Effects of calcium and aluminum chloride additions on foliar and throughfall chemistry in sugar maples

Transcription

1 Forest Ecology and Management ) 75±90 Effects of calcium and aluminum chloride additions on foliar and throughfall chemistry in sugar maples Torsten W. Berger a,*, Chris Eagar b, Gene E. Likens c, Gerhard Stingeder d a Institute of Forest Ecology, Univ. f. Bodenkultur, Peter Jordan-Strasse 82, 1190 Vienna, Austria b USDA Forest Service, Northeastern Forest Experiment Station, Durham, NH 03824, USA c Institute of Ecosystem Studies, Millbrook, NY 12545, USA d Institute of Chemistry, Univ. f. Bodenkultur, 1190 Vienna, Austria Received 15 December 1999; received in revised form 9 May 2000; accepted 29 May 2000 Abstract Calcium availability for sugar maple stands at the Hubbard Brook Experimental Forest New Hampshire, USA) was tested by experimental addition of CaCl 2 and AlCl 3. Additions of 10 g Ca m 2 represented the estimated loss from the soil exchange complex during the last 30 years due to acidic deposition. Four years of data from 12 throughfall collection sites were used to evaluate the in uence of foliar nutrient content, precipitation amount, dry deposition, precipitation acidity and precipitation solute concentrations on throughfall chemistry. Calcium additions increased Ca foliar contents signi cantly. Foliar contents indicated plant uptake of Cl. Leaching of Cl from the canopy increased with elevated Cl content of the green foliage. Leaching rates for Ca, Mg, and K were not signi cantly different between the treatments surprisingly Ca leaching tended to decrease with increasing foliar Ca content). We suggest that Ca supply to Ca de cient sugar maple trees protected the foliage from increased leaching of Ca and other elements) due to improved integrity of cell membrane and cell wall formation from Ca. Degradation of the structural material of the foliage autumnal leaf senescence, damages by ice and hail storms) caused Ca throughfall uxes in accordance to measured foliar Ca contents. Increasing acidity of precipitation caused increased leaching of Ca, Mg and K. About half of the cation leaching from these sugar maple canopies is attributable to a cation-exchange reaction driven almost entirely by H in precipitation. # 2001 Elsevier Science B.V. All rights reserved. Keywords: Throughfall chemistry; Calcium; Chloride; Acer saccharum 1. Introduction Calcium Ca), the fth most abundant element in trees, is an essential component for wood formation and the maintenance of cell walls e.g. Lawrence et al., 1995). Calcium is usually the most abundant of the * Corresponding author. Tel.: ; fax: addresses: tberger@woek.boku.ac.at, @compuserve.com T.W. Berger). alkali and alkaline earth elements on the soil exchange complex, and is important in the regulation of soil ph Bowen, 1979). In contrast to potassium, calcium is not leached readily from living foliage due to its relative immobilization in pectates and on membranes. Thus, in forest ecosystems Ca typically cycles between plants and soil through uptake±litterfall± mineralization processes Likens et al., 1998). Because Ca is a macronutrient for higher plants, spatial and temporal variations in its supply are important to the growth and vigor within a forest /01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S )

2 76 T.W. Berger et al. / Forest Ecology and Management ) 75±90 ecosystem. Soil acidi cation and accompanying leaching of Ca from the soil exchange complex is a gradual natural process. However, during the past 6 decades, concentrations of root available Ca exchangeable and acid-extractable forms) in forest- oor and soils have decreased markedly at some locations in the northeastern United States Shortle and Bondietti, 1992; Johnson et al., 1994; Likens et al., 1996, 1998). For example, long-term data from the Hubbard Brook Experimental Forest demonstrate that before the mid-1950s, annual Ca depletion from the soil exchange complex was about equal to net biomass storage, and atmospheric deposition plus weathering release nearly balanced streamwater loss. After the mid-1950s, soil depletion of Ca was up to two times greater than net biomass storage, and streamwater loss was up to 2.7 times larger than bulk precipitation plus weathering release Likens et al., 1996). Increasing concern about long-term changes in pools of available Ca for forest ecosystems and their productivity at many sites in the northeastern United States has been documented e.g. Federer et al., 1989; Lawrence et al., 1995; Bailey et al., 1996; Likens et al., 1996, 1998). A re ection of the base cation status of forest ecosystems may be seen in the element uxes in throughfall. They are measured readily, but it is dif cult to separate throughfall into its contributions. Nutrients in throughfall may result from a) incident precipitation, passing through the canopy; b) material deposited as particles, gases, or cloud droplets prior to the precipitation event being washed off during the event, and c) exchange processes within the canopy including foliage, woody parts, epiphytes, and microorganisms). Hence, according to Lovett and Lindberg 1984), net throughfall ux NTF) can be de ned as NTF ˆ TF IP ˆ DD CE 1) indicating that the difference between throughfall TF) and precipitation inputs IP) is equal to the sum of dry deposition DD) and canopy exchange CE). Eq. 1) ignores stemflow SF) flux, which is often a minor percentage of throughfall flux. Canopy exchange includes both leaching efflux from the canopy) and uptake or retention influx to the canopy). Because neither dry deposition nor canopy exchange is easily quantified, it is difficult to separate them into their respective contributions to total chemical deposition in throughfall. Distinguishing between the two is important, as dry deposition represents an input to the ecosystem, while canopy exchange is an intrasystem transfer. Quantitative separation of these mechanisms is essential for understanding pollutant effects on canopy processes, but it is notoriously dif cult Lovett, 1994). Lovett et al. 1996) pointed out that there are several variables that in uence one or both of these processes e.g. precipitation amount and rate; source strength and proximity of dry deposition; precipitation acidity; precipitation chemical concentrations; composition, age and nutrient status of the forest; epiphytes; leaf area and others). It is hypothesized that foliar nutrient contents will respond to soil treatments with calcium and that increased foliar element contents will increase leaching rates of the corresponding element, partly due to acidic deposition. Since several studies have shown that H disappearance from bulk deposition can not account for the total base cation enrichment of precipitation beneath forest canopies, organic anions are thought to be important counter ions for leached cations Eaton et al., 1973; Lovett et al., 1985; Bredemeier, 1987; Sayre and Fahey, 1999). Cronan and Reiners 1983) concluded that during the growing season as much as 30±50% of precipitation-borne strong acidity is neutralized by weak Brùnsted base leaching and the remaining neutralization was attributable to leaf surface ion exchange in a northern hardwood canopy. Hence, a further aspect of this work is the evaluation of the fate of H, which requires determining the magnitude of the cation-exchange reaction in the canopy, taking acid±base reactions into account. Calcium was applied as calcium and aluminum chloride to northern hardwood forest plots, dominated by sugar maple, to vary Ca availability manifold e.g. increased foliar Ca content due to addition of CaCl 2 and decreased Ca content due to Al induced Ca de ciency by the AlCl 3 treatment). We used 4 years of data from 12 throughfall collection sites on untreated, CaCl 2 and AlCl 3 treated sugar maple stands to evaluate the in uence of foliar nutrient content, precipitation amount, dry deposition, precipitation acidity and precipitation solute concentrations on the NTF of these sites.

