Nitrogen translocation in Sphagnum mosses: effects of atmospheric nitrogen deposition

Authors

  • Allison R. Aldous

    Corresponding author
    1. Current address: Cornell University, Department of Natural Resources, Ithaca, New York 14853, USA; The Nature Conservancy, 821 SE 14th Ave, Portland, Oregon 97214, USA
    Search for more papers by this author

Author for correspondence: Allison R. Aldous Tel: +1 503 2300707 ext. 342 Fax: +1 503 2309639 Email: aaldous@tnc.org

Summary

  • • Here, the hypothesis was tested that nitrogen (N) translocation from older to younger parts of Sphagnum decreases as N inputs from atmospheric deposition increase.
  • • Nitrogen translocation in Sphagnum mosses was compared in bogs with contrasting atmospheric N deposition (Adirondack – relatively high N deposition; Maine – relatively low) and by following the movement of a 15NH415NO3 tracer applied to plots of Sphagnum capillifolium over 2 yr.
  • • Annual N translocation ranged from 11% to > 80% in the lower and higher influx sites, respectively. Nitrogen translocation was an important process for the N budget of the Sphagnum mosses, contributing 0.5–11% of the annual N requirements. These results suggest that N translocation is as important as direct N retention from atmospheric deposition for the N budget of the mosses. Contrary to expectations, N translocation was greater in the high (Adirondack) than in the low (Maine) deposition sites.
  • • If N translocation is closely tied to water availability, the relative positions of the water tables in the sites over the course of the experiments might account for differences in N translocation among sites. The lower translocation (Maine) sites had lower water tables in the first year of the experiment and experienced a more severe drought in the second year than did the Adirondack sites.

Introduction

Sphagnum mosses are the dominant plants in most ombrotrophic bogs in the temperate and boreal climatic zones. Such bogs are notably poor in available nitrogen (N), partly because they receive all water and nutrients from atmospheric deposition that historically had low N concentrations, and partly because their anoxic, acidic conditions preclude high rates of N fixation ( Waughman & Bellamy, 1980 ; Chapman & Hemond, 1982 ; Howarth et al., 1988 ). To thrive as they do, the Sphagnum mosses must have evolved mechanisms of efficient use of N ( van Breemen, 1995 ; Li & Vitt, 1997 ).

Nutrient translocation from older senescent tissues to metabolically active ones is a key mechanism for reducing nutrient loss in vascular plants growing in nutrient-poor environments (Chapin, 1980; Vitousek, 1982; Aerts, 1990, 1996; Bridgham et al., 1995). The ectohydric bryophytes, to which the genus Sphagnum belongs, do not have vascular conducting tissues. However, several recent lines of evidence indicate that nutrient translocation may be an important mechanism for nutrient retention in Sphagnum. First, the sum of N from atmospheric deposition, mineralization, and N fixation is insufficient to account for tissue N requirements for moss primary production (Aldous, 2002). Second, the C : N ratio of living but nonphotosynthetic moss tissues and the underlying peat is greater than what would be expected from measured rates of N mineralization (Malmer, 1988, 1993). This high C : N ratio could be accounted for if some N was moved upwards to the apical capitula. Rydin & Clymo (1989) used 32P and 14C to demonstrate that Sphagnum recurvum actively moved carbon and phosphorus apically by an internal path. There is no reason to suspect that N would behave differently.

A pathway for nutrient translocation is suggested by recent cytological evidence in Sphagnum mosses (Ligrone, 1993; Ligrone & Duckett, 1998a). Nutrients are thought to be transported symplastically in the internal, living parenchyma cells via a network of plasmodesmata. The Sphagnum mosses are also renowned for their capacity to act as wicks, moving water and dissolved solutes upwards in the extracellular capillary spaces formed between pendant branches and stems (Hayward & Clymo 1982). Nitrogen probably moves acropetally in both pathways; the difference being that the mosses probably have tighter control over solutes in the intracellular pathway than in capillary spaces. In the extracellular pathway, some nutrients may be lost to microbes or vascular plant roots. While this earlier work provides evidence that N translocation does occur in the Sphagnum mosses, its extent has not been quantified (Aerts et al., 1999).

Nitrogen translocation may play a critical role for Sphagnum N nutrition. Although the mosses are assumed to be effective at capturing N from precipitation (Rosswall & Granhall, 1980; Malmer, 1988, 1990; Rydin & Clymo 1989; Rochefort et al., 1990; Li & Vitt, 1997), efficiency of N retention from atmospheric deposition may be significantly less than 100% of deposited N (Aldous, 2002). Furthermore, N in precipitation comes in irregular pulses, while N requirements for primary production do not necessarily coincide with precipitation events. Translocation can continue to supply N to the growing capitula between precipitation events.

Nutrient translocation in Sphagnum mosses may follow a seasonal pattern. If it is controlled by tissue nutrient requirements, then translocation should peak in mid-summer during the period of maximum Sphagnum growth.

Sphagnum mosses may have evolved under regimes of atmospheric deposition of fixed N that were insufficient to support moss primary production, and N translocation could have been an alternative source of this scarce nutrient. However, N deposition has increased significantly since global industrialization ( Lovett, 1994 ; Schaug, 1994 ; Clair et al., 1995 ; Driscoll et al., 1995 ; Torseth & Semb, 1998 ). As N availability increases to the point at which it is no longer scarce, do the Sphagnum mosses respond by decreasing N translocation? The hypothesis of an inverse relationship between nutrient availability and nutrient translocation has been proposed and tested several times for vascular plants. Although there are differences among individual species in their capacity for nutrient resorption, it is unclear whether there is a consistent relationship between these two variables ( Chapin & Shaver, 1989 ; Chapin, 1991 ; Aerts, 1996 ). Because bryophytes differ physiologically and biochemically from vascular plants, and because they have a different system of nutrient translocation, absence of a relationship cannot be assumed for bryophytes.

