The depletion of calcium in forest ecosystems of the northeastern USA1,2,3 is thought to be a consequence of acidic deposition and to be at present restricting the recovery of forest and aquatic systems4,5,6,7 now that acidic deposition itself is declining. This depletion of calcium has been inferred from studies1,2,3 showing that sources of calcium in forest ecosystems—namely, atmospheric deposition and mineral weathering of silicate rocks such as plagioclase, a calcium-sodium silicate—do not match calcium outputs observed in forest streams. It is therefore thought that calcium is being lost from exchangeable and organically bound calcium in forest soils. Here we investigate the sources of calcium in the Hubbard Brook experimental forest, through analysis of calcium and strontium abundances and strontium isotope ratios within various soil, vegetation and hydrological pools. We show that the dissolution of apatite (calcium phosphate) represents a source of calcium that is comparable in size to known inputs from atmospheric sources and silicate weathering. Moreover, apatite-derived calcium was utilized largely by ectomycorrhizal tree species, suggesting that mycorrhizae may weather apatite and absorb the released ions directly, without the ions entering the exchangeable soil pool. Therefore, it seems that apatite weathering can compensate for some of the calcium lost from base-poor ecosystems, and should be considered when estimating soil acidification impacts and calcium cycling.
We studied the biogeochemistry of calcium in a mixed conifer/hardwood forest within a 12-ha watershed (W-1) at the Hubbard Brook experimental forest (HBEF), New Hampshire, USA2,7,8. The main calcium-bearing minerals present in soil parent material are plagioclase, K-feldspar, hornblende, pyroxene and apatite. Sequential digestions of soil samples (see Methods) were used to give information on the weathering sources of Ca (Table 1). We first examined the molar ratios of P/Ca versus Na/Ca in C-horizon digestions in comparison with the average composition of the main Ca-bearing minerals (Fig. 1). ‘Digest 1’ consisted mainly of dissolved apatite, consistent with the much faster (>103 times) dissolution rate of apatite compared to hornblende, plagioclase and K-feldspar (at pH 2–5)9,10,11. ‘Digest 2’ consisted mainly of dissolved hornblende and pyroxene (Fig. 1), and ‘digest 3’ was mainly dissolved plagioclase and K-feldspar. About 12% of the Ca in the soil parent material occurs in apatite, another 12% is in hornblende and pyroxene, and 76% is in feldspars. The Ca/Sr ratio of digest 1 was very different to that of digests 2 and 3 (Table 1), owing to differences in Sr substitution for Ca in silicate versus phosphate minerals—providing a tracer of apatite-derived Ca. The Ca/Sr ratio of digest 1 for each soil horizon suggests that apatite was depleted from the Oa, E, Bh and Bs1 horizons, but is present in the Bs2 and C horizons (Fig. 2). If we assume that all of the Ca in the C-horizon digest 1 is derived from apatite (Fig. 1), then a Ca/Sr mixing calculation12,13 shows that 86% of the Ca released from the Bs2-horizon soils by digest 1 is derived from apatite.
We used Ca/Sr and 87Sr/86Sr ratios to ascertain the original sources of Ca in ecosystem pools13,14,15,16,17. The 87Sr/86Sr ratio is not changed by ion exchange, plant uptake or transport13,14,15,16,17,18,19, and the Ca/Sr ratio has generally been observed to be only slightly changed by these processes (a factor of 1.0–1.6)14,18,19. The C-horizon digest-1 fractions had average Ca/Sr and 87Sr/86Sr ratios (respectively 2,580 and 0.72164) that were distinct from the values for the digest-2 (279 and 0.73669) and digest-3 fractions (170 and 0.73352; Fig. 3). This pattern is consistent with our interpretation that digest 1 is dominated by dissolution of apatite, which has characteristically high Ca/Sr and low 87Sr/86Sr (refs 20, 21). In comparison, atmospheric deposition near Hubbard Brook has an average Ca/Sr of 193 and 87Sr/86Sr of 0.71060 (refs 15, 16). The ‘exchangeable’ fractions (see Methods) of soil Oa and Bs1 horizons had Ca/Sr ratios similar to atmospheric and soil-digest values, but had 87Sr/86Sr close to the digest-1 values and intermediate between atmospheric and digest-2 and digest-3 values. Thus the soil cation exchange pool is dominated by a mixture of Ca and Sr from silicate weathering and atmospheric deposition.
