France. Bruijnzeel, L.A., 2000. An ecohydrological perspective of mountain cloud forests. In: Gladwell, J.S. (Ed.), Proceedings of the Second International Colloquium on Hydrology and Water Management in the Humid Tropics, UNESCO, Paris and CATHALAC, Panama, pp. 329–359.

1304

W. Wilcke et al. / Forest Ecology and Management 257 (2009) 1292–1304

Bruijnzeel, L.A., Hamilton, L.S., 2000. Decision Time for Cloud Forests. In: IHP Humid Tropics Programme Series, 13, IHP-UNESCO, Paris. Bru¨nig, E.F., 1996. Conservation and management of tropical rainforests: an integrated approach to sustainability. CABI, Wallingford, UK. Chandrashekara, U.M., Ramakrishnan, P.S., 1994. Successional patterns and gap phase dynamics of a humid tropical forest of the western Ghats of Kerala, India – Ground vegetation, biomass, productivity, and nutrient cycling. Forest Ecol. Manage. 70, 23–40. Chazdon, R.L., 1998. Tropical Forests–Log´ em or leave´ em? Science 281, 1295– 1296. Clark, K.L., Nadkarni, N.M., Gholz, H.L., 2005. Retention of inorganic nitrogen by epiphytic bryophytes in a tropical montane forest. Biotropica 37, 328–336. Cui, M.Y., Caldwell, M.M., 1997. A large ephemeral release of nitrogen upon wetting of dry soil and corresponding root responses in the field. Plant Soil 191, 291– 299. Dawkins, H.C., Philip, M.S., 1998. Tropical moist forest silviculture and management. CABI, Wallingford, UK. De Koning, G.H.J., Veldkamp, A., Fresco, L.O., 1998. Land use in Ecuador: a statistical analysis at different aggregation levels. Agric. Ecosyst. Environ. 70, 231–247. Denslow, J.S., Ellison, A.M., Sanford, R.E., 1998. Treefall gap size effects on above- and below-ground processes in a tropical wet forest. J. Ecol. 86, 597–609. DVWK (Deutscher Verband fu¨r Wasserwirtschaft und Kulturbau), 1996. Ermittlung der Verdunstung von Land- und Wasserfla¨chen. DVWK-Merkbla¨tter zur Wasserwirtschaft 238:134 pp. Dyck, S., Peschke, G., 1995. Grundlagen der Hydrologie. Verlag fu¨r Bauwesen GmbH, Berlin, Germany. FAO, 2004. Global forest resources assessment update 2005. Terms and Definitions. Working Paper 83/E, Rome, Italy. FAO, 2006. Global Forest Resources Assessment 2005. Progress towards sustainable forest management. Food and Agricultural Organization of the United Nations, Rome, Italy. Fleischbein, K., Wilcke, W., Goller, R., Boy, J., Valarezo, C., Zech, W., Knoblich, K., 2005. Rainfall interception in a lower montane forest in Ecuador: effects of canopy properties. Hydrol. Proc. 19, 1355–1371. Fleischbein, K., Wilcke, W., Valarezo, C., Zech, W., Knoblich, K., 2006. Water budgets of three small catchments under montane forest in Ecuador: experimental and modelling approach. Hydrol. Proc. 20, 2491–2507. Forti, M.C., Neal, C., 1992. Hydrochemical cycles in tropical rainforests: an overview with emphasis on central Amazonia. J. Hydrol. 134, 103–115. Franzluebbers, K., Weaver, R.W., Juo, S.R., Franzluebbers, A.J., 1994. Carbon and nitrogen mineralization from cowpea plants part decomposing in moist and in repeatedly dried and wetted soil. Soil Biol. Biochem. 26, 1379–1387. Goller, R., Wilcke, W., Leng, M., Tobschall, H.J., Wagner, K., Valarezo, C., Zech, W., 2005. Tracing water paths through small catchments under a tropical montane rain forest in south Ecuador by an oxygen isotope approach. J. Hydrol. 