In this study, N translocation was quantified by tracing N movement with an 15N tracer in Sphagnum mosses over 2 yr. The hypothesis was tested that N translocation would be greater where N availability from atmospheric deposition was lower. The goals of this study were to: (1) measure the translocation of single doses of N tracer over time in Sphagnum mosses; (2) compare rates of N translocation in bogs located in a region of high atmospheric deposition to bogs located in a region of low atmospheric N deposition; (3) compare rates of N translocation in plots that initially received different doses of N tracer; and (4) estimate the contribution of translocated N to the N budget of the mosses.

Materials and Methods

Experimental sites

Nitrogen translocation was investigated in four bogs in the north-eastern USA. Spring Pond Bog A (SPB-A) and Spring Pond Bog B (SPB-B) are two distinct zones, approximately 400 m apart, of a 225-ha peatland complex in northern New York in the Adirondack State Park. Lindsey Brook Bog (LBB) and Vanceboro Bog (VRB) are the ombrotrophic portions of two peatlands in eastern Maine. The Adirondack sites receive 1.0–1.3 g N m−2 yr−1 wet atmospheric N deposition while the Maine sites receive 0.2–0.4 g N m−2 yr−1 (Ollinger et al., 1993; NADP/NTN, 1997–99; CASTNet, 1998–99). All four sites are in regions of granitic metasedimentary, mostly Precambrian rock. Other site characteristics are summarized in Table 1.

Table 1.  Characteristics of study sites, 1998–99 1
 (I & II) SPB-A and SPB-B Adirondacks(III) LBB Maine(IV) VRB Maine
  • 1

    SPB-A, Spring Pond Bog A; SPB-B, Spring Pond Bog B; LBB, Lindsey Brook Bog; VRB; Vanceboro Bog.

  • 2

    Mean annual temperature and precipitation data are from the North-east Regional Climate Center at Cornell University, Ithaca, NY, USA.

  • 3

    Mean peat depth data are from Davis & Anderson (1991 ) for Maine and Williams (1997 ) for the Adirondacks.

  • 4

    Precipitation chemistry data are from the National Atmospheric Deposition Program ( NADP/NTN, 1990 –2000).

Latitude44°22′ N45°25′ N45°33′ N
Longitude74°22′ W67°59′ W67°27′ W
Wetland area (ha)225221198
Mean annual temperature (°C)24.54.54.5
Mean annual precipitation (mm)298010151015
Altitude (m)486163120
Mean peat depth (m)3> 3.263.3
Rate of total N deposition, 19984 (g m−2 yr−1)1.0–1.30.2–0.4

Both vascular and bryophyte species assemblages are similar in all four sites. The bryophyte layer includes Sphagnum capillifolium var. capillifolium (Ehrh.) Hedw. and Sphagnum magellanicum Brid. on the lawns, Sphagnum cuspidatum Hoffm. in the hollows, and Sphagnum fuscum (Schimp.) Klinggr. and Polytrichum commune Hedw. on the hummocks. The vascular species include the ericaceous shrubs Ledum groenlandicum L. (Oeder), Chamaedaphne calyculata L. (Moench.), Kalmia angustifolia L., Kalmia latifolia L., Andromeda glaucophylla Link, Vaccinium oxycoccus L., Vaccinium macrocarpon Ait.; the sedges Rhynchospora alba (L.) Vahl., Eriophorum virginicum L., Carex oligosperma Michx.; the trees Acer rubrum L., Larix laricina (Du Roi) Koch. and Picea mariana (Mill.) B.S.P., and the herbs Sarracenia purpurea L. and Drosera intermedia Hayne.

The translocation experiment

At each site, nine square plots were selected randomly from site maps upon which a grid had been superimposed. All experiments were carried out using the dominant lawn species, S. capillifolium, therefore plots where this species was not dominant were rejected. Plots were 50 cm on a side with a 10-cm buffer strip. Three plots received experimental high-dose N treatments, three plots received experimental low-dose N treatments and three plots were control plots that did not receive any treatments. The 15N tracer was applied at the high dose to an additional six plots in the Adirondack sites that were used solely to measure N uptake into the vascular plants. The plots were at least 4 m apart.

A manually pressurized sprayer was used to apply 3 l of 15NH415NO3 solution (5 atom% excess 15N) to each experimental treatment plot. Because the N concentrations mimicked precipitation chemistry in each of the two regions, the doses can be considered N tracers. The high N dose was 3.9 mg N l−1, for a total dose of 0.0468 g N m−2. The low dose was 0.56 mg N l−1, for a total dose of 0.00672 g N m−2. I applied the N tracer to three replicate plots per treatment, on June 28, 1998 in the Maine sites and on July 2, 1998 in the Adirondack sites. For each tracer application, the solution was applied in two batches of 1.5 l, 1 h apart.

In all experimental treatment and control plots, Sphagnum samples were collected six times after the tracer application in the Adirondack sites and four times after tracer application in the Maine sites (Table 2). The apical capitulum (approximately 1 cm), and the top 2 cm of the stem (including leaves and branches) were sampled.

Table 2.  Details of the four N translocation experiments
 (I) Detailed sampling over time(II) Site comparison(III) High–low dose comparison(IV) Annual estimate
  • 1

    SPB-A, Spring Pond Bog A; SPB-B, Spring Pond Bog B; LBB, Lindsey Brook Bog.