In Fig. 3, the foliage samples plot along a line connecting the soil cation exchange pool and apatite, with different species occupying distinct fields. The Ca/Sr ratios were relatively low for sugar maple, intermediate for yellow birch and American beech, and high for balsam fir and red spruce. In contrast, foliar 87Sr/86Sr was nearly identical for all tree species, and was consistent with both the soil cation exchange pool and apatite weathering values (Fig. 3). Wood ferns, which are shallowly rooted and reflect forest floor values, had Ca/Sr ratios in the same range as maple and birch foliage (Fig. 3). Throughfall from the dominantly spruce/fir areas of the mixed conifer–hardwood canopy is a mixture of atmospheric deposition and spruce/fir foliage, whereas throughfall from the dominantly hardwood areas shows a stronger hardwood foliage influence. The range of soil water Ca/Sr and 87Sr/86Sr ratios for nine lysimeters sampled during each season of 1998, and the average streamwater Ca/Sr and 87Sr/86Sr for weekly samples collected in 1998, are also shown in Fig. 3.
To identify the sources of Ca in various ecosystem pools, we defined the Ca/Sr and 87Sr/86Sr ratios of three ecosystem endmembers. For the silicate weathering endmember we used the average value of the soil C-horizon digest 2, because it is the most readily weathered soil silicate fraction. For the apatite weathering endmember, we used the average of the C-horizon digest-1 fractions. The average atmospheric deposition endmember has been measured directly15,16. Apatite weathering does not contribute appreciably to exchangeable Ca in the Bs1 horizon, probably because it has been depleted in the upper horizons of the soil profile (Fig. 2). A mixing calculation for Ca/Sr and 87Sr/86Sr of two endmembers12,13 shows that the Bs1 exchange pool contained 55% Ca from silicate weathering and 45% Ca from atmospheric sources (Fig. 3), similar to findings of previous studies15,16.
The composition of the Oa exchange pool, stream water, and vegetation can be explained by a mixture of the Bs1 exchange pool and apatite (Fig. 3). Mixing calculations provide an estimate of the relative quantity of Ca derived from the Bs1 exchange pool compared to the apatite weathering pool, and show that on average ∼95% of the foliar Ca in spruce/fir foliage is derived from apatite, whereas ∼80% of beech, ∼75% of birch and ∼60% of maple and fern Ca is derived from apatite (Figs 2, 3). Also, ∼35% of the stream Ca loss from the watershed is estimated to be derived from apatite weathering and ∼20% of the exchangeable Ca in the Oa horizon is derived from apatite. Trees at HBEF appear to be accessing Ca and Sr from apatite in the Bs2 horizon, and partially bypassing the soil cation exchange complex. Studies in base-poor northern European coniferous forests have suggested that ectomycorrhizal mycelia are able to penetrate micropores in silicate minerals and take up Ca and P directly from apatite inclusions, while remaining isolated from the soil solution22. This process has only been observed in conifer forests22,23 and pot experiments23,24, but may explain the high abundance of apatite-derived Ca in the foliage of the mixed conifer/hardwood forest at HBEF. The shallower soils on which conifers tend to grow may facilitate the higher proportion of apatite-Ca. Moreover, ectomycorrhizal fungi associated with the roots of conifers may be more effective at weathering apatite. In contrast, maple—the only major endomycorrhizal tree in the forest25—is least able to access the Ca-apatite and is also the most sensitive species to Ca depletion in forest soils of the northeastern USA6,26.
An alternative explanation for the higher Ca/Sr ratios in foliage compared to those in the soil exchange complex would require that trees preferentially take up Ca over Sr by a factor of 2.3 for maple, 2.9 for birch, 3.7 for beech and 8.5 for spruce/fir. We consider this unlikely for several reasons, but acknowledge that additional research on plant uptake of Ca and Sr is needed. First, to maintain mass balance, preferential uptake of Ca over Sr would leave the soil exchange complex depleted in Ca relative to Sr compared to silicate minerals, and this is not observed (Fig. 3). Second, plants grown in nutrient solutions display only minimal preferential uptake of Ca over Sr (by a factor of 1 to 1.4)18. A study of trees growing on carbonate bedrock suggested that spruce may assimilate Sr preferentially to Ca, whereas beech and maple do not display preferential uptake27. Other studies of the uptake of Ca and Sr by trees growing on soils derived from silicate minerals show higher Ca/Sr of foliage compared to the exchange complex by a factor of 4.7 for beech27 and 7.9–8.4 for spruce16,27. We suggest that these values may reflect the presence or absence of apatite in the mineral soils, combined with rooting depth and mycorrhizal associations, rather than species-specific differences in plant uptake. Nevertheless, we consider the proportions of apatite-Ca in foliage given above to be maximum values, as some portion of the enhanced Ca/Sr may be due to preferential uptake of Ca over Sr.
