Patterns of nutrient loss from unpolluted, old-growth temperate forests: evaluation of biogeochemical theory.
Biogeochemical cycles in temperate ecosystems have been greatly influenced by human activities over dominant regions of Europe and North America. At sub-continental scales, atmospheric deposition has altered cycles of ecologically important elements such as nitrogen (N), sulfur (S), hydrogen ion ([H.sup.+]) and base cations (sum of [Ca.sup.2+], [Mg.sup.2+], [Na.sup.+], and [K.sup.+]) (e.g., Gorham 1961, Oden 1968, Likens et al. 1977, Aber et al. 1989, Last and Watling 1991, Hedin et al. 1994). At local scales, element cycles have been altered due to forestry and agricultural practices, and due to varying processes associated with urbanization (e.g., Robertson 1986, Coleman 1989, Schulze et al. 1989, McDonnell and Pickett 1990). More subtle, indirect, biogeochemical effects may occur as natural plant communities are subjected to air pollution (Findlay and Jones 1990) or the introduction of novel species (Vitousek 1990). Given the expanding impact of human activities on the global environment, it has become increasingly difficult to find regions of temperate forest that can be classified as free from human influences. Yet studies of such unpolluted areas are of fundamental importance to ecology: (1) to provide a reference point of "natural" element cycles, against which disturbed cycles can be compared, and (2) to allow for evaluation of biogeochemical theories which have previously been developed and tested only in areas subject to human influence and/or disturbance.
Our study area [ILLUSTRATION FOR FIGURE 1 OMITTED], located on the remote western slope of the Cordillera de Piuchue in southern Chile (42 [degrees] 22[minutes] S, 74 [degrees] 03[minutes] W), presents us with a rare opportunity to examine biogeochemical cycles in well-developed, old-growth temperate forests that have not been subjected to air pollution or other significant anthropogenic influences. Because of the pristine nature and old age of these forests, we believe this area can provide important baseline information on patterns of biogeochemical cycling with little or no bias due to human activities. We here examine landscape-level losses of major elements and nutrients from 31 forested watersheds in this area. Specifically, we examine whether patterns of hydrologic loss of major nutrients follow predictions from current biogeochemical theory: that unpolluted old-growth forests should exhibit minimal or no net biotic effect on patterns of element loss. This prediction (hereafter referred to as the "nutrient retention hypothesis") was developed formally by P.M. Vitousek, W. A. Reiners, and E. Gotham (Vitousek and Reiners 1975, Gorham et al. 1979; discussed in Peet ) as part of a general debate about how the degree of biotic control on biogeochemical cycles is expected to change during the course of ecosystem succession (Odum 1969, Henderson 1975, Woodmansee 1978, Bormann and Likens 1979, Peel 1992). The nutrient retention hypothesis predicts that the net effect of biotic retention on element input/output budgets, while significant in rapidly growing mid-successional forests, should be inconsequential in undisturbed old-growth forests. Specifically, it was argued that since net ecosystem production is expected to be negligible in old-growth forests, net demands on nutrients supplied by atmospheric inputs or weathering should also be negligible, resulting in little or no overall retention of nutrients (Vitousek and Reiners 1975, Gorham et al. 1979; discussed in Peet ). This view of old-growth temperate forest ecosystems as nutrient-rich relative to biotic demands and as inherently poor at retaining added nutrients (i.e., "leaky" with respect to inputs), constitutes an important assumption of current theories on how nutrient supply rates constrain plant communities throughout succession (Tilman 1988, Vitousek et al. 1989), and theories predicting how different-age temperate forests respond to increased anthropogenic N inputs (e.g., "nitrogen saturation hypothesis," Agren and Bosatta 1988, Aber et al. 1989, 1991, Van Miegroet et al. 1992). For example, if old-growth forests lack significant mechanisms for net N retention (Aber et al. 1991, Van Miegroet et al. 1992), they would be particularly sensitive to developing "nitrogen saturation" in response to anthropogenic N inputs.
Despite the importance of these hypotheses for current biogeochemical theories, direct tests have been few (Vitousek and Reiners 1975, Bormann and Likens 1979, Martin 1979), contentious (Henderson 1975, Woodmansee 1978), and limited to temperate old-growth forests in areas where element cycles have been strongly influenced by air pollution inputs. We here attempt to evaluate these general theories in old-growth temperate forest ecosystems that are free of air pollution inputs and that have remained floristically stable throughout much of the Holocene.
METHODS AND SITE DESCRIPTION
Our study area included watersheds in the Cordillera Piuchue (CP) mountain range, located within the Chiloe National Park on the western slope of the coastal Cordillera of the island of Chiloe in southern Chile [ILLUSTRATION FOR FIGURE 1 OMITTED]. This Park covers nearly 45 000 ha of well-developed, temperate evergreen forests characterized by high basal areas (up to [greater than]150 [m.sup.2]/ha; Armesto et al. 1994). "High-elevation forests" ([approximately equal to]00-800 m above sea level [a.s.l.]) are coniferous and dominated by Fitzroya cupressoides (Cupressaceae), Pilgerodendron uviferum (Cupressaceae), and Tepualia stipularis (Myrtaceae). The most common forest type on an areal basis is "mid-elevation forests" ([approximately equal to]400-600 m a.s.l.). These forests are dominated by the evergreen broad-leaved species Nothofagus nitida (Fagaceae) and Drimys winteri (Winteraceae), and may contain the conifers Podocarpus nubigena and Saxegothaea lonspicua (both Podocarpaceae). "Coastal forests" occur at low elevations (0-400 m a.s.l.) and are dominated by evergreen broad-leaved species, such as Aextoxicon punctatum (Aextoxicaceae), Luma apiculata, and Amomyrtus luma (both Myrtaceae). The CP forests contain no known exotic species of plants.
Extensive areas of CP forests have not been subjected to any significant human influence. While we selected study streams to represent undisturbed watersheds, minor removal of bark or rare selective cutting of individual trees by natives may have occurred in some high-elevation stands of Fitzroya cupressoides. Natural disturbances in mid-elevation forests mainly occur as small-scale forest gaps produced by wind storms (Veblen et al. 1981, Armesto and Fuentes 1988). Large-scale disturbances due to landslides, while common on the volcanic soils of the Andes, are extremely rare in the Coastal Cordillera (Armesto and Fuentes 1988). Fires are thought to be exceedingly rare due to the moist climate at CP (Armesto and Fuentes 1988). However, Armesto et al. (1994) speculate that natural fires may have occurred infrequently ([approximately equal to] [greater than]500-yr return interval) in high-elevation Fitzroya cupressoides stands during unusually strong droughts, causing long-term stand dynamics to be comparable to a "damped oscillation model" (sensu Peet 1992). Fitzroya cupressoides is long-lived ([greater than]2000 yr), shade-intolerant and does not, in contrast to other dominants in CP forests (e.g., Nothofagus and Podocarpus), self-regenerate under a closed forest canopy (Armesto et al. 1994).
In contrast to many European and North American temperate forests, CP forest ecosystems appear to have remained floristically and climatically stable throughout the Holocene. The study area was not covered by glaciers during the last glacial maximum (18,000 yr BP) and sheltered areas may have served as ice-free refugia for temperate forest species (Villagran 1990, 1991). Periglacial effects (e.g., solifluction) may, however, have limited the extent of vegetated area during glacial maximum (Veit and Garleff 1995). The persistent influence of westerly, oceanic, storm tracks apparently buffered the area from severe climatic changes throughout the Holocene (Villagran 1990, Markgraf et al. 1992). However, full-glacial conditions ([approximately equal to]27,000 to 14,000 yr BP) were likely somewhat colder and wetter than current conditions (Villagran 1988, Markgraf et al. 1992). Dendro-climatological analyses from a nearby region of south-central Chile do not indicate any consistent warming trend during the past [approximately equal to]3600 yr (Lara and Villalba 1993). More specifically, palynological records from [approximately equal to]700 m elevation at CP indicate a relatively stable composition of major forest taxa since the early Holocene (Villagran 1990). This floristic stability contrasts sharply with North American and European temperate forests in which dominant species have changed continuously throughout the Holocene, leading researchers to suggest insufficient time during inter- and post-glacial periods to reach "floristic equilibrium" (e.g., Davis 1976). Valdivian forest taxa associated with the low-elevation coastal forests at CP increased in abundance during the early Holocene, possibly in response to the transition to a somewhat warmer and dryer post-glacial climate (Villagran 1990).
The CP area receives [approximately equal to]250-300 cm precipitation annually and small streams and wetland areas are common (Ruthsatz and Villagran 1991). Because of the remote location of these forests, there have been no direct estimates of seasonal variations in precipitation amounts or streamwater discharge. Precipitation records from locations east of the coastal Cordillera show a summer drought period; however, it is not clear whether these data are representative also for CP. At CP, storms with significant rainfall are common during summers (L. O. Hedin and J. J. Armesto, personal observation). CP soils are wet, organic, and highly weathered, with relatively low base saturation (Holdgate 1961, Watters and Fleming 1972, Perez et al. 1991, Ruthsatz and Villagran 1991). A saturated gley layer is locally frequent at [approximately equal to]30-60 cm depth, underlain by saprolite bedrock. Fine root biomass is high (Perez et al. 1991, A. H. Johnson, personal observation), particularly in the upper 15 cm of the soil profile. While information is sparse, C:N ratios (by mass) of soil organic matter (0-10 cm depth) appear to vary among forest types, from [approximately equal to]40 in coniferous high-elevation forests, and [approximately equal to]22-36 in mid-elevation forests, to [approximately equal to]14 in coastal forests (Perez et al. 1991, Ruthsatz and Villagran 1991, Perez 1992). The geologic parent material consists of highly weathered quartz and feldspar schists of Precambrian age, dominated by quartz and albite (Watters and Fleming 1972). The lack of glaciation, presence of highly weathered soils and parent material, and low levels of Si in drainage waters suggest that supplies of elements from weathering are low (Hedin and Hetherington 1995), thus enhancing our ability to detect the role of biotic processes in controlling patterns of watershed nutrient loss.