Sites includedSPB-A, SPB-BSPB-A, SPB-B, LBBSPB-A, SPB-BSPB-A, SPB-B, LBB
N doses includedHighHighLow, highLow, high
Time periods sampledJuly 1998July 1998July 1998July 1998
 August 1998August 1998October 1998July 1999
 October 1998October 1998May 1999 
 May 1999July 1999August 1999 
 July 1999   
 August 1999   

To determine the extent to which the vascular plants were taking up tracer 15N, vascular plant foliage was sampled twice over the 2-yr experiment in the following way. Two months after tracer application, all new foliage was harvested from the six high-dose N plots in the Adirondack sites that had been identified for this purpose. After harvesting, these plots were no longer used in the experiments. Fourteen months after tracer application, at the end of the experiments in 1999, all vascular plant foliage from all high-dose experimental plots were harvested. Woody tissue was not included as this was produced before the experiment.

All plant samples were rinsed with deionized water, and dried at 60°C for 48 h. All samples were stored at room temperature, and re-dried for 24 h before grinding with liquid N and weighing. Samples from all sampling dates, except August 1998, were analysed for %N and %15N on a Europa Geo 20/20 continuous flow isotope ratio mass spectrometer (Thermo Finnigan, San Jose, CA, USA) at the Cornell Laboratory for Stable Isotope Analysis. The August 1998 samples were analysed for %N and %15N on an Integra continuous flow isotope ratio mass spectrometer (Europa PDZ) at the University of California Berkeley Department of Ecosystem Sciences. The global standards for 15N are atmospheric N.

In each experimental plot, Sphagnum growth was measured using four cranked wires per 0.25 m2 plot (Clymo, 1970). Bulk density of the top 20 cm of the Sphagnum plants and peat was also measured by extracting circular 12-cm diameter and 30-cm deep cores with a sharp knife around a circular frame used to minimize compaction.

Estimates of N translocation

Estimation of N translocation requires understanding how Sphagnum mosses grow and where new N will be found as the plants grow. A Sphagnum plant consists of an apical capitulum of expanding branches, and a stem with fully developed branches (Fig. 1). As the plants grow, the stem elongates from the apex, the branches in the capitulum become distributed along the new stem, and these branches also elongate from a subapical meristem (Ligrone & Duckett, 1998b). Therefore, new growth is found in the stem, in new branches in the capitulum, and in elongated branches along the stem. If tracer 15N is found in the new growth, it must have been translocated to those cells. In these experiments, there is no differentiation between N that was translocated in the intracellular pathway and N moved in the capillary spaces, and both are referred to collectively as translocation. However, only N incorporated into plant tissue, and not free N, is sampled.

Figure 1.

Schematic diagram of the growth of one Sphagnum plant over four hypothetical time periods. Solid lines indicate plant tissue that was present at time t x and dashed lines indicate new growth that occurred after time t x . The vertical lines indicate the samples taken at each sampling time; the distance from the top arrow to the small horizontal line is the capitulum sample, and the distance below the horizontal line to the bottom arrow is the 2 cm stem sample. The amount of growth (G, in cm) is indicated at the bottom of each plant diagram; TL, turnover length of the capitulum.

The 15N translocation between successive time periods was measured in two basic steps. First, I measured Sphagnum growth between time periods and determined the proportion of the newly formed tissues that were in the capitula compared with the stems. (Hereafter, the term ‘stem’ refers to the stem proper plus its branches and leaves.) To do this, the amount of vertical growth required for one capitulum to be replaced was determined (i.e. the turnover length of a capitulum). The proportion of new tissue in the capitulum and the stem was a function of this turnover length. By comparing actual growth to the turnover length, I was able to assign relative amounts of growth either to the capitulum or to the stem, which were sampled separately.

Second, the amount of tracer 15N in those newly formed Sphagnum tissues was determined. This required differentiating between tracer 15N in old tissues (i.e. tissues that were formed prior to the period in question) and tracer 15N in new growth (i.e. new tissues formed within the period in question). Finally, I put the estimates of N translocation in context by comparing them to the amount of N required to support Sphagnum primary production for 1 year.

Details of these eight steps are as follows. First, I estimated the net amount of tracer 15N in the capitula or stems per plot at each sampling time by subtracting the background (control) 15N from the sample 15N. The following equations are for the tracer 15N in the capitula. The same calculations were made for the stems, substituting ‘stem’ for ‘cap’.

15 N total cap  = [N] × [ 15Ncap sample] × BDcap × cap length(Eqn 1)
15 N control cap  = [N] × [ 15Ncap control] × BDcap × cap length (Eqn 2)
15 N net cap  =  15 N total cap  −  15 N control cap  × cap length (Eqn 3)

(15N totalcap, 15N controlcap, and 15N netcap are the amounts of 15N in the capitula sample, the control capitula, and the net amount of 15N in the capitula sample from the tracer application, respectively, all in g m−2; [N] is the per cent of plant tissue that is N in the capitula; BD is the bulk density of the 2-cm stem segment from the 20-cm core, measured in g m−3; cap length is the length of the capitulum sample, measured in m; and 15Ncap sample and 15Ncap control are the percentages of N that are 15N in the capitula sample and the background control, respectively.)

Second, I estimated the turnover length of the capitulum, which is the length of vertical growth after which the whole capitulum is new tissue. Because growth is continuous, capitulum branches will remain part of the capitulum for a certain turnover length until they become stem branches. To obtain a measure of turnover length, I counted the number of branches per capitulum and the number of branches per 1 cm of stem from 10 S. capillifolium plants per plot. The turnover is expressed as a vertical growth increment:

TL = (number of branches per cap)/(number of branches per 1 cm of stem)(Eqn 4)

(TL is the turnover length of the capitulum, in cm stem per capitulum; and ‘cap’ is one capitulum.)