Despite the importance of apatite-derived Ca to foliage, the Oa exchange complex contains only ∼20% Ca from apatite. To shift the Ca/Sr ratio of the Bs1 exchange complex to the value of the Oa exchange complex, as much as 40% of the Ca would need to be supplied from maple litter or as little as 22% from spruce/fir litter. Most of the Ca in the Oa exchange complex (60–78%) is derived from a combination of silicate weathering and atmospheric deposition, in a 55:45 proportion. Ferns have Ca/Sr ratios consistent with maple foliage, suggesting that maple trees may access Ca from the forest floor without it entering an exchangeable form in the Oa horizon, as has been documented for P (ref. 28). In contrast, spruce and fir trees appear to extract Ca from apatite deep in the soil in addition to utilizing the forest floor pool.
We estimate that Ca in stream water is derived as follows: ∼30% from the atmosphere, ∼35% from silicate minerals, and ∼35% from apatite. Previous research at HBEF1,2,3,7,29 suggested, on the basis of Ca/Na ratios, that silicate weathering is insufficient to account for Ca losses in stream water, and that Ca is being depleted from exchangeable and organically bound pools in forest soils of the northeastern USA. Our data suggest that apatite weathering, possibly by the direct action of root mycorrhizal systems, may account for a significant portion of this Ca, which has no accompanying Na release. However, this apatite pool itself has largely been depleted from upper soil horizons. Based on the average mass of the Bs2 horizon per hectare (ref. 30), and the amount of Ca released by the digest 1 of each soil horizon, we estimate the apatite Bs2 Ca reservoir to be 14 kmol ha-1. This pool is larger than the entire exchangeable Ca pool of the soil (5 kmol ha-1; ref. 2), and the forest floor Ca pool (8 kmol ha-1; ref. 2), but is less than the pool of Ca in above-ground biomass (15 kmol ha-1; ref. 2). Therefore, depletion of Ca from base-poor soils remains a considerable concern, but assessments and models of Ca depletion need to account for this important Ca source.
Soil pits were excavated in 1997 at 47 sites in W-1 and sampled by horizon (Oa, E, Bh, Bs1, Bs2, C). Composite samples of the <2-mm size fraction were made for each horizon for six zones within W-1. Optical microscopy and electron microprobe analysis of soil parent material was used to determine the main Ca-bearing minerals and their compositions. Soils were subjected to a four-step sequential digestion on 0.5-g samples. Solutions were removed by centrifugation after each treatment, and are referred to in the main text as follows: ‘exchangeable’, 5 ml of 1 M NH4Cl (pH = 7) for 20 h; ‘digest 1’, 5 ml of 1 M HNO3 at room temperature for 20 h; ‘digest 2’, 5 ml of 1 M HNO3 at 200 °C for 30 min (by microwave at ∼20 MPa); ‘digest 3’, 10 ml of concentrated HF and HNO3 at 210 °C for 30 min (by microwave at ∼20 MPa). Composite samples of understory foliage (UF) and overstory foliage (OF) were collected in 1999 from trees encountered along transects through each of four elevation zones. Foliage sampling included red spruce (Picea rubens, 2 UF), balsam fir (Abies balsamea, 2 UF), American beech (Fagus grandifolia, 3 UF and 2 OF), yellow birch (Betula alleghaniensis, 3 OF), sugar maple (Acer saccharum, 3 UF and 3 OF) and wood fern (Dryopteris spinulosa, 4 UF). Throughfall was collected during a single storm event on 29 July 1999 at 13 locations in the watershed; soil water was collected using zero tension lysimeters during April, June, October and December of 1998 from the Oa, Bh and Bs1 horizons at three locations in the watershed; and weekly stream water samples from the base of W-1 were analysed for the 1998 calendar year. Foliage samples (0.5 g) were ground, homogenized and digested in concentrated HNO3 at 200 °C for 30 min (by microwave at ∼25 MPa). Leaches, digests and waters were analysed for Ca, K, Mg, Na, Si, P and Sr (±2%) and 87Sr/86Sr (±0.002%) following established methods12,14.
We thank S. Bailey, D. Buso, S. Hamburg, J. Hogan and R. Yanai for discussions, and H. Wallander and L. Derry for comments and suggestions. This research was supported by the NSF, and carried out at the Hubbard Brook experimental forest, which is operated and maintained by the Northeastern Forest Experiment Station, US Department of Agriculture.