3 T.W. Berger et al. / Forest Ecology and Management ) 75± Methods 2.1. Study area The study was done in an area 300±900 m west of Watershed 6 W6) of the Hubbard Brook Experimental Forest HBEF) N, W) in the White Mountains of New Hampshire. The climate of this area is humid continental, with 1310 mm average annual precipitation. The study area spans the elevational range 700±760 m. Watershed 6 is the biogeochemical reference watershed at HBEF and is the site of long-term vegetation and biogeochemical monitoring e.g. Likens and Bormann, 1995). The HBEF is covered mostly with northern hardwood vegetation growing on Typic Haplorthods that developed in sandy till. The average ph of the humus layer Oa horizon) is 3.9, while ph values in the mineral soil increase from 4.2 to 4.7 with depth Johnson et al., 1991). However, most soil pro les of the study area were classi ed as Aquic Haplorthods and some as Aquic Haplumbrepts. Watershed 6 13 ha) and adjacent land to the west are covered by mature northern hardwood forest in which the dominant canopy tree species are sugar maple Acer saccharum Marsh.), American beech Fagus grandifolia Ehrh.), and yellow birch Betula alleghaniensis Britt.). Canopy height is generally about 20± 25 m, and epiphyte cover is negligible Study sites and treatments Within the study area 12 study sites 45 m45 m) were established, where sugar maple is the dominant canopy tree species 70±85% of stems). Calcium availability for these sugar maple stands was varied threefold: control, CaCl 2 treated and AlCl 3 treated plots in each case four replications). The study was done during four vegetation periods compare Table 1): 1995, no additions ambient conditions); 1996, part of the additions were done; 1997, before the growing season the total amount was added; 1998: no additions additional time for ecosystem response to the treatments). The total amount of 10 g Ca m mmol c Ca m 2, Table 1) represents the estimated loss from soil exchange upper mineral soil) during the last 30 years due to acidic deposition Federer et al., 1989; Likens et al., 1996, 1998). The total amounts for Ca, Table 1 Additions of calcium, aluminum and chloride mmol c m 2 ) for the CaCl 2 and AlCl 3 treated sugar maple stands a Site Element Total CaCl 2 Calcium ± ± 500 Aluminum ± ± ± ± ± Chloride ± ± 500 AlCl 3 Calcium ± ± ± ± ± Aluminum ± ± 500 Chloride ± ± 500 a No additions were performed on the control sites. Five partial treatments were done as CaCl 2 and AlCl 3 in fall and spring during the leafless periods, resulting in the given amounts added before the growing seasons of 1996 and Al and Cl are the same on an equivalent basis, however, at the beginning of the growing season in % more CaCl 2 was added than AlCl 3 compare Table 1) Foliar analysis Leaf samples of overstory sugar maple trees ve trees per plot) were collected at the end of August 1996 and 1997 with a shotgun. Individual samples were analyzed for N Kjedahl; auto-analyzer) and cations plasma emission spectrometer) after digestion with H 2 SO 4 and H 2 SeO 3 at the USDA forest service. chloride analyses amperometric endpoint titration with coulometric generated silver ions by means of a aminco chloride titrator; American Instrument Company, 1971) were done after digestion with HNO 3 one pooled sample out of ve replications per plot) at the Institute of Forest Ecology Atmospheric deposition Throughfall was collected and analyzed at all sites during the growing season 1 June to 30 September) of 1995±1998. Samples were collected weekly and the funnels of the collectors were rinsed with deionized water. Cleaned rinsed with deionized water) polyester plugs were used in the funnels to minimize particulate inputs and replaced each time of collection with plastic gloves. Six bulk samplers were placed within the inner 25 m25 m of the plot inside the 10 m wide buffer zone of the treated 45 m45 m area). The locations of the individual throughfall samplers were