Third, I estimated the amount of 15N that was translocated to new growth in the capitulum plus the stem between sampling time periods. Because these mosses grow slowly, at every sampling period the samples will contain both old 15N (what was there before) and new 15N (newly translocated N) (Fig. 1). Relative contributions of these two pools will depend on how much the mosses grew from time tx to time tx+1. To obtain an accurate estimate of translocated N, from the total 15N from tracer, I subtracted the amount of 15N that was in the sample from the previous sampling time (old 15N). A general form of this equation is:

15 N translocated t x+1  = ( 15 N net cap  +  15 N net stem )t x+1  − (α 15 N net cap  + β 15 N net stem )t x(Eqn 5)

(15N netcap + 15N netstem are from Eqn 3, in g m−2; α is the proportion of the capitulum sample from time tx that is in the total sample at time tx+1; β is the proportion of the stem sample from time tx that is in the total sample at time tx+1.)

Details of this equation are found in Table 3. If growth was less than the turnover length of the capitulum, all of the capitulum from tx would still be found in the plant sample from tx+1 and so α = 1. The amount of stem sample from tx that is lost at tx+1 depends on the growth. Since the stem sample is 2 cm, β is a fraction of 2. If growth is greater than the turnover length but less than 2 cm, then all of the 15N in the capitulum will be new 15N, but α is still 1 because all of the capitulum is found in the stem sample. The value of β is still a fraction of 2. If growth is greater than 2 cm, β is 0 because all of the stem has turned over. As long as growth is less than 2 + the turnover length, some capitulum from tx remains in the sample at tx+1. The value of α depends on the growth of the mosses. If growth exceeds 2 + turnover length, all 15N sampled in the capitulum and stem will be newly translocated N.

Table 3.  Capitula and stem components of N translocation in Sphagnum mosses. See text for detailed description
Vertical growth (cm)1,215 N in capitula, t x+115 N in stems, t x+1α3β4
  • 1

    Vertical growth of the mosses (G) determines the amount of 15 N tracer in the capitula or stems between successive time periods (t x to t x+1 ).

  • 2

    TL, turnover length of the plants from Eqn 4 (see text).

  • 3

    α is the proportion of the capitula sample from time t x that is in the total sample at time t x+1 .

  • 4

    β is the proportion of the stem sample from time t x that is in the total sample at time t x+1 .

G < TLSome old capitula 15NAll old capitula 15N1(2–G)/2
 New 15NSome old stem 15N   
  New 15N  
TL  G < 2New 15Nall old capitula 15N1(2–G)/2
  some old stem 15N  
  New 15N  
2  G < 2 + TLNew 15Nsome old capitula 15N(2 + TL − G)/20
  New 15N  
G  2 + TLNew 15NNew 15N00

Fourth, I estimated the per cent N translocated, which I define as the per cent of 15N tracer found in the plant tissues at time tx, which is translocated and found at tx+1, by dividing the amount of N translocated at any time period tx+1 by the amount in the Sphagnum tissues at time tx:

%translocation = (15N translocated tx+1/(15N net cap + 15N net stem) tx) × 100(Eqn 6)

Fifth, I estimated annual net primary production (NPP) of the Sphagnum mosses:

NPP1998–99 = growth × BD(Eqn 7)

(NPP1998–99 is primary production of the mosses from July 1998 to July 1999, in g m−2, growth is vertical elongation of the Sphagnum carpet measured in m, and BD is the bulk density of the 2-cm stem segment from the 20-cm core, measured in g biomass m−3.) These calculations assume that Sphagnum bulk density does not change over the course of the growing season (Rochefort et al., 1990).

Sixth, I estimated the amount of N required on an annual basis, as the product of NPP and N concentration:

New N1998–99 = (NPP1998–99 × (N1999))(Eqn 8)

(N1999 is the per cent of plant tissue that is N in the stem-leaves and branches in July 1999; NPP1998–99 is from Eqn 7; New N1998–99 is in units of g m−2.)

Seventh, the amount of N translocated between July 1998 and July 1999 was estimated as a percentage of the N initially retained in the mosses. I multiplied the percentage translocation between July 1998 and July 1999 obtained from Eqn 6 by values of N retention from atmospheric deposition (Table 4) to obtain an annual quantity of N translocated:

Table 4.  Nitrogen retention (%) in 1998, for the capitula and top 2 cm stems of Sphagnum mosses
Site1High N (%) Mean (± SE)Low N (%) Mean (± SE)
  1. Data are means of three replicates (± SE). Data modified from Aldous (2002). 1SPB-A, Spring Pond Bog A; SPB-B, Spring Pond Bog B; LBB, Lindsey Brook Bog.

SPB-A, Adirondacks35.20 (7.36)43.97 (5.27)
SPB-B, Adirondacks27.58 (3.52)43.98 (7.35)
LBB, Maine62.25 (1.48)77.45 (10.94)
annual N translocated = (%translocated98–99) × (N retention) (Eqn 9)

where annual N translocated and N retention are both in g m−2 yr−1.

Eighth, I divided the annual N translocated by the new N required from Eqn 8 to obtain a per cent annual contribution of translocated N to the annual N budget:

annual contribution = (annual N translocated98−99/ new N98–99) × 100(Eqn 10)

where both the annual N translocated98−99 and the new N98–99 are in units of g N m−2 yr−1.

Hydrological measurements

To verify that there was no groundwater discharge into the upper 50 cm of the bog, hydraulic head was monitored . At each site, the experimental zone that contained all plots was approximately 100 × 50 m and was surrounded by, and bisected by, clusters of piezometers. Each piezometer cluster included a water table well and open-ended piezometers at 50 cm, 100 cm, and 150 cm. The wells were monitored four or five times per year to determine the direction of groundwater flow. Once each season, I surveyed the wells to obtain elevations for each well relative to an arbitrary benchmark. Each experimental plot also had a water table well, and depth to the water table was recorded at least once each season.