Inputs of atmospheric pollutants to CP forests are negligible since strong westerly wind trajectories, directly from the Southern Pacific Ocean, dominate throughout the year due to a stationary subtropical high pressure ridge in the Southeastern Pacific ([ILLUSTRATION FOR FIGURE 1 OMITTED]; Taljaard 1972, Villagran 1990, Markgraf et al. 1992). Such unpolluted marine air trajectories are typically dominated by marine sea-salt aerosols in precipitation and dry deposition inputs (Galloway et al. 1982, Keene et al. 1986, Warneck 1988). Local air pollution emissions are few in the area surrounding CP, limited to heating (mainly wood stoves), vehicles, and infrequent biomass burning during dry periods. Any such emissions occur downwind from our study sites, on the eastern side of the Coastal Cordillera on Chiloe Island. In fact, global-scale patterns of atmospheric deposition of oxidized N, based on a Global Chemical Transport Model (GCTM; Geophysical Fluid Dynamics Laboratory, National Oceanic and Atmospheric Administration; [approximately equal to]265 km horizontal resolution; see Galloway et al. 1994), indicate that the CP area receives among the lowest worldwide rates of N deposition to terrestrial ecosystems (Galloway et al. 1994). While the Galloway et al. (1994) GCTM model predicts a N deposition rate of [less than]0.14 kg[center dot][ha.sup.-1][center dot][yr.sup.-1], we adopted the rate of [less than]1.0 kg[center dot][ha.sup.-1][center dot][yr.sup.-1] as a conservative estimate of total N deposition to the CP area. The conservative nature of this estimate is further supported by measures of wet N deposition ([approximately equal to]0.1 kg[center dot][ha.sup.-1][center dot][yr.sup.-1]) from a second remote site in southern Chile, Torres del Paine (Likens et al. 1987, Galloway et al. 1995).
Sampling approach and statistics
We used a spatially extensive sampling approach in order to discern general patterns of element loss at the landscape level. While we do not know the exact disturbance histories of individual watersheds, we assume that this approach will provide a view of the average landscape-level condition for element balances: net accumulation, net loss, or landscape-level steady-state (cf., "shifting mosaic model" sensu Bormann and Likens ). We sampled streams draining 31 small ([Mathematical Expression Omitted] size: [approximately equal to]0.3 [km.sup.2]) watersheds on the western slope of CP, representing coastal (n = 5), mid-elevation (n = 11) and high-elevation (n = 15) forests along a west-to-east transect from the coast to 15 km inland. Watershed streams were sampled during summers of 1989 (n = 5 streams), 1992 (n = 16), and 1993 (n = 19) and early autumn (March) of 1993 (n = 5). Our sampling efforts represent different hydrological regimes in that 1992 data are from a dry period, while 1993 data are from a very rainy period. Some streams were sampled two or three times to evaluate variations in chemistry between sample events (see analysis below). Because of possible bias due to differences in sampling and handling techniques, we have not included preliminary data from 1989 in our analyses. The 1989 data (Hedin and Campos 1991) generally support our present analyses and have, for comparative purposes, been included in Fig. 4.
We weighted each watershed equally in our analyses of patterns of element loss. Thus, watersheds sampled more than once were characterized by the average chemistry of individual sample events. To decrease statistical bias, we recorded any concentrations below method detection limits as one half the detection limit, rather than as zero (Newman et al. 1989). In the case of elements with non-normal distributions, we report either median or geometric mean concentrations.
In the field, duplicate water samples were filtered immediately through pre-rinsed Gelman A/E glass fiber filters ([less than]1 [[micro]meter] nominal pore-size, tested in our lab), stored in thoroughly rinsed clean polyethylene bottles, and kept on ice or as cool as possible. Rinsed bottles were randomly tested for trace-level contamination. In the field, high-pressure-liquid-chromatography-grade chloroform (1% of final volume) was immediately added to one of the duplicate samples. Samples were shipped by air to the Kellogg Biological Station (Michigan, USA), and analyzed for major ions ([Ca.sup.2+], [Mg.sup.2+], [Na.sup.+], [K.sup.+], N[H.sub.[4.sup.+]], N[O.sub.[3.sup.-]], S[O.sub.[4.sup.2-]], and [Cl.sup.-]), dissolved silica (Si), dissolved organic nitrogen (DON), dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC). Sensitive dissolved species were analyzed within 8-10 d; pH was determined in the field (range: 4.3-6.7).
Because of exceedingly low N[O.sub.[3.sup.-]] in CP streams, we used an analytical technique that allowed detection of N at very low levels ([less than]0.010 [[micro]gram]/L) during 1993. Depending on need, we pre-concentrated between 1 and 5 mL of sample on a Dionex AS4G column. The sample was injected and analyzed for N[O.sub.[3.sup.-]] on a Dionex ion chromatograph, using chemically suppressed conductivity detection and a Dionex AS4A column. Throughout this paper charge concentration is expressed in units of micromoles of charge per litre of solution. Other analytical methods and minimum detection limits (in parentheses) were: S[O.sub.[4.sup.2-]] (0.1 [[micro][mol.sub.c]]/L), [Cl.sup.-] (0.2 [[micro][mol.sub.c]]/L) and N in N[O.sub.[3.sup.-]] in 1992 (0.002 mg/L) by Dionex chemically suppressed ion chromatography using AS4A column (Burke et al. 1989); N[H.sub.[4.sup.+]]-N (0.003 mg/L in 1992 and 0.001 mg/L in 1993) and Si ([less than]0.05 mg/L) by Alpkem automated colorimetry (Alpkem 1992a, b); pH by potentiometry, using a Ross electrode (Orion Research, Boston, Massachusetts, USA); [Ca.sup.2+] ([less than]0.5 [[micro][mol.sub.c]]/L) and [Mg.sup.2+] ([less than]0.8 [micro][mol.sub.c]/L) by Perkin-Elmer Inductively Coupled Plasma Emission Spectroscopy in 1989 and 1992 and by Perkin-Elmer Flame Atomic Absorption Spectrophotometry (AAS) in 1993; [Na.sup.+] ([less than]0.4 [[micro][mol.sub.c]]/L) and [K.sup.+] ([less than]0.3 [[micro][mol.sub.c]]/L) by AAS; DOC (0.1 mg/L) and DIC (0.1 mg/L) by Ionics high-temperature (850 [degrees] C) platinum catalyst combustion; DON (0.01 mg/L) by high-temperature persulfate digestion with subsequent cadmium reduction colorimetric analysis of N[O.sub.[3.sup.-]] (Alpkem 1992c; modified from McDowell et al. 1987 and D'Elia et al. 1977). Persulfate digestion efficiency was tested against amino acid, N[O.sub.[3.sup.-]], and N[H.sub.[4.sup.+]] standards and by ensuring that all detectable DOC was digested, using a post-digestion DOC analysis by high-temperature (850 [degrees] C) platinum catalyst combustion. Following Hedin et al. (1990b), we calculated concentrations of organic anions as the discrepancy between cationic and anionic equivalent charges: [Sigma] ([[Ca.sup.2+]] + [[Mg.sup.2+]] + [[Na.sup.+]] + [[K.sup.+]] + [[H.sup.+]] + [N[H.sub.[4.sup.+]]]) - [Sigma] ([S[O.sub.[4.sup.2-]]] + [N[O.sub.[3.sup.-]]] + [[Cl.sup.-]] + [HC[O.sub.[3.sup.-]]] + [C[O.sub.[3.sup.2-]]]). A bias may exist for low-pH samples since we did not analyze for [Al.sup.n+].
We used two tests to evaluate sample integrity between times of collection and analysis. First, we collected two sets of duplicate filtered and preserved samples from eight relatively remote streams of the Cordillera Pelada in southern Chile (40 [degrees] 10[minutes] S, 73 [degrees] 35[minutes] W). One set of samples was sent by express air courier to our lab and analyzed within 5 d, while the second set was kept at cool field-conditions or refrigerated for 13 d before analysis. We found no significant differences (P [greater than] 0.05; Wilcoxon signed-rank test) in concentrations of the most sensitive chemical species (N[H.sub.[4.sup.+]], N[O.sub.[3.sup.-]], DON, and DOC) between these two sets. Second, tests using Michigan stream water of varying chemical composition (high to low N levels) showed that in situ filtration alone was sufficient to preserve samples for at least 3 wk. As an additional check on sample integrity, however, we consistently preserved one of each duplicate sample from CP by chloroform addition (1% of final volume).
Between-year variations in chemistry
We re-sampled six streams during the summers of 1992 and 1993 to evaluate the magnitude of variation in chemistry between two summer-period sampling events. We wanted to evaluate how well individual samples of watershed streams could be used to characterize strong landscape-level patterns in chemistry. In fact, we found only relatively small variations in overall chemistry between the sample events [ILLUSTRATION FOR FIGURE 2 OMITTED]. The greatest deviation from a 1:1 relationship, indicated by the line in Fig. 2, was found for N[O.sub.[3.sup.-]] in four watersheds. We attribute these differences in N[O.sub.[3.sup.-]] to changes in analytical techniques: while N levels were below detection limits in 1992 ([less than]2 [[micro]gram]/L), they were above the new detection limit for the concentrator-column technique used in 1993 (0.010 [[micro]gram]/L). Overall, this evaluation suggests that within-stream variations were low relative to the spatial patterns in chemistry that our landscape-scale analyses are based on.
Data from other studies
Biogeochemical studies of old-growth temperate forests are rare. We compared our results against data on streamwater or deep soil-solution chemistry from nine other old-growth temperate forest ecosystems. We also used estimates of atmospheric wet or bulk inorganic N deposition from these studies (see Discussion). Due to logging and other human disturbances, data from European old-growth forests are entirely lacking. While studies of old-growth forests in the eastern United States are rare, data from the Pacific northwest area of North America are more common. The biogeochemistry of the H. J. Andrews Experimental Forest (AND) in the Cascade Mountains of Oregon, USA, has been studied extensively by Sollins et al. (1980), Sollins and McCorison (1981), and others. Watershed 9 is covered by old-growth Douglas-fir (Pseudotsuga menziesii) forest with an understory of western hemlock (Tsuga heterophylla), and with soils developed on volcanic parent materials. We used average streamwater data for a 5-yr period from Sollins and McCorison (1981). Values of N[O.sub.[3.sup.-]] in Sollins and McCorison (1981) are slightly lower than earlier (Sollins et al. 1980) reported N values ([approximately equal to]3 vs. 16 [[micro]gram]/L).