4 78 T.W. Berger et al. / Forest Ecology and Management ) 75±90 not changed during the study. These collectors consisted of polyethylene funnels with a 200 mm upper diameter, placed 60 cm above the forest oor on wooden sticks. The funnels were connected to onegallon polyethylene reservoirs by black norprene tubings. All samples of the individual collectors at each site were pooled. Precipitation bulk deposition) input was sampled in a clearing at an elevation of 700 m, 350 m east of the study area using the same kind of sampler two bulk samplers, 160 cm above ground). For the regression analysis only single rain events were used, separated from other rainfall periods by at least a 12 h dry period. Of the 64 weekly collections made during the study, 16 represented single precipitation events. Individual rain events were exactly comparable for all study sites with regard to the length of the dry period before the event, duration of the rain event and time of collection after the end of the event. The volume of throughfall and precipitation was measured in the eld with a graduate cylinder, and a sample was brought back to the laboratory for analysis. An aliquot was allowed to come to room temperature and ph was measured immediately after collection with a glass electrode. Another aliquot of the sample, for chemical analysis of major ions, was preserved with 100 ml of chloroform per 100 ml of sample, and a third aliquot, for titrations, was not preserved. The samples were not ltered, but were stored in clean high-density polyethylene bottles in the dark at 48C until chemical analysis was completed, usually within 6 months; titrations were done within 2±3 weeks after collection. Sulfate, Cl and NO 3 were measured by ion chromatography, Ca and Mg by inductively coupled plasma emission ICP) spectrophotometry, Na and K by atomic absorption spectroscopy and NH 4 by auto-analyzer Institute of Ecosystem Studies). Aluminum analysis ICP Ð mass spectrometry, Institute of Chemistry) was performed after monthly volumeadjusted samples were prepared from weekly collections at all four sites of the same treatment. Throughfall uxes were calculated according to measured solution volumes per area of the collector. Titrations were performed by hand with a burette while the sample was stirred magnetic stirrer) and purged with N 2. Alkalinity was determined by endpoint titration with M HCl to ph 5.0 according to Richter et al. 1983) and Lovett et al. 1985). Total acidity was measured by titration with M NaOH standardized against HCl) and Gran plot analysis Lindberg et al., 1984). Weak undissociated) acidity was determined by subtracting the free acidity measured as ph) from total acidity measured by titration, Lovett et al., 1985). 3. Results 3.1. Foliar analysis Foliar element contents are given in Table 2. According to the proposed hypothesis Ca additions CaCl 2 ) increased Ca foliar contents signi cantly in both years. Differences between the control and AlCl 3 sites are not signi cant for Ca but Al additions AlCl 3 ) tended to cause lower Ca foliar contents. Although Mg was not added with the treatment, foliar Mg contents were signi cantly higher for the CaCl 2 treated stands than for the control 1996) and AlCl 3 sites 1996 and 1997). Since Ca and Al were added as chloride to minimize changes in soil ph, and to accelerate in ltration of Ca and Al into deeper soil horizons, the treatments had distinct effects on foliar Cl content Table 2, Cl is used for chloride throughout this paper). In 1997, after the total amounts were added, foliar Cl contents were higher for the CaCl 2 sites than for the control sites, indicating plant uptake. Aluminum chloride additions increased foliar Cl content, however, differences were not signi cant. The same trend was monitored in 1996, after two fths compare Table 1) of the total amount had been added. Table 2 Foliar element contents mg g 1 ) of sugar maple overstory trees a Year Site N Ca Mg K Cl Al 1996 Control a 0.77 a CaCl b 1.09 b AlCl a 0.69 a Control a 0.92 ab a a CaCl b 1.07 b b a AlCl a 0.77 a ab b a Results of a ScheffeÂ's multiple range test are given only, if differences were significant values by the same letter are not significantly different: Pˆ0.05; number of replications were 20 per treatment and year, except four for Cl).

5 T.W. Berger et al. / Forest Ecology and Management ) 75±90 79 Aluminum chloride treatments resulted in increased foliar Al contents of sugar maple in 1997, probably due to low soil ph, while foliar Al contents were not different between the control and CaCl 2 sites. Foliar Al contents were one order of magnitude lower than for the macro nutrients and Al throughfall concentrations were found to be unimportant in the ion balance of this study Atmospheric deposition Bulk precipitation and throughfall uxes during the growing seasons of 1995±1998 are given in Table 3. During the pre-treatment period 1995) no signi cant differences were observed between the control, CaCl 2 and AlCl 3 sites, enabling comparisons between the treatments after the additions were done. Signi cant differences Pˆ0.05) were recorded for Ca in 1996 AlCl 3 <control), for Na in 1997 CaCl 2, control<alc 3 ) and for Cl both in 1997 control<cacl 2 <AlCl 3 ) and 1998 control<cacl 2, AlCl 3 ). Since only these few comparisons four out of 52; Table 3) between the treatments revealed signi cant differences between solute uxes in throughfall, four-year averages 1995±1998) are given in Table 3 as well. Throughfall uxes of Ca, Cl and Al are shown for the years 1995±1998 from 1 June to 30 September fully developed canopy) in Fig. 1. While Ca uxes were relatively constant over the years for the control plots, the additions did change throughfall uxes in 1996, but not signi cantly in However, these changes were in contradiction to the stated hypothesis see introduction section). Only in 1998 did Ca throughfall uxes agree with measured foliar Ca contents see Table 2 for 1996 and 1997). Chloride throughfall uxes indicate increased leaching of Cl, when foliar levels are elevated. Aluminum throughfall uxes are not as accurate, because samples were Table 3 Mean fluxes of solutes mmol c m 2 ) from 1 June to 30 September in bulk precipitation open) and throughfall at untreated control) and treated CaCl 2. AlCl 3 ) sugar maple stands and precipitation amount H 2 O, mm) a Year Site Ca Mg K Na NH 4 NO 3 SO 4 Cl H Alkal WeaAc TotAc H Open Control CaCl AlCl Open Control 9.16 b CaCl ab AlCl a Open Control a a CaCl a b AlCl b c Open Control a CaCl b AlCl b ±1998 Open Control CaCl AlCl a Weak acidity weaac, undissociated) was determined by substracting the free acidity H, measured as ph) from total acidity totac, measured by titration). This difference doesn't match excactly for 95, since ph was measured for all sites, but totac and alkalinity alkal) only for one site out of four replications per treatment. A one way ANOVA factor treatment) was done for all throughfall fluxes of each year separately and results of a ScheffeÂ's multiple range test are given only Backhaus et al., 1994), if differences were significant values by the same letter are not significantly different, Pˆ0.05).