Data analysis

The design of the entire experiment included four sites, two N-dose treatments, and either six (high N) or four (low N) sampling periods. Although all of the samples in that design were collected, only a subset of these samples were analysed for %N and %15N because of time and cost constraints. Because N translocation can be estimated from any pair of sampling periods, the data were subdivided into four experimental groups for statistical analysis, in which the paired time-periods differed (Table 2). Each of the four experiments answers a slightly different question, but they overlap in the data used.

In experiment I, N translocation was compared among five pairs of sampling periods in the Adirondacks (High N deposition region). In experiment II, the sites in both regions were compared using three pairs of sampling time periods. In experiment III, the high- and low-N dose treatments in the two Adirondack sites were compared using three pairs of sampling periods. In experiment IV, annual values of N translocation for all three sites were compared using the July 1998 to July 1999 period, for both the high- and the low-dose treatments. Finally, the contribution of translocated N to the annual N budget of the Sphagnum mosses were compared using the July 1998–July 1999 sampling periods.

Within each of these experiments, the design was balanced. I tested for normality and homoscedasticity by examining residuals and using SAS PROC UNIVARIATE and SAS PROC DISCRIM (SAS Institute, 1990). Data were log-transformed to meet assumptions of normality and homoscedasticity. I compared: (1) tissue N concentration of the mosses; (2) the net amount of 15N tracer at each sampling period; (3) the amount of N translocated; and (4) the per cent contribution of translocated N between dose treatments and among time periods either with a two-way or a three-way analysis of variance using SAS PROC GLM (SAS Institute, 1990). In the analyses of net N tracer in the capitula and stems, a repeated measures analysis of variance was performed using three periods (July 1998, October 1998 and July 1999) with ‘time’ as the repeated measure (SAS Institute, 1990). All variables were fixed in all models.

Results

Plants initially were sampled at four sites: two in Maine and two in the Adirondacks. However, at one of the Maine sites, Vanceboro Bog, many of the plants in the site inexplicably died over the course of the 1998 field season, including Sphagnum mosses that were not in any experimental plots. Analysis of 15N from the tracer experiments confirmed that these plants were extremely stressed, because they lost all 15N tracer, whereas the tracer was translocated in all experimental plots in all other sites. Therefore, these data were eliminated from the analyses and will not be discussed further.

Turnover length of capitulum

For S. capillifolium, I found a mean turnover length of 1.65 cm stem per capitulum. Because no significant differences in turnover among plots or among sites were found, this value is a mean from pooled data.

Sphagnum growth and new N required

Values of Sphagnum NPP and the amounts of N required to support that new growth are shown in Table 5. Both Sphagnum NPP and the new N required to support that growth were significantly greater in the Adirondack sites than in the Maine sites (P < 0.05). Data from both dosage treatments were pooled because no significant differences were found either among treatments for NPP or for the new N required.

Table 5.  Net primary production (NPP), N concentration of upper stems, and new N required for the Sphagnum mosses, from July 1998 to July 1999
Site1NPP (g m−2 yr−1) Mean (± SE)[N] (%) Mean (± SE)New N required (g N m−2 yr−1)
  1. Data are means of three replicates (± SE). Means preceded by the same lower case superscript were not significantly different in pairwise comparisons in the analysis of variance; P < 0.05. NPP and N concentration data are from Aldous (2002) and Aldous (2001), respectively. 1SPB-A, Spring Pond Bog A; SPB-B, Spring Pond Bog B; LBB, Lindsey Brook Bog.

SPB-A, Adirondacksa 520 (5) a 0.99 (0.06) 5.15 (0.31)
SPB-B, Adirondacksa 498 (9) a 1.02 (0.05) 5.08 (0.43)
LBB, Maineb 245 (13) b 0.93 (0.06) 2.28 (0.82)

Net 15N in capitula and stems

The net amount of tracer 15N in the capitula increased over the first sampling season, to a maximum in early October for both the high-N dose (Fig. 2a) and for the low-N dose (Fig. 2b). The net amount of tracer in the growing capitula then declined over the autumn, winter, and the following spring and summer. This pattern was the same for all three sites and for both high- and low-dose treatments, although the magnitudes differed. Both the site and time effects were statistically significant in the repeated measures analysis of variance (Table 6).

Figure 2.

Net 15 N in the capitula (±1 SE) at six (or five) sampling times for Spring Pond Bog A (SPB-A, diamonds), Spring Pond Bog B (SPB-B, squares), and Lindsey Brook Bog (LBB, circles). (a) High-dose N treatment; (b) low-dose N treatment. Note different scales on y-axes.

Table 6.  Results from a repeated measures analysis of variance of net tracer 15 N in the capitula and stems of Sphagnum mosses
 dfHigh doseFPLow doseFP
  1. The repeated measures cover three time-periods. Values of F and associated probabilities in bold are statistically significant at the P < 0.05 level.

Capitula
Site (July 98)2 7.94  0.021 2.29  0.18
Site (October 98)2 2.64  0.15 5.29  0.048
Site (July 99)2 7.51  0.023 4.32  0.069
Site (overall)2 8.63  0.017 7.76  0.022
Time250.21< 0.000184.41< 0.0001
Site × time4 0.49  0.75 1.46  0.27
Stems
Site (July 98)217.66  0.0031 6.66  0.03
Site (October 98)229.05  0.0008 1.51  0.29
Site (July 99)2 1.88  0.2324.59  0.0013
Site (overall)221.54  0.001824.61  0.0013
Time219.27  0.0002 1.13  0.36
Site × time415.07  0.0001 0.37  0.83

Relative net amounts and patterns of tracer 15N in the stems differed among sites. For both dose treatments in the Adirondack sites, and for the low-dose treatment in the Maine site, the net amount of stem 15N tracer did not change very much as the plants grew over the course of the experiment (Fig. 3). By contrast, for the high-dose treatment in the Maine site (Fig. 3a), the pattern of stem 15N tracer was similar to that in the capitulum: there was an initial increase in 15N tracer, followed by a drop over the winter. For both high- and low-dose treatments, the site effect was significant in the repeated measures analysis of variance (Table 6). The time effect was statistically significant only for the high dose treatment. In all cases, the magnitudes of net 15N tracer in both capitula and stems were greater in the high than in the low-dose plots.