The Jamieson Creek watershed (JAM) in British Columbia, Canada, (Zeman and Slaymaker 1978) is located in the western hemlock (Tsuga heterophylla) biogeoclimatic zone and drains acid igneous rock (quartz diorite) parent material. We used volume-weighted annual mean streamwater data from weekly samples over a 1-yr period (Zeman and Slaymaker 1978). The Indian River watershed (IND) drains a Sitka spruce (Picea sitchensis) - western hemlock (Tsuga heterophylla) forest in a mixed sedimentary and volcanic terrane in southeastern Alaska, USA (Stednick 1981). Alders (Alnus rubra and A. sinuata) line some streams. We used volume-weighted annual average streamwater data from a 4-yr period based on biweekly to monthly samples. Data from a Douglas-fir (Pseudotsuga menziesii) - western hemlock (Tsuga heterophylla) forest at Findley Lake (FIN), Washington, USA, were summarized in Johnson and Lindberg (1992) as part of the Integrated Forest Study project. Because streamwater data are lacking, we based comparisons on soil-solution chemistry from lysimeters located in the Bs horizon ([approximately equal to]40 cm depth). Data from the western Olympic Peninsula (OLY) in Washington State, USA, derive from a 2-yr watershed study of an old-growth western hemlock and Pacific silver fir (Abies amabilis) forest (Larson 1979). Douglas-fir (Pseudotsuga menziesii), western red cedar (Juniperus virginiana), Sitka spruce, and red alder (Alnus rubra) were less common. Average volume-weighted streamwater chemistry was collected approximately every 2 wk over 2 yr. Because N[H.sub.[4.sup.+]] often was at or below detection (Larson 1979), we used the N detection limit (0.04 mg/L) as a lower limit approximation of N[H.sub.[4.sup.+]] in stream water and precipitation. N[H.sub.[4.sup.+]] may thus be slightly overestimated relative to N[O.sub.[3.sup.-]].
The Bowl forest (BOW), New Hampshire, USA, is dominated by an old-growth deciduous forest of sugar maple (Acer saccharum), yellow birch (Betula lutea), and beech (Fagus grandifolia), with stands of red spruce (Picea rubens) and balsam fir (Abies balsamea) at high-elevations (Martin 1979). The geology consists of syenite bedrock overlain by glacial till derived from granite. We chose unweighted average data from four streams, based on collections every 2 wk over 18 mo (Martin 1979). Concentrations of N[O.sub.[3.sup.-]] did not vary seasonally. We used streamwater data from an old-age spruce-fir (Picea rubens and Abies balsamea) watershed at 1143 m elevation in New Hampshire, USA (Vitousek 1977: [ILLUSTRATION FOR FIGURE 6 OMITTED]). While the exact location is not given, we infer from Vitousek (1977) that this watershed is located on Mount Moosilauke (MOU). We used N deposition values from Lovett et al. (1982). However, these values are from a site with particularly high cloud water N deposition and thus most likely overestimates N deposition to the Vitousek site. We used unweighted averages based on collections every 2 wk to monthly over 16 mo. While N[H.sub.[4.sup.+]] was generally not detectable, we used the detection limit (1 [[micro][mol.sub.c]]/L) as a conservative estimate for our analyses of N[O.sub.[3.sup.-]]:N[H.sub.[4.sup.+]] ratios. In addition, we used data from an old-growth subalpine spruce-fir (Picea rubens and Abies balsamea) forest on Whiteface Mountain (WHI), New York, USA, located on anorthosite bedrock (Friedland et al. 1991). Because drainage streams are lacking, we based comparisons on soil-solution chemistry from lysimeters located in the Bs horizon ([approximately equal to]45-60 cm depth). Data are averages for the period 1986-1990. Cloud-water deposition was a significant additional input of N (Friedland et al. 1991). We used data from an old-growth red spruce (Picea rubens) forest in the Great Smoky Mountains (SMO), North Carolina, USA, which is subject to high atmospheric N deposition (Johnson et al. 1991). We used soil-solution chemistry samples from lysimeters located in the Bw2 horizon ([approximately equal to]49-55 cm depth) at the Tower site on Clingman's Dome. Our values derive from average volume-weighted concentrations based on monthly samples over 3 yr. Studies at the Tower site showed that atmospheric N deposition is strongly ([greater than]80%) dominated by dry deposition and cloud-water deposition inputs (Johnson et al. 1991).
Use of [Cl.sup.-] as a watershed-level tracer
We used [Cl.sup.-] as a watershed-level hydrologic tracer for atmospheric aerosol inputs of major elements to CP forests. In silicate terranes, [Cl.sup.-] is not notably produced by silicate weathering and does not participate to any consequential degree in biological cycling or soil anion exchange reactions (Gorham 1961, Gibbs 1970, Likens et al. 1977). Because major elements ([Ca.sup.2+], [Mg.sup.2+], [Na.sup.+], [K.sup.+], and S[O.sub.[4.sup.2-]]) occur in strict and constant ratios relative to [Cl.sup.-] in sea-salt aerosols of unpolluted marine air masses (Keene et al. 1986, Warneck 1988), we used watershed-level changes in element:[Cl.sup.-] ratios to indicate net retention or net release of elements within the forest ecosystem. Streamwater chemistry was corrected for the influence of sea-salt aerosols based on the following sea-salt charge ratios relative to [Cl.sup.-] (Keene et al. 1986): [Na.sup.+] = 0.862; [K.sup.+] = 0.0188; [Mg.sup.2+] = 0.196; [Ca.sup.2+] = 0.0378; S[O.sub.[4.sup.2-]] = 0.104.
Net loss of an element from the watershed ecosystem - e.g., due to weathering release or mineralization during organic matter decomposition - would result in streamwater outputs with element: [Cl.sup.-] ratios ([R.sub.out]) greater than the element: [Cl.sup.-] ratio ([R.sub.ss]) of sea-salt aerosol (i.e., [R.sub.out] [greater than] [R.sub.ss]). Conversely, net retention of an element within the watershed - e.g., due to immobilization by vegetation, microbes, or secondary mineral formation - would result in streamwater element:[Cl.sup.-] ratios lower than sea-salt aerosol ([R.sub.out] [less than] [R.sub.ss]). An overall lack of net retention of atmospheric inputs by the forest ecosystem would produce, as predicted by the nutrient retention hypothesis, streamwater outputs with element: [Cl.sup.-] ratios that remain unchanged relative to sea-salt aerosol ([R.sub.out] = [R.sub.ss]). This last situation is consistent with the earliest formulation of the nutrient retention hypothesis: that ecosystem outputs should equal atmospheric inputs at the condition of no or negligible net ecosystem production (Vitousek and Reiners 1975). This prediction was, however, later modified to explicitly include weathering as an input term (Henderson 1975, Gorham et al. 1979), although Gorham et al. (1979) argued that weathering inputs may likely be minimal in geologically old, highly weathered, silicate terranes such as at CP.
Chloride as a tracer for aerosol inputs
Streams draining forested Cordillera de Piuchue (CP) watersheds were dilute with low, but varying, levels of total dissolved inorganic salts (median = 0.3 m[mol.sub.c]/L; range: 0.2-4.2). Dissolved Si was low (median = 1.0 mg/L; range: 0.4-2.4 mg/L), indicating a modest influence by silicate weathering reactions (discussed below in Element loss. . .; Hedin and Hetherington 1995). The chemistry of watershed streams was dominated [TABULAR DATA FOR TABLE 1 OMITTED] by [Cl.sup.-] and [Na.sup.+], occurring at a median Na:Cl charge ratio (0.93) near that expected if sea-salt aerosols were the dominant source for these elements (0.86). Further evidence that atmospheric inputs of marine aerosols were the dominant source of [Cl.sup.-] to CP watersheds comes from the conspicuous pattern of exponential decline in streamwater levels of [Cl.sup.-] as a function of increasing distance of watersheds from the ocean ([ILLUSTRATION FOR FIGURE 3 OMITTED]; [r.sup.2] = 0.96; P [less than] 0.001; log-log regression).
Patterns of loss of major elements
Because most sample efforts occurred during summer, a time of high biological activity, we assume that our samples reflect the watershed-level integration of water that interacted with a biologically active over-story and soil column before emerging as streamwater output. Streamwater outputs from most CP watersheds contained exceedingly low levels of dissolved elements. For example, median concentrations of [Ca.sup.2+], [K.sup.+], S[O.sub.[4.sup.2-]], N[H.sub.[4.sup.+]] and N[O.sub.[3.sup.-]] (Table 1) were substantially lower than in waters draining feldspathic areas (e.g., Likens et al. 1977, Stauffer 1990, Hedin and Hetherington 1995), and were even lower than, or similar to, levels found in precipitation in areas of North America and Europe (National Research Council 1986, Hedin et al. 1990a). Despite these dilute levels, however, concentrations of major anions and cations in CP watershed streams showed strong and linear relationships when regressed against [Cl.sup.-] [ILLUSTRATION FOR FIGURE 4 OMITTED]; for [Na.sup.+] [r.sup.2] = 0.98, for S[O.sub.[4.sup.2-]] [r.sup.2] = 0.88, for [K.sup.+] [r.sup.2] = 0.92, for [Mg.sup.2+] [r.sup.2] = 0.98, and for [Ca.sup.2+] [r.sup.2] = 0.61; P [less than] 0.001 for all regressions). Thus, hydrologic losses of major elements from CP watersheds occurred in an overall pattern of strict ratios relative to [Cl.sup.-] [ILLUSTRATION FOR FIGURE 4 OMITTED]. Compared to other elements, [Ca.sup.2+] showed a more variable relationship against [Cl.sup.-] [ILLUSTRATION FOR FIGURE 4B OMITTED].