6 80 T.W. Berger et al. / Forest Ecology and Management ) 75±90 Fig. 1. Mean throughfall fluxes of Ca, Cl and Al from 1 June to 30 September of each year. Standard errors are given for fluxes of Ca and Cl four replications per treatment), but not for Al fluxes, since samples were pooled within treatment and month before Al analysis. pooled before analysis. However, a two way ANOVA factors year and treatment; 4 monthly values per treatment were used as replicates) indicated that the factor year 1998, 1996<1995) surpassed effects of the Al treatment not signi cant). Mean solute uxes in NTF are given for the control sites in Table 4 and differences between the years were tested by a ScheffeÂ's multiple range test. The use of NTF instead of TF for comparisons between years seems more appropriate, because effects of year-to-year

7 T.W. Berger et al. / Forest Ecology and Management ) 75±90 81 Table 4 Mean net throughfall fluxes throughfall flux minus bulk precipitation flux) for solutes mmol c m 2 ) from 1 June to 30 September 1995± 1998) at the control sites a Year Ca Mg K Na NH 4 NO 3 SO 4 Cl H a 4.12 b a 0.24 b 1.53 b 0.09 b 1.51 ab 0.82 a a a 4.29 b b 0.17 b 0.33 b 1.94 a 2.95 b 0.99 a c a 4.29 b ab 0.14 b 0.81 b 0.23 b 2.60 ab 1.01 a c a 2.81 a ab 0.24 a 5.59 a 3.38 a 0.63 a 0.62 a b a A massive ice storm struck the study area in January 1998, reducing the average leaf area index LAI) from approximately 6.5 to 4.7. A severe hail storm on 24 August 1998 further reduced LAI by 1.0±1.4 units. A ScheffeÂ's multiple range test was performed to test differences between the years values with the same letter are not significantly different, Pˆ0.05). variations in amounts and patterns of precipitation are minimized. 4. Discussion and conclusions This study revealed interesting effects of CaCl 2 and AlCl 3 treatments on foliar nutrition of sugar maple and subsequently on the chemistry of throughfall. Additions of 10 g Ca m mmol c Ca m 2 ) represented the estimated loss from soil exchange upper mineral soil) during the last 30 years due to acidic deposition. The total amounts for Ca, Al and Cl were the same on an equivalent basis Foliar analysis Calcium additions increased foliar Ca contents signi cantly Table 2). This fact is not surprising, since Ca uptake into roots occurs principally by passive movement in the mass ow of soil water driven by transpiration McLaughlin and Wimmer, 1999). Increased Al tended to decrease foliar Ca content. Aluminum induced Ca de ciency is well documented e.g. Rost-Siebert, 1985) and explained by the fact that Al interferes with Ca uptake and root growth. The increase of Ca foliar content at the control sites from mg g 1 ) to mg g 1 )is in accordance with observed large year-to-year differences at the HBEF Likens et al., 1998). These authors also reported that Ca foliar contents of sugar maple in the forest west of W6 were higher in mg g 1 ) than averages for 1992± mg g 1 ) and that foliar Ca content of sugar maple at high elevations 715 m) was signi cantly lower 3.8 mg g 1 ) than in low and mid-elevations 5.9 mg g 1 ). Hence, we conclude that Ca supply of sugar maple at the high elevation sites of this study is below the optimum. In addition, Ca de ciency is suggested in these trees by the high rates of increase 29% for 1996) in foliar Ca content resulting from relatively low amounts of Ca fertilization. The nitrogen/calcium ratio 4.3±4.6; Table 2) for the control sites is within the `harmonious range' 2±7; Stefan and FuÈrst, 1998), rejecting the hypothesis of N induced Ca de ciency via enhanced atmospheric N deposition e.g. Gundersen, 1998) and supporting the hypothesis of long-term loss of Ca from the ecosystem via natural or anthropogenic disturbances e.g. forest harvest, acid rain) on these base poor soils of the HBEF Likens et al., 1996, 1998). Although Mg was not added with the treatment, foliar Mg contents were signi cantly higher for the CaCl 2 treated stands than for the control and AlCl 3 sites Table 2). We don't know, whether this increase was caused by positive impact of the treatment on the root system and consequently increased uptake, by higher Mg soil solution concentrations due to soil exchange reactions or by elevated soil ph depression of Mg uptake by H ; Marschner, 1986). Foliar Cl contents re ected plant uptake Table 2). According to Marschner 1986), chloride is readily taken up by plants and the mobility of Cl is high, both in short- and long-distance transport. Since Cl and bromide have similar physiochemical properties, substitution of Cl by Br is of no practical difference, except that Cl is much more abundant in forest ecosystems than Br. Hence, signi cant uptake of experimentally added Br by a northern hardwood forest at the HBEF Berger et al., 1997; Berger and