Figure 3.

Net 15 N in the stems (±1 SE) at six (or five) sampling times for Spring Pond Bog A (SPB-A, diamonds), Spring Pond Bog B (SPB-B, squares), and Lindsey Brook Bog (LBB, circles). (a) High-dose N treatment; (b) low-dose N treatment. Note different scales on y-axes.

Translocation experiment I: N translocation over five periods

In the Adirondack sites, the per cent N translocated, which is defined as the per cent of 15N tracer found in the plant tissues at time tx that is translocated and found again at time tx+1, was greatest in the first summer (Fig. 4). There was a net loss of 15N over the autumn and winter, between October 1998 and May 1999. There was a net positive N translocation over the second summer, albeit small. In the analysis of variance, the sampling period was significant (F = 58.6, P < 0.001), with the first two periods having the greatest net N translocation, the two 1999 summer periods having the second most, and the autumn and winter period the least. The two Adirondack sites were not significantly different in the analysis of variance.

Figure 4.

Estimates of per cent N translocation (±1 SE) for the high dose treatments at Spring Pond Bog A (SPB-A, light tinted columns) and Spring Pond Bog B (SPB-B, darker tinted columns) over the over six periods between 1998 and 1999. Data are least-squares means. Pooled means of both sites for time periods with the same lower case letter are not significantly different; P  < 0.05.

Translocation experiment II: site comparison

The pattern of per cent N translocated in experiment II for all three sites was similar to experiment I, with translocation being greatest during the first summer, followed by a net loss of N tracer over the autumn and winter (Fig. 5). A significant site effect was found in the analysis of variance, and the two Adirondack sites had greater per cent N translocation than the Maine site (F = 10.6, P < 0.05). The effect of time was also statistically significant (F = 37.5, P < 0.001), with the first two summer periods having greater translocation than the autumn and winter period. No significant site–time interaction was found.

Figure 5.

Estimates of N translocation (±1 SE) at Lindsey Brook Bog (LBB, light tinted columns), Spring Pond Bog A (SPB-A, open columns) and Spring Pond Bog B (SPB-B, darker tinted columns) over four periods. Data are least-squares means. Pooled means of all three sites for periods with the same lower case letters are not significantly different; P  < 0.05.

Translocation experiment III: high–low dose comparison

The pattern of N translocation in experiment III for both doses was similar to the other experiments, with translocation being greatest in the first summer (Fig. 6). While per cent translocation from the low-dose treatment was greater than the high-dose treatment, this difference was not statistically significant. The only statistically significant effect in the analysis of variance was the time-period (F = 74.3, P < 0.01). Per cent translocation was greatest in the first summer, lower during the second summer, and there was a net loss of N over the autumn and winter.

Figure 6.

Estimates of N translocation (±1 SE) over four periods for the high-dose N treatment (light tinted columns) and the low-dose N treatment (darker tinted columns). (a) Spring Pond Bog A (SPB-A); (b) Spring Pond Bog B (SPB-B). Pooled means of both treatments for periods with the same lower case letters are not significantly different; P  < 0.05.

Translocation experiment IV: annual estimate

Annual estimates of percent of N initially retained that was translocated between July 1998 and July 1999 ranged from 11% in Maine to over 80% in the Adirondacks (Table 7). Per cent N translocated was significantly greater in the Adirondack sites than in the Maine site (F = 103, P < 0.01). Neither the N dose effect nor the site–dose interaction was significant in the analysis of variance.

Table 7.  Per cent of N initially retained that was translocated between July 1998 and July 1999
Site1High-N doseLow-N dose
  1. Data are means of three replicates (± SE). Means within a dose group preceded by the same lower case superscript were not significantly different in pairwise comparisons in the analysis of variance; P < 0.05. 1SPB-A, Spring Pond Bog A; SPB-B, Spring Pond Bog B; LBB, Lindsey Brook Bog.

SPB-A, Adirondacksa 69.9 ( 13.3) a 83.2 (4.2)
SPB-B, Adirondacksa 64.5 (14.4) a 74.8 (10.1)
LBB, Maineb 11.1 (15.5) b 32.8 (0.3)

Contribution of translocated N to the N budget

Contribution of translocated N to the annual N budget of the Sphagnum mosses ranged from 0.5% in Maine to over 11% in the Adirondacks (Table 8). This contribution was significantly greater for the Adirondack sites than for the Maine sites (F = 11, P < 0.05). Neither the dose effect nor the site–dose interaction was significant in the analysis of variance.

Table 8.  Per cent contribution of translocated N to the Sphagnum N budget from July 1998 to July 1999
Site1High-N doseLow-N dose
  1. Data are least-squares means. Means within a dose group preceded by the same lower case superscript are not significantly different, from pairwise comparisons in the analysis of variance; P < 0.05. 1SPB-A, Spring Pond Bog A; SPB-B, Spring Pond Bog B; LBB, Lindsey Brook Bog.