Element loss relative to aerosol inputs
The solid lines in Fig. 4 show expected concentrations of individual ions, assuming they occur in exact sea-salt aerosol ratios relative to the [Cl.sup.-] tracer (i.e., [R.sub.out] = [R.sub.ss]). With the exception for [Ca.sup.2+] [ILLUSTRATION FOR FIGURE 4B OMITTED], the observed relationships between major ions and CI in CP watershed streams were close to the expected sea-salt ratios [ILLUSTRATION FOR FIGURE 4A OMITTED]. In fact, subtraction of the contribution of atmospheric sea-salt aerosol to streamwater chemistry resulted in strong reductions of all major elements (Table 1), with aerosol inputs alone explaining close to 100% of S[O.sub.[4.sup.2-]], [Cl.sup.-], [K.sup.+], and [Na.sup.+], and [approximately equal to]75% of [Mg.sup.2+], but a lesser amount of [Ca.sup.2+] (25%). After such aerosol correction, concentrations of S[O.sub.[4.sup.*]], [K.sup.*] and [Cl.sup.*] (* denotes sea-salt-corrected value) were not significantly different from zero (t tests: P [greater than] 0.26 for S[O.sub.[4.sup.2-]], P [greater than] 0.14 for [K.sup.+], [Cl.sup.-] by assumption; Table 1). Concentrations of [Ca.sup.*], [Mg.sup.*] and [Na.sup.*], while statistically different from zero (t tests: P [less than] 0.001 for all; Table 1), were only slightly so when expressed on the basis of absolute concentrations ([Ca.sup.*] = 14.1 [[micro][mol.sub.c]]/L or 0.28 mg/L; [Mg.sup.*] = 7.9 [[micro][mol.sub.c]]/L or 0.096 mg/L; and [Na.sup.*] = 8.1 [[micro][mol.sub.c]]/L or 0.19 mg/L). Subtraction of the contribution from aerosol inputs thus resulted in much reduced levels of base cations in CP streams, even more substantially below other naturally dilute waters draining glaciated or unglaciated feldspathic areas (Stauffer 1990). Overall, correction for sea-salt aerosol produced a 95% decrease in total dissolved inorganic solids - from 10.4 to 0.56 mg/L - in the median chemistry of CP watershed streams. The general pattern of hydrologic loss of major elements in CP watersheds was thus very close (within 93% for S[O.sub.[4.sup.2-]], [Cl.sup.-], [K.sup.+], and [Na.sup.+]; within 75% for [Mg.sup.2+]) to predictions based solely on atmospheric aerosol inputs, with the notable exception for [Ca.sup.2+], which was elevated [approximately equal to]75% above aerosol predictions.
While slight on the basis of absolute concentration, the statistically significant concentrations of [Ca.sup.*], [Mg.sup.*], and [Na.sup.*] indicate a weak source for these elements in CP watersheds. It is, however, difficult to interpret these corrected values since, with the exception of [Ca.sup.*], they represent small differences between two relatively large values (i.e., observed and predicted). However, two lines of evidence suggest that the slight residual levels of base cations were contributed from silica weathering reactions. First, variations in [Ca.sup.*], [Mg.sup.*], and [Na.sup.*] were all positively correlated with streamwater concentrations of dissolved Si (Pearson's r [greater than] 0.54 and P [less than] 0.002 for each element). Second, the average molar ratio of Si:[Na.sup.*] from 29 watersheds (two extreme values excluded) was 2.15 (SE = 0.31; n = 29), close to the theoretical expectation of 2.0 for the weathering of plagioclase feldspar to kaolonite (Taylor and Velbel 1991). While more variable, ratios of [Ca.sup.*]:[Na.sup.*] were particularly low ([Mathematical Expression Omitted]; SE = 0.4; n = 29) when compared to values from other feldspathic areas (Stauffer 1990), suggesting the presence and slight weathering of anorthite or Ca-rich minerals such as epidote (Watters and Fleming 1972).
Patterns of nitrogen loss
Streamwater losses from CP watersheds were characterized by exceptionally low concentrations of inorganic forms of N ([ILLUSTRATION FOR FIGURE 5 OMITTED] and Table 2). Levels of N[O.sub.[3.sup.-]]-N were particularly low (geometric [Mathematical Expression Omitted]; range: 0.05-12 [[micro]gram]/L; Table 2). In fact, as far as we are aware, high- and mid-elevation forests in our study area showed the lowest efflux concentrations of N[O.sub.[3.sup.-]]-N reported from any old-growth temperate forest ecosystem (Table 2). While also low, levels of N[H.sub.[4.sup.+]]-N (geometric [Mathematical Expression Omitted]) were [approximately equal to]20 times higher than N[O.sub.[3.sup.-]], and varied only slightly among watersheds (range: 3.0-25 [[micro]gram]/L; [ILLUSTRATION FOR FIGURE 5 OMITTED] and Table 2).
In contrast to inorganic forms of N, we found dissolved organic forms of N (DON) to be the dominant hydrologic vector of N loss from CP watersheds [ILLUSTRATION FOR FIGURE 5 OMITTED]. While we did not measure hydrologic losses of particulate organic N (PON), this vector is small ([less than]3% of total N) in other temperate forests (Likens et al. 1977). Average concentrations of DON (geometric [Mathematical Expression Omitted]; range: 83-421 [[micro]gram]/L) were [approximately equal to]400 and 20 times higher, respectively, than N[O.sub.[3.sup.-]] or N[H.sub.[4.sup.+]]. Thus [approximately equal to]95% (range: 87-99%) of all hydrologic N losses from CP watersheds occurred as DON (Table 2). In contrast, N[O.sub.[3.sup.-]] contributed only [approximately equal to]0.2% (range: 0.04-8.9%), and N[H.sub.[4.sup.+]] only [approximately equal to]4.8% (range: 0.8-13%) to overall hydrologic N losses. Levels of DON varied approximately six-fold among watershed streams, and were significantly correlated with among-watershed variations in DOC ([ILLUSTRATION FOR FIGURE 6A OMITTED]; [r.sup.2] = 0.51; P [less than] 0.001; log-log regression). A slightly stronger relationship [ILLUSTRATION FOR FIGURE 6B OMITTED] ([r.sup.2] = 0.56; P [less than] 0.001; log-log regression) existed between DON and the calculated anion discrepancy ([Sigma] cationic charge - [Sigma] anionic charge), an approximate measure of the concentration of total organic anions (Hedin et al. 1990a). These correlations suggest that watershed-level differences in DON loss followed variations in leaching losses of organic acid anions associated with dissolved humic materials.
With the exception of N[O.sub.[3.sup.-]], the overall pattern of hydrologic N loss from CP watersheds did not vary markedly among the three major forest types. Losses of N were uniformly dominated by DON, comprising 94% of total N in coastal forests, 95% in high-elevation, and 96% in mid-elevation forests (Table 2). Streamwater levels of N[H.sub.[4.sup.+]] also did not differ significantly between forest types (P [greater than] 0.14; ANOVA), with N ranging from 8 [[micro]gram]/L in high-elevation forests to 10 [[micro]gram]/L in coastal forests (Table 2). In contrast, N[O.sub.[3.sup.-]] concentrations differed significantly between forest types (P [less than] 0.0001; ANOVA), with levels in coastal forest streams [approximately equal to]40 and 14 times higher than high-elevation and mid-elevation forest streams, respectively (Table 2). While severely acidic conditions may in some cases inhibit nitrification (Schmidt 1982; but see Robertson 1982, Stams et al. 1991), we found no significant correlation (P [greater than] 0.99; n = 31) between streamwater N[O.sub.[3.sup.-]] and acidity (pH range: 4.3-6.5), nor did streamwater pH vary consistently among forest types (Table 2). The observed pattern of N[O.sub.[3.sup.-]] loss appears, however, to correlate with coarse-scale variations in soil C:N ratios. While data are sparse (Perez et al. 1991, Ruthsatz and Villagran 1991, Perez 1992), soil organic matter C:N ratios show a successive decrease from high-elevation gymnosperm forests ([approximately equal to]40) to mid-elevation ([approximately equal to]22-36) and coastal angiosperm-dominated forests ([approximately equal to]14). Irrespective of forest type, however, N[O.sub.[3.sup.-]] comprised only a negligible fraction of total N in CP watershed streams, from a minimum of 0.06% in high-elevation forests to a maximum of 1.8% in coastal forests (Table 2).
Lack of net biotic effect on inorganic element losses
There is long-standing interest among ecologists in the extent to which biotic processes control patterns of nutrient loss from terrestrial ecosystems (Gorham 1961, Odum 1969, Vitousek and Reiners 1975, Bormann and Likens 1979, Gorham et al. 1979). Several studies have shown strong retention of key nutrients (e.g., N, [K.sup.+], or [Ca.sup.2+]) by vegetation and soil microbes in aggrading early and mid-successional temperate forests (e.g., Vitousek and Reiners 1975, Bormann et al. 1977, Likens el. al. 1977, Stone and Kszystniak 1977, Monk and Day 1988). For old-growth temperate forests that show no or minimal net ecosystem production (NEP), however, current theory predicts that the net biotic effect on element retention should be minimal (Vitousek and Reiners 1975, Gorham et al. 1979). Information from such old-growth forests is presently sparse, and does not necessarily reflect truly "undisturbed" patterns of nutrient cycling since major forested regions in Europe, North America, and other industrial areas are subject to atmospheric deposition of N and other pollutants. Information from temperate old-growth forests in truly unpolluted areas of the world is particularly important for assessing the biogeochemical conditions that have acted as evolutionary constraints on plant and microbial communities before the advent of regional-scale anthropogenic impacts on biogeochemical cycles.
The Cordillera de Piuchue (CP) forests of southern Chile provide an example of well-developed, unpolluted old-growth forests, subject to low disturbance levels. These forests have developed over time within the biogeochemical constraints of low weathering rates from the local geology and of natural atmospheric conditions, dominated by dilute sea-salt aerosol inputs. While direct measures of NEP are currently lacking, we assume that our extensive sampling approach allows for characterization of the average landscape-level condition of CP forest communities (cf. "shifting mosaic model" sensu Bormann and Likens ).
For major inorganic elements our results support the prediction of the nutrient retention hypothesis, that net ecosystem demands on elements should be minimal in old-growth forests, resulting in little or no retention of externally added nutrients (Vitousek and Reiners 1975, Gorham et al. 1979). In fact, with the exception of [Ca.sup.2+] and [Mg.sup.2+], losses of major inorganic elements from CP watersheds were [greater than]93% of predictions based solely on atmospheric deposition inputs of sea-salt aerosol (i.e., [R.sub.out] [congruent] [R.sub.ss], where [R.sub.out] = element: [Cl.sup.-] streamwater output ratio, and [R.sub.ss] = element: [Cl.sup.-] ratio of sea-salt aero-sol) ([ILLUSTRATION FOR FIGURE 4 OMITTED] and Table 1). Furthermore, direct subtraction of the sea-salt aerosol contribution to element losses from watersheds resulted in the near-complete reduction of total inorganic salts (Table 2), with only trace levels of [Ca.sup.2+], [Mg.sup.2+], and [Na.sup.+] remaining. Thus, overall patterns of hydrologic loss of major elements from CP forest ecosystems were primarily controlled by atmospheric inputs of dilute sea-salt aerosol [ILLUSTRATION FOR FIGURE 4 OMITTED]. While exceedingly low, residual levels of [Ca.sup.*], [Mg.sup.*] and [Na.sup.*] (* denotes sea-salt-corrected value; see Table 1) were positively correlated with between-watershed variations in Si. We interpret these correlations to indicate that, with the exception of [Ca.sup.*], weathering processes contributed only minimally to overall patterns of element loss from CP watersheds.
