Author for correspondence:R. J. Mitchell Tel: +1 229 734 4706 Fax: +1 229 734 4707 Email: firstname.lastname@example.org
•N2 fixation rates of three legume species and the impact of fire regime are reported.
•Summer, winter, and no burn treatments were applied. N 2 fixation rates ( 15 N isotope dilution) and C trade-offs with flowering and fine root turnover were examined in response to season of burn.
•Tephrosia and Centrosema had uniformly high percentage N dfa across all treatments (74–92% N dfa ), whereas Rhynchosia showed limited N 2 fixation activity (18% and 0%). No evidence for decreased N 2 fixation due to loss of leaf area following growing season burns was found. Moreover, no consistent evidence for decreased N 2 fixation with greater flowering or fine root turnover was observed.
•Despite species differences in response to fire regime, the following patterns emerged: when increased N 2 fixation is associated with decreased growth rates, legumes show limited N 2 fixation rates (as seen in Rhynchosia ). Alternatively, if greater N 2 fixation is related to increased growth rates, then legumes experience C limitations to N 2 fixation only in small individuals or during periods of rapid growth (as in Centrosema ). Reproduction may influence N 2 -fixation, but, as in the case of Tephrosia , the relationship was positive, opposite to patterns indicative of C trade-offs.
Although fire can cause short-term increases in N availability (Christensen, 1977; Hulbert, 1988; Dudley & Lajtha, 1993), burning leads to substantial N losses in longleaf pine ecosystems through volatilization, ash transport, and leaching (Christensen, 1981). Repeated burning can also promote the development of a positive feedback between nitrogen availability and litter substrate quality, which may diminish long-term N availability (Christensen, 1977; Ojima et al., 1994). Frequent fires lead to low-quality litter inputs, resulting in rapid decreases of soil organic N and microbial biomass (Ojima et al., 1994).
Through symbiotic N2 fixation and high-quality litter inputs, legumes may offset N losses and increase N-availability (Hendricks & Boring, 1992, 1999; Towne & Knapp, 1996). Although Hainds et al. (1999) found legume species to be widely distributed and stem densities to be uniformly high across a moisture gradient (Hainds et al., 1999), very little is known about the N2-fixing ability of perennial legumes found in the longleaf pine-wiregrass (Aristida beyrichiana) ecosystem. Because N2 fixation rates can vary considerably among herbaceous legume species, legume population size alone is not a sufficient indicator of N2 fixation activity (Hendricks & Boring, 1999).
Fire season can affect fixed N2 use efficiency in legumes by consuming leaf area of legumes and by altering carbon partitioning to tissues above and below ground, as well as to reproductive structures. Within-plant C competition for growth, reproduction, defense (Antonovics, 1980) and, for legumes, N2 fixation (Pate, 1996) is a well-established concept. Fire regime and, specifically, season of burn can significantly influence growth and reproduction in native forbs (Platt et al., 1988; Brewer & Platt, 1994; Hiers et al., 2000). The pre-Columbian fire regime in longleaf pine savannas is thought to have been characterized by frequent, low-intensity, surface fires that occurred every 1–5 yr (Christensen, 1981), and season of burn is hypothesized to have been an important element of this disturbance regime (Platt et al., 1988; Brewer & Platt, 1994). In the south-east, fires traditionally have been set from late-February to mid-April for game management (Lemon, 1949; Stoddard, 1950; Grelen & Epps, 1967). This late winter/early spring fire season, however, differs from the putative pre-Columbian fire regime when wildfires, ignited by lightning, are thought to have occurred between May and August (Robbins & Myers, 1992).
Trade-offs among reproduction, N2 fixation, and fine roots have been documented for crop legumes (Pate, 1996), but have yet to be investigated for perennial legumes.
To begin to assess the importance of legumes in maintaining N in longleaf pine savannas under different burn regimes, the objectives of this study are: first to determine N2 fixation rates of dominant native legume species under season of burn and unburned treatments; and second to explore the effects of fire and season of burn on N2 fixation and C partitioning in perennial legumes (Fig. 1). We hypothesize that fire season will affect fixed N2 use efficiency through tradeoffs in allocation of C among plant functions. Specifically, we predict that: first crown development (above-ground biomass) will be positively correlated with N2 fixation; second that lower N2 fixation rates will accompany increased reproduction; and that third increased fine root dynamics (increased mortality followed by replacement of roots) and canopy replacement following fire in the growing season will decrease N2 fixation rates.