8 82 T.W. Berger et al. / Forest Ecology and Management ) 75±90 Likens, 1999) is in accordance with the observed Cl uptake of this study Atmospheric deposition Average precipitation amounts and bulk precipitation uxes 1995±1998, Table 3) were in the same range as reported by Lovett et al. 1996) for the same area 1989±1992) for all major solutes, except SO 4 and H uxes which were 30 and 40% lower in this study. This decrease is undoubtedly due to lower inputs of SO 4 and H in recent years in atmospheric deposition Likens and Bormann, 1995). Different stands and elevations had lower solute throughfall uxes as well e.g. mean Ca throughfall uxes of this study were about 50% of uxes measured in a mature hardwood forest; Lovett et al., 1996). A massive ice storm struck the study area in January 1998, reducing the average leaf area index LAI) from approximately 6.5 to 4.7 and a severe hail storm on 24 August 1998 further reduced LAI by 1.0 to 1.4 units Fahey, personal communication). Since the bulk samplers were not moved during the experimental period this research represents a case study for evaluating effects of LAI on solute NTF. In fact in 1998, reduced LAI affected NTF of most solutes signi cantly Table 4). Lower NTF for Mg, Na and NH 4 in 1998 were probably caused both by the reduced ltering capacity of the canopy and lower foliar leaching rates due to reductions of LAI by up to 50%. Thus, the 1998 data are not taken into consideration for further discussions of differences between the treatments Factors regulating throughfall flux Regression analysis on the event-to-event variation in NTF is used in this study to resolve the dry deposition DD) and canopy exchange CE) components see Eq. 1)), as proposed by Lovett and Lindberg 1984). This approach is based on the hypothesis that for single-event collections of throughfall, the dry deposition component of the NTF is correlated to the length of the dry period before the event antecedent period) and the canopy exchange component is correlated with the amount of precipitation of that event. Multiple linear regression is performed using the NTF value for each event as the dependent variable and the antecedent period and the precipitation amount as independent variables. Several assumptions and caveats have been discussed by Lovett and Lindberg 1984); Lovett et al. 1996), and Berger and Glatzel 1998). To be consistent with Lovett et al. 1996), concentrations of H and a particular solute in the incident precipitation were added, which increased signi cance of the overall regressions in most cases. Thus, the regression equations were of the form: NTF x ˆ a b 1 A b 2 P b 3 C H b 4 C x 2) see abbreviations and units on Table 5) The regression results indicated that Eq. 2) explained a major portion of the variance in NTF for most sites and solutes Table 5). Unlike the results of Lovett and Lindberg 1984) and Berger and Glatzel 1998), many of the intercept terms in this analysis were signi cant. However, these results correspond with those of Lovett et al. 1996) for comparable sites at the HBEF. They conclude that the apparent intercept may be a result of the nonlinearity in the relationship with independent variables, although this nonlinearity was not evident in plots of NTF versus the independent variables in both studies Dry deposition The antecedent period A) term was highly signi cant for Ca, Mg and K, indicating that dry deposition or other processes that cause accumulation of substances on the canopy or in the collectors between rain events do contribute signi cantly to the NTF. This result contrasts to Lovett et al. 1996), who used the same approach and concluded that dry deposition does not play a major role in controlling the chemistry of throughfall at HBEF compare Likens et al., 1998). Given the coef cients of Table 5 for A and P see below), measurements of the days of rain-free weather and amount of rainfall, the relative contributions of dry deposition and canopy exchange can be estimated. By doing so only for the single rain events 1996± 1997) of this study, dry deposition of base cations amounts to 24±45% Ca 38±45%, Mg 24±37%, K 41± 44%) and the remainder is attributed to foliar leaching additional positive cation exchange was caused by precipitation chemistry). The contribution of dry deposition to NTF for Ca in the same range as measured adjacent to a lime quarry in Austrian oak

9 Table 5 Number of observations n), adjusted coefficients of determination R 2 ), and regression coefficients for single event regressions for untreated control) and treated CaCl 2, AlCl 3 ) sugar maple stands during the treatment period 1996±1997) as well as for all sites during the entire experimental period 1995±1998) a Site n Parameter Ca Mg K Na NH 4 NO 3 SO 4 Cl H Alkal WeaAc TotAc Control 32 R *** 0.85 *** 0.83 *** ns 0.33 ** 0.32 ** 0.29 ** 0.58 *** 0.91 *** ns 0.41 ** 0.29 ** 96±97 a 397 *** 299 *** 1068 *** ns 417 *** ns ns 162 ** 697 *** ns ns ns A 41.6 *** 16.3 *** 99.8 *** ns ns 38.1 ** ns 12.0 *** 41.1 *** ns *** ** P 22.1 *** 12.0 *** 54.4 *** ns 16.4 ** ns ns 7.4 ** 33.4 *** ns ns 61.0 * C H 2.4 *** 1.3 *** 5.4 *** ns 3.3 ** ns ns 0.4 * 8.0 *** ns ns ns C x 10.1 ** 40.2 *** 62.6 ** ns 8.5 * ns ns 7.1 * ns 12.2 *** 12.1 ** CaCl 2 32 R *** 0.84 *** 0.84 *** ns 0.37 ** 0.36 ** 0.43 ** 0.40 ** 0.85 *** ns 0.36 ** 0.31 ** 96±97 a 384 *** 303 *** 1036 *** ns 395 *** ns ns 193 * 590 *** ns ns 1626 * A 33.8 *** 10.7 ** 79.2 *** ns ns 42.2 ** ns 12.2 ** 46.7 *** ns *** ** P 21.8 *** 11.8 *** 44.2 *** ns 15.9 ** ns ns 10.5 ** 25.1 *** ns ns 87.9 * C H 2.4 *** 1.1 *** 4.4 *** ns 3.8 ** ns ns 0.7 * 6.8 *** ns ns ns C x 12.8 *** 50.3 *** *** ns 8.2 * ns ns ns ns 17.6 *** 17.6 *** AlCl 3 32 R *** 0.78 *** 0.77 *** ns 0.53 *** 0.33 ** 0.21 * 0.29 * 0.70 ** ns 0.49 *** 0.36 ** 96±97 a 344 *** 254 *** 967 *** ns 455 *** ns ns 389 * 775 *** ns ns 1557 ** A 31.3 *** 9.0 * 86.6 *** ns 15.3 * 33.5 ** ns ns 50.8 *** ns *** ** P 21.8 *** 12.4 *** 53.9 *** ns 15.1 ** ns ns 19.6 ** 35.8 *** ns 52.9 * 87.6 ** C H 2.6 *** 1.4 *** 5.6 *** ns ns ns ns 1.3 * 7.4 *** ns ns ns C x ns 31.3 *** 46.4 * ns ns ns ns 19.2 ns 15.1 *** 14.9 *** Mean 192 R *** 0.67 *** 0.70 *** 0.21 *** 0.43 *** 0.20 *** 0.26 *** 0.30 *** 0.85 *** 0.09 ** 0.39 *** 0.23 *** 95±98 a 284 *** 111 *** 403 *** 8 ** 340 *** ns 67 * 56 ** 493 *** 125 *** ns 651 ** A 35.3 *** 11.2 *** 83.4 *** 0.5 * ns 15.6 *** ns 9.6 *** 39.9 *** ns *** 48.3 *** P 18.9 *** 8.0 *** 29.9 *** 0.5 *** 13.1 *** ns 11.6 *** 4.4 *** 28.6 *** 3.6 * 31.9 ** 53.3 *** C H 2.5 *** 0.9 *** 2.6 *** 0.1 *** 2.6 *** 2.5 ** ns ns 7.2 *** 0.7 ** ns 4.2 * C x 3.8 ** 16.5 *** 24.7 ** 0.9 ** 4.9 ** 5.6 ** ns ns ns 11.9 *** 10.3 *** a Regressions are of the form NTF xˆa b 1 A b 2 P b 3 C H b 4 C x ; NTF x, net throughfall of element x mmol c m 2 ); a, intercept term; A, antecedent period d); P, precipitation amount mm); C H and C x are the concentrations mmol c l 1 )ofh and solute x in precipitation. Units of coefficients are mmol c m 2 per day for A representing mean dry deposition rates) and mmol c m 2 per mm for P representing mean canopy exchange rates) and mmol c m 2 per mmol c l 1 for C H and C x representing effects of acid precipitation on the NTF). Significance of overall regression and individual coefficients. ns, P>0.05, * P<0.5, ** P<0.01, *** P< T.W. Berger et al. / Forest Ecology and Management ) 75±90 83