The overall lack of element retention within CP watersheds located [greater than] 1 km inland deserves special attention. Because many dissolved elements were exceptionally dilute in these streams (total salts [less than] 0.4 m[mol.sub.c]/L), we would expect ion ratios to be particularly sensitive to internal watershed processes such as biotic demand. Yet, observed element ratios remained close to sea-salt predictions even in these streams [ILLUSTRATION FOR FIGURE 4 OMITTED], indicating little or no net effect of biological processes on the retention of these elements. For example, while [K.sup.+] levels were exceedingly dilute (median = 2.4 [[micro][mol.sub.c]]/L), there was no statistical evidence for watershed retention of [K.sup.+] relative to aerosol inputs (i.e., [R.sub.out] = [R.sub.ss]; Table 2). Yet, [K.sup.+] is a critical nutrient subject to exceptionally strong biotic demands in aggrading early- and mid-successional temperate forest ecosystems (Likens et al, 1977, Stone and Kszystniak 1977, Monk and Day 1988). Furthermore, the observed levels of [K.sup.+], [Ca.sup.2+], and to a lesser extent [Mg.sup.2+] in inland CP watershed streams should theoretically be sensitive to moderate rates of biotic demand. Based on net vegetation demands of elements reported from Northern Hemisphere temperate forests of varying successional ages ([K.sup.+] = 150-1400 [mol.sub.c][center dot][ha.sup.-1][center dot][yr.sup.-1]; [Ca.sup.2+] = 275-3000 [mol.sub.c][center dot][ha.sup.-1][center dot][yr.sup.-1]; and [Mg.sup.2+] = 70-1400 [mol.sub.c][center dot][ha.sup.-1][yr.sup.-1]; Johnson and Lindberg 1992), and the conservative assumption that biotic demands occur evenly throughout the year, we calculate the following theoretical reductions of yearly average element concentrations in CP streams: [K.sub.+]: 6-55 [[micro][mol.sub.c]]/L; [Ca.sup.2+] = 11-120 [[micro][mol.sub.c]]/L; and [Mg.sup.2+] = 3-55 [[micro][mol.sub.c]]/L based on 250 cm of annual streamflow. Thus, with the exception of [Na.sup.+], the patterns of element loss in Fig. 4 should be sensitive to even modest rates of net biotic demand.
Our results indicate that background S[O.sub.[4.sup.-]] levels are near zero in CP forests. We found no statistically significant levels (positive or negative) of S[O.sub.[4.sup.2-]] remaining after sea-salt correction (Table 1), indicating that watershed-level S budgets are, on average, balanced relative to atmospheric inputs to these old-growth forest ecosystems. Even when uncorrected for aerosol inputs, levels of S[O.sub.[4.sup.2-]] were an order of magnitude lower in CP streams than surface waters draining acid-deposition-impacted temperate forests in the northeastern U.S. (e.g., Brakke et al. 1988) and in Europe (e.g., Henriksen and Brakke 1988). When compared to these polluted areas, our results support contentions that atmospheric deposition has caused substantial changes in the biogeochemical cycle of S in soils and waters of European and North American temperate ecosystems (e.g., Oden 1968, Last and Watling 1991).
Controls on nitrogen loss
In contrast to major inorganic elements, patterns of N loss from CP watersheds were more complex. Atmospheric deposition of N is exceptionally low ([less than]1 kg[center dot][ha.sup.-1][center dot][yr.sup.-1]) in remote unpolluted areas, such as CP, where marine wind trajectories prevail (Galloway et al. 1982, 1994, Likens et al. 1987). GCTM model results (Galloway et al. 1994) indicate that our study area has among the lowest rates of deposition of oxidized N in the world. For such conditions, the nutrient retention hypothesis predicts that low rates of atmospheric input would cause low streamwater losses of N. In addition, recent North American and European studies have advanced the idea - the "N-saturation hypothesis" - that inorganic N levels should be naturally low in streams that drain temperate forests with no atmospheric pollution inputs of N (Agren and Bosatta 1988, Aber et al. 1989). Our results support both these hypotheses, in that we found only very low levels of N[O.sub.[3.sup.-]]-N (geometric [Mathematical Expression Omitted]) as well as N[H.sub.[4.sup.+]]-N (geometric [Mathematical Expression Omitted]) in all streams draining the CP watersheds ([ILLUSTRATION FOR FIGURE 5 OMITTED] and Table 2). In contrast, waters draining old-growth forests in areas subject to elevated atmospheric N deposition, such as in the Northeastern U.S., show several orders of magnitude higher levels of N[O.sub.[3.sup.-]]-N (up to [approximately equal to]1400 [[micro]gram]/L), and up to 6 times higher levels of N[H.sub.[4.sup.+]]-N ([approximately equal to]50 [[micro]gram]/L) throughout the growing season [ILLUSTRATION FOR FIGURE 7 OMITTED].
In Fig. 7A we compare concentrations of N[O.sub.[3.sup.-]] in high-(CPH), mid-(CPM) and low-elevation (CPL) CP forests against nine other studies of undisturbed, old-growth temperate forests. Four sites, BOW (angiosperm dominance; Martin 1979), MOU (gymnosperm dominance; Vitousek 1977), SMO (gymnosperm dominance; Johnson et al. 1991), and WHI (gymnosperm dominance; Friedland et al. 1991) are located in regions of high N-deposition in the eastern U.S. The remaining sites, AND (Sollins et al. 1980), FIN (Johnson and Lindberg 1992), IND (Stednick 1981), JAM (Zeman and Slaymaker 1978), and OLY (Larson 1979) are located in gymnosperm-dominated old-growth forests of the Pacific Northwestern U.S. While deposition rates are less than the northeastern U.S. or Europe, areas of southwestern British Columbia and northwestern Washington are also subject to anthropogenic N and S deposition (Vong et al. 1985, Feller 1987).
By using wet or bulk deposition of inorganic N as a coarse correlate for total atmospheric N deposition, we constructed an approximate ranking of sites from low to high N deposition [ILLUSTRATION FOR FIGURE 7 OMITTED]. Because three high-elevation sites (WHI, MOU, and SMO) were subject to substantial cloud water N inputs (Friedland et al. 1991, Johnson et al. 1991, Lovett et al. 1982), we included cloud water N deposition (WHI = 6, MOU [greater than] 20, and SMO = 9 kg[center dot][ha.sup.-1][center dot][yr.sup.-]) in the N deposition values used to rank these sites in Fig. 7. Levels of N[O.sub.[3.sup.-]] increased by 4 orders of magnitude along this N deposition gradient, with averages for high-elevation spruce forests in the eastern U.S. up to [approximately equal to]10000 times higher than the coniferous high-elevation forests of CP [ILLUSTRATION FOR FIGURE 7 OMITTED]. Other variables such as precipitation amount, vegetation type, and quality of soil organic matter have been suggested as important controls on N[O.sub.[3.sup.-]] losses from terrestrial ecosystems (Aber et al. 1991, Aber and Melillo 1991). However, hydrologic dilution cannot explain the observed pattern in N[O.sub.[3.sup.-]] concentrations in Fig. 7, since precipitation amounts varied only 3.5-fold while N[O.sub.[3.sup.-]] varied nearly 4 orders of magnitude between sites. It is more difficult to assess any direct or indirect effects by vegetation on patterns of N[O.sub.[3.sup.-]] loss. While nutrient use efficiency, tissue or litter quality (defined as C:N or lignin:N ratios) may differ between conifer- and angiosperm-dominated forests (Vitousek 1982, Aber and Melillo 1991), it does not appear that between-site variations in N[O.sub.[3.sup.-]] were due to such effects. Gymnosperm-dominated forests showed among the lowest (CPH and AND) as well as highest (SMO and MOU) levels of N[O.sub.[3.sup.-]], while angiosperm-dominated forests varied almost as widely (e.g., CPM vs. BOW) [ILLUSTRATION FOR FIGURE 7A OMITTED].
Long-term differences in atmospheric N deposition may contribute importantly to the observed between-site variations in N[O.sub.[3.sup.-]] levels [ILLUSTRATION FOR FIGURE 7A OMITTED]. Sites with the highest levels of N[O.sub.[3.sup.-]] (SMO, MOU, BOW, and WHI) are all located in areas of the eastern U.S. which have been subject to strong anthropogenic atmospheric N deposition for several decades (e.g., Likens et al. 1977, National Research Council 1986, Hedin et al. 1987). Conversely, CP old-growth forests showed the lowest efflux concentrations of N[O.sub.[3.sup.-]], and are located in an area that is not subject to anthropogenic N inputs. Studies of streams draining various forest types (successional ages not specified) subject to low N deposition in New Zealand similarly found N[O.sub.[3.sup.-]] to be low (N [less than] 20 [[micro]gram]/L) or below detection limits (inferred by us to be [approximately equal to]0.5 [[micro]gram]/L) except in areas of sheep farming (Stenzel and Herrmann 1990). This analysis of CP and other old-growth forests provides strong support for the N-saturation hypothesis (Agren and Bosatta 1988, Aber et. al. 1991) that solution losses of N[O.sub.[3.sup.-]] from forested ecosystems strongly increase as a function of increased atmospheric pollution inputs of N. We hypothesize that old-growth forest ecosystems may be particularly sensitive indicators of ecological effects due to N deposition since, barring any fundamental shift in community composition and growth rates, these ecosystems are inherently poor at retaining added nutrients. Results from second-growth forests are likely to be more variable owing to the ability of aggrading ecosystems to retain N by net plant uptake and net immobilization in soils during humification. Yet, increased N[O.sub.[3.sup.-]] has also been reported in runoff waters from European and North American second-growth forests subject to elevated N loadings (e.g., Henriksen and Brakke 1988, Wright and Haus 1991).