Materials and Methods
The study site was established in an abandoned agricultural field at the Joseph W. Jones Ecological Research Center in Baker County, GA, USA. The field was of the Wagram soils series (USDA, 1968), a soil type on which longleaf pine savannas commonly occur in south-western Georgia. This area had not been cultivated for nearly a decade. Soils were analyzed for residual phosphorus via double acid extraction before plot establishment. The value of 8.8 kg P ha−1 is only slightly higher than values reported for Wagram soils supporting native groundcover (Hainds et al., 1999), and these values are within the range of P found in intact longleaf pine savannas of south-western Georgia (Hainds et al., 1999; Jacqmain et al., 1999).
For this study, we chose the three most abundant perennial legume species based on stem density and frequency of occurrence in longleaf pine communities at the Jones Center: Tephrosia virginiana (L.) Persoon, Rhynchosia reniformis DC, and Centrosema virginianum (L.) Bentham (Hainds et al., 1999). Although a survey of longleaf pine savannas found more than 45 species of legumes, these three species collectively represented > 25% of the total legume stems (Hainds et al., 1999). They differ greatly in size, growth form, reproductive phenology, and morphology. Tephrosia virginiana (Goat's Rue) is a long-lived, perennial forb that forms discrete clusters of stems emanating from a subsurface root crown. Centrosema virginiana (Butterfly Pea) is a perennial herbaceous vine, several meters in length, which spreads clonally by rhizomes. Rhynchosia reniformis (Dollar Pea) is a small forb that produces several stems from a small, perennating taproot, but it is widespread in its distribution. Non-N2-fixing native forbs, Asclepias tuberosa L. and Dyschoriste oblongifolia (Michx.) Kuntze, were selected as reference plants to be used in the estimation of N2 fixation rates based on similarity of growth characteristics and reproductive phenology to the legumes studied. Reference plants were randomized within the study design and subjected to all experimental manipulations.
Garden plot establishment
In 1997 a garden plot was established to obtain estimates of N2 fixation using 15N isotope dilution. This approach was chosen, in part, because N2 fixation in naturally established populations has been difficult to assess due to methodological shortcomings (Hamilton et al., 1993; Handley & Scrimgeour, 1997). Low 15N enrichment levels, unpredictable sources of isotope fractionation, and/or spatial variation can lead to spurious results using techniques based on 15N natural abundance to estimate N2 fixation (Handley & Scrimgeour, 1997). Nonuniform labeling and unequal rooting volumes potentially confound the use of the isotope dilution technique in intact communities (Schoeneberger et al., 1989). Using isotope dilution in a garden plot setting provided some level of control for these factors and facilitated total recovery of below-ground biomass, which was important to test our hypotheses.
Root cuttings of each species, including reference plants, were collected in the winter of 1997 from legumes within intact longleaf pine savannas and were standardized to a similar size. Cuttings were potted in the glasshouse, allowed to sprout, and exposed to native Rhizobium using a native soil/sand mixture. Plants were randomly assigned a position in the garden plot and were planted in May 1997, allowing one full growing season for acclimation. Observation of fine roots at the time of outplanting showed the presence of nodules on both T. virginiana and C. virginiana, indicating sufficient exposure to native Rhizobium. For each individual, a 30-cm wide by 75-cm deep hole was excavated. A 25-cm diameter cylindrical bag constructed of landscape cloth was buried to a depth of 75 cm, confining legume roots (and nutrient uptake) to the soil volume within the bag. Bags were filled with the displaced soil from the old field, and a cubic litre of topsoil from an intact longleaf pine savanna was added to the top of each bag to ensure continued exposure to native Rhizobium. A 25-cm diameter PVC collar was placed at the top 10-cm of each bag to confine lateral root growth and to facilitate the application of 15N-enriched fertilizer.
15 N labeled (NH 4 ) 2SO4 fertilizer was applied in May 1997, September 1997, and March 1998 with 5.6, 10.0, and 99 atom percentage excess 15N, respectively. A total of 5.47 g N m−2 was added in 1997, but only 0.03 g N m−2 were added in 1998. This application helped to maintain high levels of 15N enrichment throughout the study, while adding minimal total N to the study plants in the growing season of 1998. Labeled fertilizer was applied with enough water to saturate the top 10 cm of soil to distribute the 15N label throughout the rooting zone.