10 84 T.W. Berger et al. / Forest Ecology and Management ) 75±90 stands by Berger and Glatzel, 1998), Mg and K is surprisingly high. We believe that dry deposition was overrated, because samples were collected weekly, systematically overestimating dry periods between single rain events used for the performance of this technique e.g. the mean dry period between the seleced single rain events was 6.5 days, while the mean long-term dry period amounts 2±3 days). Other ions for which the NTF would be expected to re ect dry deposition, e.g. Na and SO 4, have small NTF in these stands compare Table 4, NTF from June to September). Hence, the given coef cients of Table 5 are not useful for upscaling e.g. to total growing season) but provide relative comparisons between the treatments. Regression analyses Table 5) showed high dry deposition rates for weak undissociated) acidity, which was determined by subtracting the free acidity measured as ph) from the total acidity measured by titration). However, we do not believe that the forest gained weak acidity via dry deposition, but we consider weak acidity an internal source. One reason, why this method was not useful for analyzing weak acidity data, might be that organic anions migrate to the surface of the leaf in the transpiration stream exudations) during dry periods Cronan and Reiners, 1983; Reiners et al., 1986). Another reason might be that natural weathering of the cuticle during dry periods produces organic material on the surface which is washed off by the following rain event Likens et al., 1994). Both processes would show up as dry deposition in this analysis, although they represent internal sources. To our knowledge this technique has not been used before regarding weak acidity NTF Canopy leaching The precipitation amount term P) was signi cant for all base cations except Na, indicating that canopy leaching is a dominant process controlling event-toevent variation in NTF Table 5). Leaching rates for Ca see Fig. 2), Mg, and K were not signi cantly different between the treatments. Leaching coef cients indicate a trend of reduced leaching for Mg despite elevated foliar Mg contents, Table 2) and K on the CaCl 2 sites during 1996 and This trend is supported by throughfall uxes for all base cations Table 3; 1996±1997). The performance of a Co- ANOVA for solute NTF with solute 1995 NTF as Fig. 2. Regression coefficients for P see Table 5, 1996±1997) for regressions of Ca and Cl, representing mean canopy exchange rates positive values indicate leaching). Dry deposition rates for weak acidity coefficients for A) were converted into canopy exchange rates adjusted by given significant terms for P of the model), assuming only internal sources for weak acidity see text). Error bars are S.D. covariate did not change statistical differences between the treatments given for solute TF Table 3), except for 1998, when Ca NTF were signi cantly higher for the CaCl 2 sites than for the control and AlCl 3 sites. Hence, our stated hypothesis of increased Ca foliar leaching due to elevated Ca foliar contents must be rejected for the treatment period 1996±1997), there even seems to be an opposite trend. Because Ca is