If we consider all major forms of dissolved N in CP streams - that is, organic as well as inorganic - our results indicate a more complex pattern of N loss from temperate forests than suggested by the nutrient retention hypothesis, or by current conceptual and simulation models (e.g., Vitousek and Reiners 1975, Liu et al. 1991, Johnson 1992, Van Migroet et al. 1992). Efforts to understand controls on hydrologic N losses from temperate watershed ecosystems have mainly focused on inorganic forms of N (i.e., [N[O.sub.3].sup.-] and [N[H.sub.4].sup.+]). In contrast, our results identify dissolved organic nitrogen (DON) as the major hydrologic vector of N loss in unpolluted forests. In fact, DON contributed on average 95% of total hydrologic N losses from CP watersheds [ILLUSTRATION FOR FIGURE 5 OMITTED], with little difference between forest types (Table 2). Studies by Sollins et al. (1980) similarly highlighted DON as the major vector of hydrologic N loss from old-growth Douglas-fir forests at the Andrews Experimental Forest. While data are sparse, DON may also comprise a significant ([greater than] 10%) fraction of hydrologic N losses from temperate forests in areas of high N deposition (Qualls et al. 1991; L. O. Hedin, unpublished data). Despite this quantitative importance, little is known about effects of DON losses on long-term budgets or cycling of N in temperate forest ecosystems. Johnson (1992) argued that N retention in soils may be strongly affected by the poorly understood mechanism of non-biological immobilization of N into soil humus. Our results indicate that the subsequent mechanism of organic N dissolution from soil humic fractions is a similarly important, yet poorly known, process of the N cycle in temperate forest soils.
Variations in DON among CP watershed streams were positively correlated with dissolved organic carbon (DOC) [ILLUSTRATION FOR FIGURE 6A OMITTED] and organic anions [ILLUSTRATION FOR FIGURE 6B OMITTED], suggesting that DON losses were associated with leaching of soil humic substances. Organic anions in natural waters are typically dominated by fulvic acids, with lesser contributions from hydrophilic and humic acids (McKnight et al. 1985, 1992). Based on average chemical characteristics, we can provide a preliminary evaluation of the contribution of fulvic acids to losses of DON from CP watersheds. Assuming an average fulvic acid charge density of [approximately equal to] 14 [[micro]mol.sub.c]/mg C (McKnight et al. 1985, 1992) and that our measured organic anion charges (geometric [Mathematical Expression Omitted]; range: 25-194 [[micro]mol.sub.c]/L) are mainly due to fulvic acids, we calculate that the carbon in fulvic acid is present in an average concentration of [approximately equal to] 5.4 mg/L, or 80% of average DOC in CP watershed streams. Such percentage contributions of fulvic acids to DOC are slightly elevated compared to other natural waters ([approximately equal to] 35-67%; McKnight et al. 1985, 1992). Furthermore, based on N contents of 1.6 to 3.0% relative to carbon for natural fulvic acids (McKnight et al. 1985, 1992; D. M. McKnight, personal communication) we calculate fulvic acids to contribute between 86 to 161 [[micro]gram]/L of N to DON, or between 57 to 107% of average measured DON levels (N geometric [Mathematical Expression Omitted]; range: 83-350) in CP streams. These calculations, together with relatively high DOC: DON mass ratios ([approximately equal to] 40-50; Table 2), support the idea that hydrologic DON losses from CP watersheds occur mainly due to the dissolution of highly refractory fulvic acids from soil organic pools.
The strong contrast in patterns of loss between inorganic vs. organic forms of N in CP streams [ILLUSTRATION FOR FIGURE 5 OMITTED] indicates the existence of different mechanisms of ecosystem-level control on N retention. Our results suggest that while biologically available forms of N ([N[H.sub.4].sup.+] and [N[O.sub.3].sup.-]) are retained within CP forest ecosystems, biologically unavailable and hydrologically mobile forms (DON) dominate in streamwater outputs. Thus, ecosystem-level losses of N are not exclusively subject to traditional mechanisms of direct biotic control (i.e., mineralization or biotic uptake), but also to indirect biotic control associated with the long-term accumulation and humification of soil organic N during succession. Losses of DON are likely more directly linked to hydrological parameters, such as variations in water flow paths through soils and dissolution kinetics of humic soil components. Long-term N budgets of these old-growth ecosystems are thus constrained by the continuous "escape" or "leak" of N in the form of biologically unavailable, non-labile DON. While the exact ecological consequences are unclear, such losses may act over time to limit soil N pools as well as internal ecosystem supplies of N from mineralization.
The exceptionally low levels of [N[O.sub.3].sup.-] in CP streams (Table 2) may, in addition to the lack of anthropogenic N deposition, be due to: (1) strong biotic demand for [N[O.sub.3].sup.-], (2) lack of nitrification in CP soils (either due to low [N[H.sub.4].sup.+] supply or to inhibition of nitrifiers), or (3) high rates of denitrification in soils. Measures of soil N levels in forests of different successional ages led Rice and Pancholy (1972) to suggest that [N[H.sub.4].sup.+] dominates over [N[O.sub.3].sup.-] as a N source for vegetation of old-growth forests. While Rice and Pancholy did not measure nitrification directly, they attributed this dominance of [N[H.sub.4].sup.+] over [N[O.sub.3].sup.-] to the inhibition of nitrification in old-growth forests. This hypothesis has been met with skepticism, with subsequent studies documenting that such inhibition is not necessarily a general property of old-growth forest ecosystems (Lamb 1980, Robertson and Vitousek 1981, Bremner and McCarthy 1988).
Our watershed-level results from CP, however, concur with Rice and Pancholy's results to the extent that we found a strong dominance of [N[H.sub.4].sup.+] over [N[O.sub.3].sup.-] in streamwater outputs. Furthermore, a comparison of [N[O.sub.3].sup.-]-N : [N[H.sub.4].sup.+]-N ratios in solution losses from CP and nine additional old-growth forests shows a consistent dominance of [N[H.sub.4].sup.+]-N (ratio [less than]1) in areas with low atmospheric N deposition [ILLUSTRATION FOR FIGURE 7B OMITTED]. In contrast, ratios in waters that drain high N-deposition forests in the eastern U.S. (BOW, MOU, SMO, or WHI) were consistently above 8 [ILLUSTRATION FOR FIGURE 7B OMITTED]. We conclude that high ratios of [N[O.sub.3].sup.-]-N : [N[H.sub.4].sup.+]-N in solution losses from old-growth temperate forests appear linked to areas of high N deposition [ILLUSTRATION FOR FIGURE 7B OMITTED].
While this comparative analysis supports the idea that [N[H.sub.4].sup.+] dominates over [N[O.sub.3].sup.-] in losses from unpolluted old-growth forests, we cannot conclude which mechanism(s) is responsible for the apparent low [N[O.sub.3].sup.-]-N : [N[H.sub.4].sup.+]-N ratios associated with low N deposition. Our preliminary studies of CP soils show a strong nitrification potential during laboratory incubations (C. Perez, L. O. Hedin, and J. J. Armesto, unpublished manuscript) indicating that demands by heterotrophs, vegetation, or denitrifiers may limit levels of [N[O.sub.3].sup.-] in CP watersheds. The presence of saturated soils at depth (gley layer at [approximately equal to] 30-60 cm) in CP watersheds may contribute to the lack of [N[O.sub.3].sup.-] relative to [N[H.sub.4].sup.+], either by inhibiting nitrification or by stimulating denitrification. However, preliminary results from 23 low-tension lysimeters showed that [N[H.sub.4].sup.+]-N consistently dominated over only low levels of [N[O.sub.3].sup.-]-N (median [N[O.sub.3].sup.-]-N : [N[H.sub.4].sup.+]-N ratio = 0.04) in shallow as well as deep lysimeters (15 cm. vs. [approximately equal to] 40 cm depth, n = 11 vs. 12 lysimeters, respectively). That streamwater [N[O.sub.3].sup.-] remained low despite strong variations in soil water flow paths during wet vs. dry sampling periods further indicates a lack of relation between soil depth, denitrification, and levels of [N[O.sub.3].sup.-] at CP. We conclude that while low efflux concentrations of [N[O.sub.3].sup.-] from old-growth forests appear linked to areas with low N deposition [ILLUSTRATION FOR FIGURE 7A OMITTED], it is not clear why [N[H.sub.4].sup.+]-N consistently dominates over [N[O.sub.3].sup.-]-N in solution losses from the same forests [ILLUSTRATION FOR FIGURE 7B OMITTED].
Biogeochemistry of unpolluted old-growth forests
It is difficult to ascertain the degree to which CP forest ecosystems are representative of other old-growth temperate forests, such as those studied in the Northern Hemisphere. For example, soils at CP are highly weathered and supply less base cations than many younger, glaciated temperate soils (Hedin and Hetherington 1995). Yet the acidic nature of CP soils and drainage waters is similar to oligotrophic, poorly buffered, temperate ecosystems found in many regions of Europe and North America. Furthermore, plant species of CP forests have different biogeographical and evolutionary histories than species of Northern Hemisphere temperate forests (Axelrod et al. 1991). It is not known whether such historical floristic differences might translate into ecologically relevant differences in biogeochemical cycles.
While some aspects of CP forests may differ from other temperate forests, our study provides the opportunity to evaluate general theories about nutrient cycling in unpolluted and floristically stable old-growth forests. If these theories are of a truly general nature, we would expect that they apply to CP forests as well as Northern Hemisphere forests. For major elements, our results support the current view that old-growth forest ecosystems should exhibit minimal or no net biotic retention of element inputs. In contrast, patterns of N loss were more complex than current theories suggest, because old-growth CP forests can alternatively be classified as "non-leaky" or "leaky" depending on whether inorganic or organic forms of N are considered. Yet patterns of [N[O.sub.3].sup.-] loss at CP and other old-growth forests [ILLUSTRATION FOR FIGURE 7A OMITTED] supported predictions from the N saturation hypothesis of Abet et al. (1989, 1991). We suggest that old-growth forests may be particularly appropriate sites to evaluate the link between atmospheric deposition and element outputs, since such mature ecosystems are likely to be less retentive due to low net internal nutrient demands.
Assuming our results from CP are representative for unpolluted old-growth temperate forests, the observed patterns of N loss may have significant conceptual importance also at levels of communities and populations. For example, the exceedingly low levels of [N[O.sub.3].sup.-] and the low [N[O.sub.3].sup.-]-N : [N[H.sub.4].sup.+]-N ratios at CP and other low N-deposition old-growth forests [ILLUSTRATION FOR FIGURE 7 OMITTED] raise the fundamental question of to what extent [N[O.sub.3].sup.-] was available as a significant source of N to biota before the advent of regional-scale anthropogenic N pollution. While nitrification can be an important internal source of [N[O.sub.3].sup.-], low hydrologic losses of [N[O.sub.3].sup.-] from unpolluted forests indicate that biotic demands for [N[O.sub.3].sup.-] are high relative to supplies. Tamm (1991) recently argued that increased anthropogenic N deposition has caused an increased selection for nitrophilous plant species in natural ecosystems. Our analyses support Tamm's contention, and furthermore suggest that the relative abundance of [N[O.sub.3].sup.-] vs. [N[H.sub.4].sup.+] may have changed as a function of N loading [ILLUSTRATION FOR FIGURE 7B OMITTED]. In theory, such change would have important consequences for the relative competitive abilities of species of plants and microorganisms that favor obtaining N as [N[O.sub.3].sup.-] rather than as [N[H.sub.4].sup.+] (cf. Tilman 1984, 1988, Tilman and Wedin 1991). Current studies of plant and microbial communities in regions subject to anthropogenic N deposition may therefore not reveal ecological interactions that are historically characteristic of these communities, but evolutionarily novel interactions that have been biased by regional-scale human influences on the N cycle.