A preliminary harvest of plants in August of 1997 revealed that vigorous production of lateral roots by T. virginiana had caused small tears in the landscape cloth bags. For the remaining plants, roots outside the cloth bags were pruned to a depth of 35 cm. Additional plastic cylinders (30 cm diameter) were sunk to a depth of 35 cm to prevent future lateral root escape. By the end of the experiment, comparisons of pruned T. virginiana vs six control plants showed no significant effect of pruning on flowering (F1,11 = 1.09, P = 0.32) or biomass (F1,11 = 0.025, P = 0.88).
Summer burn (mid-June), late winter burn (mid-March), and no burn treatments were randomly assigned to individuals of each species. Approximately 50 g of fine fuels (longleaf pine needles) were added to individual plants and ignited using a burn tool described by Van Eerden (1997). With a crank mounted on a tripod, a propane torch was lowered onto each individual plant. Time-temperature profiles were generated on a laptop computer using thermocouples placed at ground level. Real-time temperature data displayed on the computer in the field helped to control both maximum temperature and residence time of each fire (Van Eerden, 1997). This process allowed us to minimize variation between treatment fires, avoiding confounded factors related to fire intensity and season of burn (Streng et al., 1993).
Flower production was censused weekly. To test the hypothesis that investment in reproduction should decrease N2 fixation, we randomly selected six individuals of each species burned in March and clipped all reproductive buds soon after initiation, preventing flowering throughout the study. Clipped buds were dried and weighed as part of the final plant biomass for clipped individuals. N2 fixation and growth were compared between the clipping treatment for the March burn treatment. Due to significant mortality in R. reniformis, this species was excluded from the clipping experiment.
Six individual plants of each legume species in each treatment were destructively harvested three times during the 1998 growing season (March, June, and August). Due to high mortality following transplanting, R. reniformis was also excluded from the June harvest. Immediately before excavation, above-ground biomass was clipped. After excavation of the bags, soil containing the rootstock was passed through a 2-mm mesh sieve. Roots were carefully washed and sorted into coarse roots (> 2 mm in diameter), fine roots (< 2 mm in diameter), and nodules. All samples were dried at 70°C to a constant mass and weighed.
We determined fine root production and mortality rates with a 15N mass balance approach described by Hendricks et al. (1997), which utilized the 15N enrichment of the soil. This method estimates fine root production and mortality without modifying root growth. Because of hypothesized treatment differences in fine root mortality, we could not rely on standing crop alone to estimate N from fine root inputs. Although there is no standard way of estimating the 15N concentration of plant-available soil N (Hendricks et al., 1997), we chose to use the nondiazotrophic reference plants as estimates of soil 15N concentration. The assumption that reference plants are accurate measures of plant available soil N is commonly made when using reference plants to estimate N2 fixation rates (Schoeneberger et al., 1989), and the extension of this assumption to estimate fine root turnover is reasonable within the constraints of the mass balance approach. N2-fixing plants had access to both soil and atmospheric N pools. It was necessary to account for this additional pool of 15N when calculating fine root turnover for legumes. Thus, we diluted estimated available 15N for legumes by the amount of mass of N fixed by each individual, assuming atmospheric 15N enrichment to be zero. Because we excluded R. reniformis from the June harvest, it was not possible to calculate fine root production and mortality for that species.
N2 fixation measurements
All harvested individuals were analyzed for N2 fixation rates using 15N isotope dilution. 15N-enrichment of plant tissue was determined using a Finnegan Ratio Mass Spectrometer at the University of Georgia's Institute of Ecology Chemical Analytical Laboratory. To improve the accuracy of N2 fixation estimates, shoots, coarse roots, and fine roots were analyzed individually for 15N concentration, and N2-fixation was calculated by the weighted atom percentage 15N excess (WAE) for whole plants using the following equations described by Danso & Kumarasinghe (1990):
where A.E. (S), A.E. (C), and A.E. (F) refer to the atom percentage 15N excesses in shoots, coarse roots, and fine roots, respectively; TN(S), TN(C), and TN(F) refer to total N in shoots, coarse roots, and fine roots, respectively. Percent N2 derived from the atmosphere (% Ndfa) and total N2 fixed were calculated using the following equations:
Although percentage Ndfa is often used as a comparative measure of N2 fixation (Danso & Kumarasinghe, 1990; Hendricks & Boring, 1999), it cannot account for potential differences in tissue N between legumes. Estimates of total N2 fixed that was incorporated into tissue N overcomes this, but makes comparisons among plants that differ in size difficult. By dividing total N2 fixed by total biomass,
Fixed-N2 use efficiency (NFUE) = Total N2 fixed/Total biomass(Eqn 4)
NFUE is measured as g of N2 fixed per g of biomass. As calculated in eqn 4, NFUE incorporates differences in tissue N while providing a relative metric of fixed N2 use efficiency between plants of different sizes. This metric is analogous to nitrogen use efficiency.