308, 67– 80. ¨ kologie und Verju¨ngung von Mahagoni (Swietenia macrophylla Gu¨nter, S., 2001. O King) in Naturwa¨ldern Boliviens. Dissertation. University of Go¨ttingen, Go¨ttingen, Germany. Gu¨nter, S., Mosandl, R., 2003. Nachhaltige Naturwaldbewirtschaftung in Bergregenwa¨ldern Su¨decuadors. Eine Option zur Erhaltung von Biodiversita¨t? In: Mosandl, R., El Kateb, H., Stimm, B. (Eds.), Waldbau – weltweit, 192. Forstl. Forschungsber, Mu¨nchen, pp. 10–23. Gu¨nter, S., Cabrera, O., Weber, M., Stimm, B., Zimmermann, M., Fiedler, K., Knuth, J., Boy, J., Wilcke, W., Iost, S., Makeschin, F., Werner, F., Gradstein, R., Mosandl, R., 2008. Natural forest management in neotropical mountain rain forests – an ecological experiment. In: Beck, E., Bendix, J., Kottke, I., Makeschin, F., Mosandl, R. (Eds.), Gradients in a tropical mountain ecosystem of Ecuador. Ecological Studies 198. Springer, Berlin, Germany, pp. 347–359. Hamilton, L.S., Juvik, J.O., Scatena, F.N., 1995. The Puerto Rico tropical cloud forest symposium: introduction and workshop synthesis. In: Hamilton, L.S., Juvik, J.O., Scatena, F.N. (Eds.), Tropical montane cloud forests. Ecological Studies 110. Springer, Berlin, Germany, pp. 1–23. Henderson, A., Churchill, S.P., Luteyn, J.L., 1991. Neotropical plant diversity. Nature 351, 21–22. Hering, K.G., 2003. A scientific formulation of tropical forest management. Ecol. Model. 166, 211–238. Homeier, J., 2004. Baumdiversita¨t, Waldstruktur und Wachstumsdynamik zweier tropischer Bergregenwa¨lder in Ecuador und Costa Rica. Ph.D. Thesis. University of Bielefeld, Germany. Homeier, J., Dalitz, H., Breckle, S.-W., 2002. Waldstruktur und Baumartendiversita¨t im montanen Regenwald der Estacio´n Cientifica San Francisco in Su¨decuador. Ber. d. Reinhold-Tu¨xen-Ges. 14, 109–118.

Hungerbu¨hler, D., 1997. Neogene basins in the Andes of southern Ecuador: evolution, deformation and regional tectonic implications, Ph.D. Thesis. ETH Zu¨rich, Zu¨rich, Switzerland. Hutchinson, I.D., 1993. Puntos de partida y muestreo silvicultural de bosques naturales del tro´pico hu´medo. Coleccio´n silvicultura y manejo de bosques naturales No 7. CATIE, Turrialba, Costa Rica. Serie Tecnica: Informe Tecnico No. 204. Huth, A., Ditzer, T., 2001. Long-term impact of logging in a tropical rain forest – a simulation study. For. Ecol. Manage. 142, 33–51. Joslin, J.D., Wolfe, M.H., 1993. Temperature increase accelerates nitrate release from high elevation red spruce soils. Can. J. For. Res. 23, 756–759. Lamprecht, H., 1986. Waldbau in den Tropen. Paul Parey, Hamburg, Germany. Likens, G.E., Bormann, F.H., 1995. Biogeochemistry of a forested ecosystem, 2nd edn. Springer, New York, USA. Lukewille, A., Wright, R.F., 1997. Experimentally increased soil temperature causes release of nitrogen at a boreal forest catchment in southern Norway. Global Change Biol. 3, 12–21. Matzner, E. (Ed.), 2004. Biogeochemistry of a forested catchment in a changing environment. A German case study. Ecological Studies 172. Springer, Berlin, Germany. Muscolo, A., Sidari, M., Mercurio, R., 2007. Influence of gap size on organic matter decomposition, microbial biomass and nutrient cycle in Calabrian pine (Pinus laricio Poiret) stands. Forest Ecol. Manage. 