This is contribution number 790 to the program at the W. K. Kellogg Biological Station, and the Cordillera de Piuchue Ecosystem Study (CPES). Financial support was provided by the A. W. Mellon Foundation and by Fondecyt (Chile) grants 860-88 and 92-1135. We thank M. Brown, C. Perez, D. Soto, M. Velbel, J. von Fischer, and B. Kennedy. R. Serrano, C. Smith, and S. Agila assisted in the field. We appreciate the logistic assistance of Chiloe Park personnel. We thank D. Johnson for advice and D. McKnight for her valuable review, and for advice about the role of fulvic acids.
Abet, J. D., and Melillo, J.M. 1991. Terrestrial ecosystems. Saunders, Philadelphia, Pennsylvania, USA.
Aber, J. D., J. M. Melillo, K. J. Nadelhoffer, J. Pastor, and R. Boone. 1991. Factors controlling nitrogen cycling and nitrogen saturation in northern temperate forest ecosystems. Ecological Applications 1:303-315.
Abet, J. D., K. J. Naedelhoffer, P. Steudler, and J. M. Melillo. 1989. Nitrogen saturation in northern forest ecosystems. BioScience 39:378-386.
Agren, G. I., and E. Bosatta. 1988. Nitrogen saturation of the terrestrial ecosystem, Environmental Pollution 54:185-197.
Alpkem. 1992a. Silica in seawater. Document No. 000671. Alpkem, Wilsonville, Oregon, USA.
-----. 1992b. Ammonia in seawater. Document No. 000674. Alpkem, Wilsonville, Oregon, USA.
-----. 1992c. Nitrate + nitrite nitrogen. Document No. 000589. Alpkem, Wilsonville, Oregon, USA.
Armesto, J. J., J. C. Aravena, C. Perez, C. Smith-Ramirez, M. Cortes, and L. Hedin. 1994. Conifer forests of the Chilean coastal range: history and ecology. In N. J. Enright and S. Hill, editors. Ecology of the Southern Conifers. Melbourne University Press, Melbourne, Victoria, Australia, in press.
Armesto, J. J., and E. R. Fuentes. 1988. Tree species regeneration in a mid-elevation, temperate rain forest in Isla de Chiloe, Chile. Vegetatio 74:151-159.
Axelrod, D. I., M. T. K. Arroyo, and P. H. Raven. 1991. Historical development of temperate vegetation in the Americas. Revista Chilena de Historia Natural 64:413-446.
Binkley, D., and D. Richter. 1987. Nutrient cycles and budgets of forest ecosystems. Advances in Ecological Research 16:1-51.
Bormann, F. H., and G. E. Likens. 1979. Pattern and process in a forested ecosystem. Springer-Verlag, New York, New York, USA.
Bormann, F. H., G. E. Likens, and J. M. Melillo. 1977. Nitrogen budget for an aggrading northern hardwood forest ecosystem. Science 196:981-983.
Brakke, D. F., D. H. Landers, and J. M. Eilers. 1988. Chemical and physical characteristics of lakes in the Northeastern United States. Environmental Science and Technology 22:155-163.
Bremner, J. M., and G. W. McCarthy. 1988. Effect of terpenoids on nitrification in soil. Soil Science Society of America Journal 52:1630-1633.
Burke, E. M., F. X. Suarez, D.C. Hillman, and E. M. Heithmar. 1989. The evaluation and comparison of ion chromatography, segmented flow analysis and flow injection analysis for the determination of nitrate in natural surface waters. Water Research 23:519-521.
Coleman, D. C. 1989. Ecology, agroecosystems, and sustainable agriculture. Ecology 70:1590.
Davis, M. G. 1976. Pleistocene biogeography of temperate deciduous forests. Geoscience and Man 13:13-26.
D'Elia, C. F., P. A. Steudler, and N. Corwin. 1977. Determination of total nitrogen in aqueous samples using persulfate digestion. Limnology and Oceanography 22:760-764.
Feller, M. C. 1987. The influence of acid precipitation on stream chemistry in a small forested basin in southwestern British Columbia. Pages 33-47 in R. H. Swanson, P. Y. Bernier, and P. D. Woodard, editors. Forest hydrology and watershed management. International Association for Hydrological Sciences Press, Vancouver, British Columbia, Canada.
Findlay, S. F., and C. G. Jones. 1990. Exposure of cottonwood plants to ozone alters subsequent leaf decomposition. Oecologia 82:248-250.
Friedland, A. J., E. K. Miller, J. J. Battles, and J. E Thorne. 1991. Nitrogen deposition, distribution and cycling in a subalpine spruce-fir forest in the Adirondacks, New York, USA. Biogeochemistry 14:31-55.
Galloway, J. N., H, Levy II, and P. S. Kasibhatala. 1994. Year 2020: consequences of population growth and development on deposition of oxidized nitrogen. Ambio 23:120-123.
Galloway, J. N., G. E. Likens, W. C. Keene, J. Gonzales, and C. Yancz. 1995. Composition of precipitation in a remote Southern Hemisphere location: Torres del Paine National Park, Chile. In W. L. Franklin, W. E. Johnson, and A. Iriarte, editors. A patagonia gem: the ecology and natural history of a world biosphere reserve. Santiago, Chile, in press.
Galloway, J. N., G. E. Likens, W. C. Keene, and J. M. Miller. 1982. The composition of precipitation in remote areas of the world. Journal of Geophysical Research 87:8711-8786.
Gibbs, R. J. 1970. Mechanisms controlling world water chemistry. Science 170:1088-1090.
Gorham, E. 1961. Factors influencing supply of major ions to inland waters, with special reference to the atmosphere. Geological Society of America Bulletin 72:795-840.
Gorham, E., P. M. Vitousek, and W. A. Reiners. 1979. The regulation of chemical budgets over the course of terrestrial ecosystem sucession. Annual Review of Ecology and Systematics 10:53-84.
Hedin, L. O., and H. Campos. 1991. Importance of small streams in understanding and comparing watershed ecosystem processes. Revista Chilena Historia Natural 64:583-596.
Hedin, L. O., L. Granat, G. E. Likens, T. A. Buishand, J. M. Galloway, T. J. Butler, and H. Rodhe. 1994. Steep declines in atmospheric base cations in regions of Europe and North America. Nature 367:351-354.
Hedin, L. O., L. Granat, G. E. Likens, and H. Rodhe. 1990a. Strong similarities in seasonal concentration ratios of S[O.sub.4], N[O.sub.3] and N[H.sub.4] in precipitation between Sweden and the northeastern U.S. Tellus 42B:454-462.
Hedin, L. O., and E. Hetherington. 1995. Atmospheric and geologic constraints on the biogeochemistry of North and South American temperate rain forests. In R. G. Lawford, E. Fuentes, and H. Mooney, editors. High latitude forest and riverine systems of the west coasts of the Americas: environment, ecology and human use. Springer-Verlag, New York, New York, USA, in press.
Hedin, L. O., G. E. Likens, and F. H. Bormann. 1987. Decrease in precipitation acidity resulting from decreased concentration. Nature 325:244-246.
Hedin, L. O., G. E. Likens, K. M. Postek, and C. T. Driscoll. 1990b. A field experiment to test whether organic acids buffer acid deposition. Nature 345:798-800.
Henderson, G. S. 1975. Element retention and conservation. BioScience 25:770.
Henriksen, A., and D. F. Brakke. 1988. Increasing contributions of nitrogen to the acidity of surface waters in Norway. Water, Air, and Soil Pollution 42:183-201.
Holdgate, M. W. 1961. Vegetation and soils in the south Chilean islands. Journal of Ecology 49:559-580.
Johnson, D. W. 1992. Nitrogen retention in forest soils. Journal of Environmental Quality 21:1-12.
Johnson, D. W., and S. E. Lindberg. 1992. Atmospheric deposition and forest nutrient cycling. Ecological Studies 91. Springer-Verlag, New York, New York, USA.
Johnson, D. W., H. Van Miegroet, S. E. Lindberg, and D. E. Todd. 1991. Nutrient cycling in red spruce forests of the Great Smoky Mountains. Canadian Journal of Forest Research 21:769-787.
Keene, W. C, A. A. P. Pszenny, J. N. Galloway, and M. E. Hawley. 1986. Sea-salt corrections and interpretation of constituent ratios in marine precipitation. Journal of Geophysical Research 91:6647-6658.
Lamb, D. 1980. Soil nitrogen mineralisation in a secondary rain-forest succession. Oecologia 47:257-263.
Lara, A., and R. Villalba. 1993. A 3620-year temperature record from Fitzroya cupressoides tree rings in southern South America. Science 260:1104-1106.
Larson, A. G. 1979. Origin of the chemical composition of undisturbed forested streams: Western Olympic peninsula, Washington State. Dissertation. University of Washington, Seattle, Washington, USA.
Last, F. T., and R. Watling, editors. 1991. Acidic deposition: its nature and impacts. The Royal Society of Edinburgh, Edinburgh, Scotland.
Likens, G. E., F. H. Bormann, R. S. Pierce, J. S. Eaton, and N. M. Johnson. 1977. Biogeochemistry of a forested ecosystem. Springer-Verlag, New York, New York, USA.
Likens, G. E., W. C. Keene, J. M. Miller, and J. N. Galloway. 1987. Chemistry of precipitation from a remote, terrestrial site in Australia. Journal of Geophysical Research 92:299-313.
Liu, S., R. Munson, D. Johnson, S. Gherini, K. Summers, R. Hudson, K. Wilkinson, and L. Pitelka. 1991. Application of a nutrient cycling model (NuCM) to a northern mixed hardwood and a southern coniferous forest. Pages 173-184 in M. R. Kaufmann and J. J. Landsberg, editors. Advancing toward closed models of forest ecosystems. Heron, Victoria, British Columbia, Canada.