11 T.W. Berger et al. / Forest Ecology and Management ) 75±90 85 important for maintenance of cell walls Marschner, 1986), we suggest that Ca supply to Ca de cient sugar maple trees protects the foliage from increased leaching of Ca, e.g. due to acidic deposition. When Ca is supplied in excessive amounts, Ca foliar leaching will occur in accordance to Ca foliar contents, as measured by Berger and Glatzel 1998) for oak stands on calcareous soils. The fact that Mg foliar leaching was slightly reduced despite the coupled increase of Mg together with Ca in the green foliage supports this new hypothesis. Calcium throughfall uxes from 1 to 15 October 1997 leaf senescence and beginning of litter fall, no October data were available for 1996 and 1998) were higher for the CaCl 2 treated stands than for the control and AlCl 3 treated stands, despite the opposite trend for the fully developed canopy from 1 June to 30 September. Hence, we conclude that during leaf senescence, when Ca foliar contents are elevated Ryan and Bormann, 1982), Ca is present in a more leachable form. The deterioration of a broad spectrum of essential physiological processes in Ca-de cient plants has led to recognition of the important role of Ca supply in delaying plant senescence Poovaiah, 1988). Symptoms of Ca de ciency that are commonly associated with senescence include loss of protein, loss of chlorophyll, and reduced integrity of cell membrane and cell wall McLaughlin and Wimmer, 1999). The delay of plant senescence in the CaCl 2 treated sugar maple stands might be an explanation of the fact that increased Ca foliar levels did not cause higher but lower Ca leaching of the non senescent foliage. The same explanation is attributable to the observation that during the growing season of 1998, Ca foliar leaching was in accordance to Ca foliar contents. The massive ice storm in January 1998 branches of partly damaged trees died after spring ush) and the severe hail storm in August 1998 degraded the structural material of the foliage during the growing season June Ð September) similar to autumnal leaf senescence. Lovett and Hubbell 1990) also measured higher not signi cant) leaching of Ca and Mg in damaged sugar maple branches than in undamaged branches. Interpretations are dif cult, however, because we do not know the fraction of Ca which might be insoluble-stored as Ca oxalate crystals or Ca pectate as reported for conifer needles Fink, 1991). According to DeHayes et al. 1997, 1999) the dominant, but insoluble, extracellular Ca pool re ected in measured total foliar Ca contents is not a meaningful surrogate for the physiologically important and labile pool associated with the plasma membrane-cell wall compartment of red spruce mesophyll cells. Canopy exchange coef cients were positive leaching) for Cl, indicating signi cant impacts of the Cl treatments on Cl foliar leaching rates Table 5, Fig. 2). As indicated by measured Cl throughfall uxes compare Fig. 1) the Cl leaching coef cient was higher for the AlCl 3 treated sites than for the CaCl 2 treated sites, although Cl foliar contents Table 2) showed the opposite trend. However, this fact strengthens the above hypothesis that Ca supply to Ca de cient sugar maple trees protects the foliage from increased leaching of Ca and other elements, e.g. Cl, due to Ca improved integrity of cell membrane, cell wall formation and maybe cuticle formation. Plant uptake of experimentally added Cl into the foliage of sugar maple and increased leaching of Cl via throughfall are especially interesting, since such studies are rare and contradictory. For example, Ulrich 1983) assumed that canopy exchange for Cl was zero, however, Kazda 1990) attributed one-third of the Cl in the stem ow of a beech stand to crown leaching. As discussed above, we do not believe that the forest gained weak acidity via dry deposition, but we consider weak acidity an internal source. Dry deposition rates for weak acidity coef cient for A) were converted into canopy exchange rates adjusted by signi cant terms for P of the model) by multiplying by the days of rain-free weather and dividing by the amount of precipitation only for the selected single rain events of 1996±1997, Fig. 2). Differences between the treatments are not signi cant but leaching of organic anions tends to increase with increasing foliar Ca content. Organic anions are frequently assessed by the anion de cit of a solution, but this approach involves large potential errors when the de cit is small compared with the cation and anion totals Lovett et al., 1985). Anion de cit throughfall uxes were between 8.7 and 13.4 mmol c m 2 per growing season 1996±1997) and amounted to 43± 63% of weak acidity throughfall uxes. The anion de cit TF was signifcantly lower for the AlCl 3 treated sites than for the control sites 1996: P<0.10; 1997: P<0.05) supporting the observed trend for roughly calculated weak acidity leaching rates of Fig. 2. Weak

12 86 T.W. Berger et al. / Forest Ecology and Management ) 75±90 acidity represented 38±45% of the total acidity in bulk deposition, but increased to 61±79% in throughfall 1996±1997, compare Table 3). These results are in the same range as reported by Lovett et al. 1985) for deciduous forest stands in eastern Tennessee weak acidity as percent of total acidity was 44% in wet deposition and 58±66% in throughfall). It may not be justi ed to interpret these weak acidity data, since none of the results are statistically signi cant. However, when increased leaching of protonated organic anions reduces H driven cation exchange, the new hypothesis would be supported, that is, Ca supply diminishs base cation foliar leaching due to acidic rain Precipitation acidity All base cations except Na showed signi cant positive coef cients for precipitation H concentrations Table 5). This nding indicated that, in general, increasing acidity of precipitation caused increased leaching of Ca, Mg and K. The coef cients of the C H term i.e. amount of Ca, Mg or K leached per unit increase in H concentration in precipitation) did not differ signi cantly between the treatments but was different between the cations in decreasing order: K>Ca>Mg Fig. 3). This surprising result was con- rmed by regressing Ca NTF against H NTF, using all rain events during the growing seasons 1996±1997 nˆ112 per treatment): for all base cations regression coef cients were more negative for the control than for the CaCl 2 treated sites signi cant coef cients for control, CaCl 2 and AlCl 3 sites: Ca 0.45, 0.38, and 0.36; Mg 0.23, 0.21, and 0.23; K 0.81, 0.73, and 0.79). Again, there seems to be a trend of reduced acidity effects on cation leaching for the CaCl 2 treated sites. Experimental spraying of arti cial rain on plants has shown acid-induced leaching, but in general effects on leaching were more pronounced for Ca and Mg than for K or Na e.g. Lovett and Hubbell, 1990; Sayre and Fahey, 1999). To our knowledge the only demonstration that foliar leaching of cations increases in response to increases in precipitation acidity in naturally occurring rain events was given by Lovett et al. 1996). However, these authors found signi cant acidity effects on leaching of Ca and Mg only, and not, as we report, for K Precipitation ion concentrations Calcium, Mg and K showed signi cant ef ux from the canopy with increasing ion concentration in bulk) precipitation. While the relative contribution of the C x term see Eq. 2)) to NTF is small for Ca <9% of Fig. 3. Regression coefficients for the precipitation acidity term b 3 ) in Eq. 2), for regressions of Ca, Mg and K. Units for the coefficients are mmol c m 2 of Ca, Mg or K) mmol H l 1 ) 1. Error bars are S.D.