242, 412–418. Nadkarni, N.M., Schaefer, D., Matelson, T.J., Solano, R., 2004. Biomass and nutrient pools of canopy and terrestrial components in a primary and a secondary montane cloud forest, Costa Rica. Forest Ecol. Manage. 198, 223–236. Ostertag, R., 1998. Belowground effects of canopy gaps in a tropical wet forest. Ecology 79, 1294–1304. Parker, G.G., 1983. Throughfall and stemflow in the forest nutrient cycle. Adv. Ecol. Res. 13, 57–133. Paulsch, A., 2002. Development and application of a classification system for undisturbed and disturbed tropical montane forests based on vegetation structure. Ph.D. Thesis. University of Bayreuth, Bayreuth, Germany. Schaefer, D.A., McDowell, W.H., Scatena, F.N., Asbury, C.E., 2000. Effects of hurricane disturbance on stream water concentrations and fluxes in eight tropical forest watersheds of the Luquillo Experimental Forest, Puerto Rico. J. Trop. Ecol. 16, 189–207. Schrumpf, M., Axmacher, J.C., Zech, W., Lehmann, J., Lyaruu, H.V.C., 2007. Long-term effects of rainforest disturbance on the nutrient composition of throughfall, organic layer percolate and soil solution at Mt. Kilimanjaro. Sci. Tot. Environ. 376, 241–254. Silver, W.L., Scatena, F.N., Johnson, A.H., Siccama, T.G., Watt, F., 1996. At what temporal scales does disturbance affect belowground nutrient pools? Biotropica 28, 441–457. Soethe, N., Lehmann, J., Engels, C., 2006. Root morphology and anchorage of six native tree species from a tropical montane forest and an elfin forest in Ecuador. Plant Soil 279, 173–185. Ulrich, B., 1983. Interactions of forest canopies with atmospheric constituents: SO2, alkali and earth alkali cations and chloride. In: Ulrich, B., Pankrath, J. (Eds.), Effects of accumulation of air pollutants in forest ecosystems. D. Reidel Publishing, Dordrecht, The Netherlands, pp. 33–45. USDA-NRCS (United States Department of Agriculture – National Resource Conservation Service), 1998. Keys to Soil Taxonomy, 8th edn. Pocahontas Press, Washington D.C., USA. Wilcke, W., Yasin, S., Valarezo, C., Zech, W., 2001. Change in water quality during the passage through a tropical montane rain forest in Ecuador. Biogeochemistry 55, 45–72. Wilcke, W., Yasin, S., Abramowski, U., Valarezo, C., Zech, W., 2002. Nutrient storage and turnover in organic layers under tropical montane rain forest in Ecuador. Eur. J. Soil Sci. 53, 15–27. Wilcke, W., Yasin, S., Fleischbein, K., Goller, R., Boy, J., Knuth, J., Valarezo, C., Zech, W., 2008a. Water relations. In: Beck, E., Bendix, J., Kottke, I., Makeschin, F., Mosandl, R. (Eds.), Gradients in a tropical mountain ecosystem of Ecuador. Ecological Studies 198. Springer, Berlin, Germany, pp. 193–201. Wilcke, W., Yasin, S., Fleischbein, K., Goller, R., Boy, J., Knuth, J., Valarezo, C., Zech, W., 2008b. Nutrient status and fluxes at the field and catchment scale. In: Beck, E., Bendix, J., Kottke, I., Makeschin, F., Mosandl, R. (Eds.), Gradients in a tropical mountain ecosystem of Ecuador. Ecological Studies 198. Springer, Berlin, Germany, pp. 203–216. Williams, M.R., Fisher, T.R., Melack, J.M., 1997. Solute dynamics in soil water and groundwater in a central Amazon catchment undergoing deforestation. Biogeochemistry 38, 303–335. Yasin, S., 2001. Water and nutrient dynamics in microcatchments under montane forest in the south Ecuadorian Andes. Bayreuther Bodenkundliche Berichte 73, University of Bayreuth, Germany.