SPB-A, Adirondacksa 9.5 (1.8) a 11.4 (0.6)
SPB-B, Adirondacksa 6.6 (1.5) a7.6 (1.0)
LBB, Maineb 0.5 (0.6)  b 1.3 (0.01)

15N tracer in vascular plant foliage

Some of the 15N tracer was detected in the vascular plant foliage (Table 9). There was no statistically significant difference in the amount of tracer 15N in the foliage between the 2-month and 14-month periods. However, the site term was significant in the analysis of variance for the 2-month samples (F = 49.64, P < 0.01) and for the 14-month samples (F = 9.05, P < 0.01). In both cases, tracer 15N in the SPB-B plots in the Adirondacks was greater than all other plots (P < 0.05). The Sphagnum capitula and upper stems contained over 100 times more tracer than the vascular foliage, both at 2 months and 14 months.

Table 9.  Tracer 15 N in vascular plant foliage and ratio of tracer in the sum of Sphagnum capitula and upper stems to tracer in vascular foliage, either 2 or 14 months after tracer application
Site12 Months14 Months
15 N content (g cm −2 ) Sphagnum : vascular 15 N content (g cm −2 ) Sphagnum : vascular
  1. Data are means (± SE) of three experimental plots. Means within a time group preceded by the same lower case superscript are not significantly different, from pairwise comparisons in the analysis of variance; P < 0.05. 1SPB-A, Spring Pond Bog A; SPB-B, Spring Pond Bog B; LBB, Lindsey Brook Bog.

SPB-A, Adirondacksb 3.2 × 10 −9 (3 × 10 10 ) 143 : 1b 6.9 × 10 −9 (7 × 10 −10 ) 143 : 1
SPB-B, Adirondacksa 6.4 × 10 −9 (3 × 10 10 ) 111 : 1a 1.2 × 10 −8 (4 × 10 10 ) 200 : 1
LBB, Mainen/an/ab 7.4 × 10 −9 (2 × 10 −9 ) 100 : 1

Site hydrology

In all three sites, the surface 50 cm was a zone of groundwater recharge during the 1998 and 1999 growing seasons. Thus, no N could have been supplied by groundwater. Groundwater movement below 50 cm was both spatially and temporally heterogeneous.

During the summer 1998 N-tracer application, the water table position was similarly high in all three sites (Fig. 7a). However, over the 1998 growing season, the water table dropped in the Maine site to 15 cm below the surface, while it remained high in the Adirondacks. In 1999, a regional drought caused the water tables to drop in all sites to more than 17 cm below the surface (Fig. 7b). This drought may have been more extreme in the Maine sites, where climate models indicated complete depletion of soil water by mid-summer (Keith Eggleston, North-east Regional Climate Center, Cornell University).

Figure 7.

Depth to water table (±1 SE) for Spring Pond Bog A (SPB-A, diamonds; n  = 10), Spring Pond Bog B (SPB-B, squares; n  = 9), and Lindsey Brook Bog (LBB, circles; n  = 6) in (a) 1998 and (b) 1999. Data are from water table wells at piezometer clusters and in experimental plots. Arrows indicate the dates of tracer application.

Discussion

These data show unequivocally that Sphagnum mosses translocate N upwards to the growing capitula. While such a process has been postulated for Sphagnum mosses in the past (Malmer, 1988; Rydin & Clymo, 1989; Malmer, 1993), these data are the first quantitative estimates of N translocation over time.

I initially hypothesized that N translocation would be lower in the Adirondack sites where N deposition was higher. These results did not support this hypothesis. The per cent N translocated (Fig. 5; Table 7) and the contribution of translocated N to the annual N budget (Table 8) were less for the Maine mosses than for the Adirondack mosses.

Although rates of N deposition over time might be an important factor in the development of Sphagnum communities, other factors, particularly climate, may overwhelm expected patterns. Depth to the water table in all sites was similar at the time of tracer application, but the water tables dropped in the Maine sites throughout the 1998 growing season, whereas they remained higher in the Adirondack sites (Fig. 7). In that first summer, N translocation was significantly lower in the Maine sites both for the July–August and the August–October sampling periods. If translocation of plant nutrients such as N depends to some extent on the availability of water for vertical transport, the position of the water table with respect to the surface of the Sphagnum mats will be a critical factor in determining the extent of nutrient translocation.

Patterns of N translocation in the low- and high-dose plots were similar (Fig. 6). No differences were found between these two doses, confirming that both behaved similarly as tracers and that N was equally translocated at doses comparable to atmospheric N deposition in both regions.

Patterns of N translocation over time

Translocated N as a source for the growing mosses was most important shortly after deposition. Initial N retention at the time of deposition can range from 50% to 90% of deposited N (Li & Vitt, 1997; Aldous 2002). If initial retention is at the low end of this range and a large proportion of the N is not taken up immediately in the capitulum and upper branches, it will percolate down through the mat of moss and peat. Translocation is analogous to a ‘second net’ for N capture by the mosses, because the mosses move that missed N back up to the growing apex, thus increasing the overall efficiency of N capture.

Over time, the amount of N translocated from the initial pulse declined. It is likely that the quantity of labile N remaining from the application decreased as it became incorporated into the moss tissue, was taken up by microbes and plant roots, and was adsorbed as ammonium onto cation exchange sites in the peat.

All plots lost 15N over the course of the autumn and winter (Fig. 5), mostly from the stems rather than the capitula (Fig. 3a). These losses may have been due to leaching of N associated with the freeze–thaw cycle and/or snow melt. N may be lost either by physical removal by melting snow or by replacement with new N in snowmelt. The snowpack accumulates deposited N over the winter and so snowmelt delivers a large pulse of N to the ecosystem (Galloway et al., 1987; Schaefer & Driscoll, 1993; Baisden et al., 1995), which may displace the old N.