This experiment was arranged in a completely randomized design and analyzed using two-way anova with treatment and species as fixed effects in PROC MIXED (SAS, 1996). Simple effect slices of species by treatment interactions were used to examine treatment differences at each level of species. A priori mutually orthogonal planned contrasts were used to compare treatment differences between season of burn (March vs June) and burn (March + June) vs no burn treatments. Correlations were used to explore whether fire regime altered fixed NFUE through changes in internal C allocation to different plant functions. R. reneformis was excluded from most correlations due to excessive mortality. Both linear and nonlinear regressions and correlations were performed using Sigma-plot/Sigma Stat (SigmaPlot, 1997). Nonlinear regressions were calculated iteratively with a maximum likelihood, best-fit curve. Goodness of fit between linear and curvilinear regressions was determined based on coefficients of variation and statistical significance.
Direct effects of fire on NFUE
Both percentage Ndfa and g N2 fixed g−1 biomass varied significantly among the three species of legumes (P < 0.0001) but showed no statistical difference with respect to the main effect of burn treatment (P > 0.05) (Fig. 2a,b). Tephrosia virginiana and C. virginianum had uniformly high percentage Ndfa across all treatments (74–92% Ndfa), whereas R. reniformis showed limited NFUE (18% and 7% in March and June burn treatments, respectively) and no discernible fixed NFUE in unburned plants.
Grams of N2 fixed per gram of biomass (a relative measure of fixed NFUE that incorporates tissue N) indicated a significant species by treatment interaction (P < 0.01) (Fig. 2b). Statistical contrasts of treatment means for C. virginianum indicated that burned individuals had significantly higher NFUE than unburned individuals (F1,37 = 4.09, P < 0.05), but the season of burn effect was not significant (F1,37 = 1.15, P > 0.05). In T. virginiana, contrasts showed that significantly more NFUE occurred following March burns than June burns (F1,37 = 16.03, P < 0.001) (Fig. 2a).
Total N2 fixed by legumes in this study was primarily related to patterns of biomass production. Due to differences in plant size, there was a significant main effect of species on total N2 fixed (P < 0.0001) (Fig. 2c). Season of burn contrasts for total N2 fixed were also significant (F1,37 = 11.94, P < 0.005). For all species, total N2 fixed was highest in March burn treatments, but the relationship of total N2 fixed across treatments differed by species, as indicated by a significant interaction (P < 0.0001). For C. virginianum, lowest total N2 fixed was observed in unburned individuals, with approximately equal means for both burn treatments. For T. virginiana, individuals burned in June fixed the least total N2 relative to individuals that were unburned or burned in March Total N2 fixed by R. reniformis was negligible across all treatments.
Tissue N concentration also varied significantly among species (P < 0.0001) (Fig. 2d). Rhynchosia reniformis had significant lower tissue N than did C. virginianum or T. virginiana, but species responded differently to burn treatments as shown by a highly significant species by treatment interaction (P < 0.0001). For treatment means of both C. virginianum (P > 0.05) and R. reniformis (P > 0.05), tissue N was relatively constant, with individuals burned in June containing slightly higher percent N. Tephrosia virginiana had significantly elevated tissue N following March burns (P < 0.001), averaging 2.76% N, compared with 2.34% and 2.15% N for no-burn and June – burn treatments, respectively.
Effects of fire regime on C trade-offs
Correlations of reproduction and growth against NFUE were used to explore the species-specific responses of these plant functions to fire regime. In C. virginianum, patterns among plant functions did not differ with respect to season of burn; rather, across treatments, NFUE and reproduction appeared to be positively related to above-ground biomass. NFUE showed a strong asymptotic relationship to above-ground biomass (r2 = 0.75, P < 0.0001) (Fig. 3a). Fixed N2 use efficiency rapidly reached peak values at relatively small plant sizes and became insensitive to further increases in biomass. Plant reproduction (i.e. total number of flowers) also showed a strong relationship with above-ground biomass (r2 = 0.81, P < 0.0001) and increased linearly throughout the range of plant sizes studied (Fig. 3b).