13 T.W. Berger et al. / Forest Ecology and Management ) 75±90 87 leaching term P) and moderate for Mg <40% of P), ef ux from the canopy for K was in uenced almost in the same range up to 90%; CaCl 2 treatment 1996± 1997) by K bulk precipitation chemistry C x term) as by the precipitation amount P term). These high contributions of ion bulk precipitation chemistry to NTF is probably caused by the fact that bulk deposition collectors are continuously open, collecting substantial amounts of atmospheric particles, especially during the long dry periods between the single rain events the average antecedent period was 6.5 days for 1996±1997, however, usually it rains 2±3 times a week at Hubbard Brook; Likens and Bormann, 1995). Particles can be formed on plant surfaces by degradation of leaf tissue or by migration of salts from the interior to the exterior of leaves e.g. Lovett and Lindberg, 1984). These aerosol particles can be transported by wind and deposited to the bulk deposition collectors in clearings within the forest. Volume weighted mean bulk concentration of K was 2.7 times higher than the mean value of wet only collectors at HBEF Likens et al., 1994). They also concluded that some bulk deposition of K likely represents internal cycling e.g. pollen or plant debris from local vegetation) rather than a true ecosystem input. In all cases P<0.001), H showed increased retention more negative NTF) with increasing concentration of H in precipitation Table 5). This nding suggests that the retention of H responds to the concentration gradient between incoming precipitation and absorption sites on or in the plants. As will be discussed below, H retention is mostly attributable to passive cation exchange on the leaf surface or interior see Lovett et al., 1996). Increasing NTF for weak acidity with decreasing solute concentration in precipitation is contradictory to meachnisms of dry deposition as falsely given by the model, see above) but supports the assumption that weak acidity in NTF represents an internal source Cation exchange and effects of acidic deposition Regression analyses of NTF were useful for relative comparisons between the chemical manipulated study sites. However, canopy exchange rates of these calculations were not used for estimating total amounts of canopy exchange, since the selected single rain events for this technique did not properly represent the entire growing season. Hence, Na was used to calculate particulate interception deposition Ulrich, 1983) and the mean ratio for Na NTF/bulk precipitation ux; mean of all sites during 1996±1997) was used for all other elements, assuming the same deposition velocity for all atmospheric constituents except that interception dry) depostion of alkalinity and weak acidity was assumed to be zero). Since dry deposition does not appear to play a major role at the HBEF Lovett et al., 1996; compare small NTF for Na and SO 4 in Table 4, which are expected to re ect dry deposition), the error of this assumption is considered small. According to Ulrich 1983) canopy exchange for Na and SO 4 was assumed zero. This method provided our best estimate for total amounts of canopy exchange Table 6) for the growing seasons of 1996±1997. Table 6 Mean canopy exchange mmol c m 2 ) per growing season 1 June to 30 September) for the years 1996 and 1997, estimated according to Ulrich 1983) a Control CaCl 2 AlCl 3 SO NO Cl Alkalinity Weak acidity H K Ca Mg Na NH Cation leaching Anion leaching Cation exchange I) Cation retention Anion retention Cation exchange II) a Positive values represent leaching; negative values represent retention. Dry deposition of alkalinity and weak acidity to the forest was set at zero, assuming only internal sources for these compounds. AlkalinityˆHCO 3 weak bases. Anion leaching includes weak acidity, assumed to leach as organic anion. Cation exchange I) was calculated as the difference between cation and anion leaching, cation exchange II) as the difference between cation and anion retention. For calculating cation retention, H was adjusted for protonation of organic anions.

14 88 T.W. Berger et al. / Forest Ecology and Management ) 75±90 Charge balance consideration requires that the net exchange of ions between the canopy and impinging deposition be electrically balanced according to the following equation Lovett et al., 1985): cation retention anion retention ˆ cation leaching anion leaching 3) Taken individually, each side of 3) represents the magnitude of the cation-exchange reaction in the canopy. We give in Table 6 all of the terms of 3), from which we calculate both estimates: cation exchange I)ˆcation leaching anion leaching; cation exchange II)ˆcation retention anion retention. Calculations were described by Lovett et al. 1985), but bear repeating here: titrations were done on samples purged with N 2,sonoH 2 CO 3 is accounted for in the weak acidity values. However, any HCO 3 left in the solution would be accounted for in the alkalinity titrations. Organic anions appear in two forms in TF: those that were protonated on contact with the acidic rainfall appear in throughfall as weak acidity, while those organic anions that remained unprotonated appear as alkalinity. The actual H retention is equal to the H consumption indicated in Table 6, minus the H consumed in protonation of organic anions. These estimates of cation exchange both sides of Eq. 3)) with the total cation leaching Ca Mg K) are plotted in Fig. 4. Cation exchange accounted for 43±67% of the total cation leaching from these canopies, which is in the same range 40±60%) as estimated for deciduous stands in Tennessee by Lovett et al. 1985). Differences between the treatments were not signi cant, but the estimated amount of leached cations per growing season was lowest for the CaCl 2 Fig. 4. Mean canopy exchange from 1 June to 30 September 1996 and Positive values represent leaching, negative values represent retention.