Although all plots lost 15N over the winter, losses were greater in the Maine plots than in the Adirondack plots. It is difficult to explain this difference. The three sites vary in background N loadings from atmospheric deposition. If N displacement resulting from a large pulse of N in snowmelt was the primary cause of N loss, then greater losses in the Adirondacks where N deposition is greater would be expected. However, the opposite was true. Perhaps differences in snow cover, temperature, the freeze-thaw cycle or the previous summer's drought may have caused greater N leaching in Maine than in the Adirondacks. Alternatively, a difference in the morphology of the mats of Sphagnum mosses may have resulted in differences in N leaching. Bulk density of the mosses in the Maine sites was twofold greater than in the Adirondack sites (Aldous, 2002) and therefore these plants had a greater volume and surface area from which N might be leached.

During the second summer, there was again a net positive, albeit small, N translocation from the initial pulse of N tracer. At this point, there are three likely sources of the translocated N. The first is N lost from microbes during turnover of the microbial pool. Although I did not measure retention of tracer 15N by the microbial community, microbes compete with plants for inorganic N in peatlands and can be assumed to have taken up some of the tracer (Kaye & Hart, 1997; Gilbert et al., 1998a,b; Jaeger et al., 1999).

The second possible source of translocated N is from fine roots. Vascular plants also may be competing with the mosses for N and the time scale of fine-root turnover is short enough to be a source of mineralized N (Backeus, 1990; Jackson et al., 1990; Saarinen, 1996). This source is probably not large. In my experiments, above-ground vascular plant tissues contained less than 1% of the 15N tracer compared with the mosses (Table 9). In a similar experiment, Li & Vitt (1997) recovered less than 2% of applied N in the above-ground shrubs.

The third possible source of translocated N is ammonium from cation exchange sites in the peat or on the mosses. The bonds formed between ammonium and the exchange sites are weaker than those with many other cations, such as potassium, magnesium, and calcium (Sposito, 1984; Schlesinger, 1991). While these base cations are normally found at very low concentrations in the pore-water of ombrotrophic bogs (Vitt, 1990; Vitt & Chee, 1990), small amounts might displace the ammonium from exchange sites, at which point the ammonium could be transported acropetally to the growing apex.

A source of translocated N that probably does not play a role within the time-frame of this experiment is from Sphagnum plant tissue that is no longer metabolically active. Amino acids either may be broken down intracellularly by the moss itself or mineralized by microbes, and then become available for vertical transport. Because insufficient time passed between the initial N tracer addition and sampling, the tissue that was being synthesized at the beginning of the experiment will not yet have senesced and decomposed. However, this process is likely to become a source of N for translocation over longer periods.

Relative importance of N translocation

Nitrogen translocation probably plays a significant role in the Sphagnum annual N budget, assuming that the water table is sufficiently high to support this process. In Maine, 1.3% of the annual N budget from the low dose tracer was estimated to have been derived from translocated N from one pulse of 15N tracer. Similarly, in the Adirondacks, 9.5% of the annual N budget from the high-dose tracer was estimated to have been derived from translocated N. In comparison, estimates of N retention from wet atmospheric deposition accounted for less than 10% of N required to support annual primary production (Aldous, 2002). Therefore, initial N retention from atmospheric deposition and N translocation might be of equal importance for the N budget of the Sphagnum mosses.

The combination of N retention and N translocation, however, cannot account for all N required by the Sphagnum mosses. This conclusion contrasts with the assumption that Sphagnum mosses in bog ecosystems obtain all N required for plant production from direct atmospheric inputs (Tamm, 1954; Lee & Woodin, 1988; Brown & Bates, 1990; Malmer, 1993). Other potential sources of N include N mineralization, N fixation, and groundwater inputs. Studies of similar ecosystems have shown that N mineralization in bogs can be as high as 4–5 g N m−2 yr−1 (Urban & Eisenreich, 1988; Bridgham et al., 1998; Aerts et al., 1999) but it is not clear whether mineralized N is an important component of the Sphagnum N budget. For example, the vascular plants, with their fine roots scavenging the peat for mineralized N, may be better competitors for this source of N (Backeus, 1990; Jackson et al., 1990). While N-fixing cyanobacteria have been observed to colonize the water-holding hyaline cells of Sphagnum mosses in some bogs (Granhall & Selander, 1973; Basilier et al., 1978; Henriksson et al., 1987; Sheridan, 1991), the acidic conditions of bogs usually preclude N fixation (Howarth et al., 1988). Furthermore, I examined Sphagnum hyaline cells for cyanobacterial colonization, and did not find any (unpublished). In this study, groundwater supplied no N to the plant-rooting zone; the hydrology data show that these four bogs are precipitation-fed during the summer months and water moves down through the peat during this time. Therefore, these results suggest that a combination of internal N cycling via mineralization, N capture from atmospheric deposition and N translocation are the three key processes supplying N to the growing mosses.

Although these data did not support the hypothesis of an inverse relationship between N translocation and N availability, they do point to N translocation as a key player in bog nutrient cycling. That nonvascular Sphagnum mosses can play such a role only adds to their reputation as ‘ecosystem engineers’ (van Breemen, 1995). Further, these data suggest the greater role of internal cycling processes over external nutrient acquisition (Hobbie, 1992) and of natural selection in nutrient-poor environments on characteristics that favour nutrient retention (Aerts, 1996; Aerts et al., 1999).

Acknowledgements

Several organizations provided funding for this research, including the Fonds Pour la Formation de Chercheurs et l’Aide a la Recherche (Government of Québec), the Society of Wetland Scientists, the Kieckhefer Foundation, and Cornell's National Science Foundation-funded Program in Biogeochemistry and Environmental Change. This manuscript was significantly improved by reviews by Barbara Bedford, Todd Dawson, Tim Fahey, Richard Clymo and one anonymous referee. Tremendous field and laboratory assistance was provided by Mace Vaughan, Jill Clougherty, Kathy Bailey, Carmen Chapin, Jim Burdett, and Roman Pausch.

Ancillary