Lovett, G. M., W. A. Reiners, and R. K. Olsen. 1982. Cloud droplet deposition in subalpine balsam-fir forests: hydrological and chemical inputs. Science 218:1303-1304.
Markgraf, V., J. R. Dodson, A. P. Kernshaw, M. S. McGlone, and N. Nicholls. 1992. Evolution of late Pleistocene and Holocene climates in the circum-South Pacific land areas. Climate Dynamics 6:193-211.
Martin, C. W. 1979. Precipitation and streamwater chemistry in an undisturbed forested watershed in New Hampshire. Ecology 60:36-42.
McDonnell, M. J., and S. T. A. Pickett. 1990. Ecosystem structure and function along urban-rural gradients. Ecology 71:1232-1237.
McDowell, W. H., J. J. Cole, and C. T. Driscoll. 1987. Simplified version of the ampoule-persulfate method for determination of dissolved organic carbon. Canadian Journal of Fisheries and Aquatic Science 44:214-218.
McKnight, D. M., K. E. Bencala, G. W. Zellweger, G. R. Aiken, G. L. Feder, and K. A. Thorn. 1992. Sorption of dissolved organic carbon by hydrous aluminum and iron oxides occuring at the confluence of Deer Creek with the Snake River, Summit county, Colorado. Environmental Science and Technology 26:1388-1396.
McKnight, D. M., E. M. Thurman, and R. L. Wershaw. 1985. Biogeochemistry of aquatic humic substances in Thoreau's bog, Concord, Massachussetts. Ecology 66:1339-1352.
Monk, C. D., and F. P. J. Day. 1988. Biomass, primary production, and selected nutrient budgets for an undisturbed watershed. Pages 151-159 in W. T. Swantk and D. A. J. Crossley, editors. Forest hydrology and ecology at Coweeta. Springer-Verlag, New York, New York, USA.
National Research Council. 1986. Acid deposition: long-term trends. National Academy Press, Washington, D.C., USA.
Newman, M. C., P. M. Dixon, B. B. Looney, and J. E. Pinder. 1989. Estimating mean and variance for environmental samples with below detection limit observations. Water Resources Bulletin 25:905-916.
Oden, S. 1968. The acidification of air and precipitation and its consequences on the natural environment. Bulletin number 1. Swedish National Science Research Council, Ecology Committee, Stockholm, Sweden.
Odum, E. P. 1969. The strategy of ecosystem development. Science 164:262-270.
Peet, R. K. 1992. Community structure and ecosystem function. Pages 103-151 in D. L. Glenn-Lewin, R. K. Peet, and T. T. Veblen, editors. Plant succession: theory and prediction. Chapman & Hall, London, England.
Perez, C. 1992. Los bosques de "Olivillo" (Aetoxicon punctatum) de la cordillera de la costa de Chile: Interaccion clima-suelo-vegetation. Dissertation. Universidad de Chile, Santiago, Chile.
Perez, C., J. J. Armesto, and B. Ruthsatz. 1991. Descomposicion de hojas, biomasa de raices y caracteristicas de los suelos en bosques mixtos de coniferas y especies laurifolias en el Parque Nacional Chiloe, Chile. Revista Chilena de Historia Natural 64:479-490.
Perez, C., L. O. Hedin, and J. J. Armesto. 1994. Patterns of nitrogen mineralization in unpolluted, old-growth temperate forests of southern Chile. Abstract for 1994 annual meeting of the Ecological Society of America. Knoxville, Tennessee, USA.
Qualls, R. G., B. L. Haines, and W. T. Swank. 1991. Fluxes of dissolved organic nutrients and humic substances in a deciduous forest. Ecology 72:254-266.
Rice, E. L., and S. K. Pancholy. 1972. Inhibition of nitrification by climax ecosystems. American Journal of Botany 59:1033-1040.
Robertson, G. P. 1982. Nitrification in forested ecosystems. Philosophical Transactions of the Royal Society London B 296:445-457.
----- 1986. Nitrogen: regional contributions to the global cycle. Environment 28:16-21.
Robertson, G. P., and P. M. Vitousek. 1981. Nitrification potentials in primary and secondary succession. Ecology 62:376-386.
Ruthsatz, B., and C. Villagran. 1991. Moorland vegetation and soil nutrients in Chiloe island. Revista Chilena de Historia Natural 64:461-478.
Schmidt, E. L. 1982. Nitrification in soil. Pages 253-288 in F. J. Stevenson, editor. Nitrogen in agricultural soils. Soil Science Society of America, Madison, Wisconsin, USA.
Schulze, E.-D., O. L. Lange, and R. Oren, editors. 1989. Forest decline and air pollution. Springer-Verlag, Berlin, Germany.
Sollins, P., C. C. Grier, F. M. McCorison, K. Cromack, R. Fogel, and R. L. Fredriksen. 1980. The internal element cycles of an old-growth Douglas-fir ecosystem in western Oregon. Ecological Monographs 50:261-285.
Sollins, P., and F. M. McCorison. 1981. Nitrogen and carbon solution chemistry of an old growth coniferous forest watershed before and after cutting. Water Resources Research 17:1409-1418.
Stams, A. J. M., H. W. G. Bootink, L. J. Lutke-Schipholt, B. Beemsterboer, J. R. W. Woittiez, and N. Van Breemen. 1991. A field study of the fate of 15N-ammonium to demonstrate nitrification of atmospheric ammonium in an acid forest soil. Biogeochemistry 13:241-255.
Stauffer, R. E. 1990. Granite weathering and the sensistivity of alpine lakes to acid deposition. Limnology and Oceanography 35:1112-1134.
Stednick, J. D. 1981. Precipitation and streamwater chemistry in an undisturbed watershed in southeast Alaska. Research Paper PNW-291. USDA Forest Service Pacific Northwest Forest and Range Experiment Station, Portland, Oregon, USA.
Stenzel, A., and R. Herrmann. 1990. Comparing the effects of acidic deposition on the chemistry of small streams in the South Island of New Zealand with those in the Fichtelgebirge, F.R.G. Catena 17:69-83.
Stone, E. L., and R. Kszystniak. 1977. Conservation of potassium in the Pinus resionosa ecosystem. Science 198:192-194.
Taljaard, J. J. 1972. Synoptic meteorology of the Southern Hemisphere. Meteorological Monographs 13:139-213.
Tamm, C. O. 1991. Nitrogen in terrestrial ecosystems. Springer-Verlag, Berlin, Germany.
Taylor, A. B., and M. A. Velbel. 1991. Geochemical mass balances and weathering rates in forested watersheds of the southern Blue Ridge. II. Effects of botanical uptake terms. Geoderma 51:29-50.
Tilman, D. 1988. Plant strategies and the dynamics and structure of plant communities. Princetin University Press, Princeton, New Jersey, USA.
Tilman, D., and D. Wedin. 1991. Plant traits and resource reduction for five grasses growing on a nitrogen gradient. Ecology 72:685-700.
Tilman, G. D. 1984. Plant dominance along an experimental nutrient gradient. Ecology 65:1445-1453.
Van Miegroet, H., D. W. Cole, and N. W. Foster. 1992. Nitrogen distribution and cycling. Pages 178-196 in D. W. Johnson and S. E. Lindberg, editors. Atmospheric deposition and forest nutrient cycling. Springer-Verlag, New York, New York, USA.
Veblen, T. T., C. Donoso, F. M. Schlegel, and R. Escobar. 1981. Forest dynamics in South-central Chile. Journal of Biogeography 8:211-247.
Veit, H., and K. Garlef. 1995. Evolucion del paisaje cuatenario y desarrollo de suelos en Chile sur-central. In J. J. Armesto, M. K. Arroyo, and C. Villagran, editors. Ecologia del bosque templado de Chile. Editorial Universitaira, Santiago, Chile, in press.
Villagran, C. 1988. Expansion of magellanic moorland during the late Pleistocene: palynological evidence from northern Isla de Chiloe, Chile. Quaternary Research 30:304-314.
-----. 1990. Glacial climates and their effects on the history of the vegetation of Chile: a synthesis based on palynological evidence from Isla de Chiloe. Review of Palaeobotany and Palynology 65:17-24.
-----. 1991. Historia de los bosques templados del sur de Chile durnate el Tardiglacial y Postglacial. Revista Chilena de Historia Natural 64:447-460.
Vitousek, P. M. 1977. The regulation of element concentrations in mountain streams in the northeastern United States. Ecological Monographs 47:65-87.
----- 1982. Nutrient cycling and nutrient use efficiency. American Naturalist 119:553-572.
----- 1990. Biological invasions and ecosystem processes: towards an integration of population biology and ecosystem studies. Oikos 57:7-13.
Vitousek, P. M., P. A. Matson, and K. Van Cleve. 1989. Nitrogen availability and nitrification during sucession: primary, secondary and old-field seres. Plant and Soil 115:229-239.
Vitousek, P.M., and W. A. Reiners. 1975. Ecosystem succession and nutrient retention: a hypothesis. BioScience 25:376-381.
Vong, R. J., T. V. Larson, D. S. Covert, and A. P. Waggoner. 1985. Measurement and modeling of western Washington precipitation chemistry. Water, Air, and Soil Pollution 26:71-84.
Warneck, P. 1988. Chemistry of the natural atmosphere. Academic Press, San Diego, California, USA.
Watters, W. A., and C. A. Fleming. 1972. Contributions to the geology and palaeontology of Chile Island, Southern Chile. Philosophical Transactions of the Royal Society London 263(B):370-408.
Woodmansee, R. G. 1978. Additions and losses of nitrogen in grassland ecosystems. BioScience 28:448-453.
Wright, R. E, and M. Haus. 1991. Reversibility of acidification: soils and surface waters. Proceedings of the Royal Society of Edinburgh 97B:169-191.
Zeman, L. J., and O. Slaymaker. 1978. Mass balance model for calculation of ionic input loads in atmospheric fallout and discharge from a mountainus basin. Hydrological Sciences Bulletin 23:103-116.
|Printer friendly Cite/link Email Feedback|
|Author:||Hedin, Lars O.; Armesto, Juan J.; Johnson, Arthur H.|
|Date:||Mar 1, 1995|
|Previous Article:||Experimental analysis of intermediate disturbance and initial floristic composition: decoupling cause and effect.|
|Next Article:||Water losses in the Patagonian steppe: a modelling approach.|