Reproduction differed substantially across burn treatments in T. virginiana, as did the relationship between NFUE and growth. In unburned T. virginiana, NFUE was not related to increases in above-ground biomass (r2 = 0.36, P = 0.28). Reproduction, which was very low in unburned plants, also showed no relationship with above-ground biomass (r2 = 0.06, P = 0.85) or NFUE (r2 = 0.02, P = 0.83). In contrast, plants burned in March produced three times as many flowers as unburned plants. In the March burn treatment, NFUE and above-ground biomass were positively related (r2 = 0.66, P < 0.05), whereas reproduction showed no significant relationship to above-ground biomass (r2 = 0.40, P = 0.24) (Fig. 4a,b). NFUE and reproduction were strongly positively related (r2 = 0.83, P < 0.05) (Fig. 4c).
Of all three treatments, individuals burned in June showed greatest variation in flower production, ranging from almost no flower production to the greatest observed flower production. When flowers were few, NFUE and flower production were closely related but, as flower number increased, this relationship reached a threshold of fixed N2 use efficiency (Fig. 5a). Following June burns, total number of flowers showed no significant relationship with above-ground biomass (r2 = 0.48, P = 0.12) (Fig. 5b). When plotted against below-ground biomass, however, reproduction showed a significant positive relationship (r2 = 0.89, P < 0.001) (Fig. 5c). Fixed N2 use efficiency, however, remained linearly positively related to above-ground biomass (r2 = 0.93, P < 0.001) (Fig. 5d). With all treatments pooled, above-ground biomass and NFUE showed a curvilinear, asymptotic relationship similar to that of C. virginianum, despite the overall complex response to season of burn by T. virginiana (Fig. 6).
Following March burns, individuals of C. virginianum and T. virginiana whose reproductive buds had been clipped showed no statistical differences in total biomass, root:shoot ratios, and percentage Ndfa from individuals that had been allowed to reproduce (Table 1). Moreover, C. virginianum showed no change in NFUE rates, tissue N, and fine root biomass. Although biomass of clipped plants was slightly higher than individuals that were allowed to flower, this trend was not statistically significant (P > 0.05) for either species. For T. virginiana, flowering individuals showed significantly higher tissue N (P < 0.001) and greater NFUE per unit biomass (P < 0.01) than plants prevented from flowering. In flowering T. virginiana, this increased NFUE was accompanied by significantly lower fine root biomass relative to clipped plants (P < 0.05).
Table 1. Treatment means for clipping experiment: biomass, root to shoot ratios, % N dfa , grams of N 2 fixed per gram biomass, tissue N, and fine root biomass
Root : shoot ratio
N2 fixation (%Ndfa)
N2 fixation (g N g−1 biomass)
Tissue N (%)
Fine Root Biomass
P < 0.05;
P < 0.01;
P < 0.001. Rhynchosia was excluded due to high mortality.
Effects of fire regime on legume production and tissue inputs
By the end of the second growing season, patterns of total legume biomass showed significant main effects of species (P < 0.0001) and of treatment (P < 0.05), as well as significant treatment by species interaction (P < 0.01). Planned contrasts showed significantly higher total plant biomass in March burns relative to June burns (F1,38 = 7.97, P < 0.01).
Biomass allocation patterns also differed significantly among burn treatments (P < 0.01) and species (P < 0.005) (Fig. 7). Proportionally less biomass was allocated above-ground following June vs March burns (F1,37 = 11.79, P < 0.005), with the mean root:shoot ratio increasing for all species in June burns relative to March burns. Despite proportionally more biomass being allocated below-ground following growing season fires, total root biomass was significantly greater in March vs June burns (F1,37 = 4.17, P < 0.05).
Both fire regime (P < 0.05) and legume species (P < 0.0001) significantly influenced the annual inputs of legume litter, collected above-ground and fine root biomass (Table 2). June burn treatments showed significant decreases in mean litter inputs relative to March burns (F1,36 = 8.12, P < 0.01). For C. virginianum, primary decreases in turnover biomass resulted from lower investment in above-ground biomass following June burns, as fine roots showed no differences in production and mortality across treatments (Fig. 8a). Fine root production was significantly different among burn treatments for T. virginiana (P < 0.05), but treatments did not affect fine root mortality (Fig. 8b) (P > 0.05). For all three species, above-ground biomass was the largest pool (79–92%) of potential legume N inputs (Table 2).
Table 2. Comparison of annual litter inputs (collectively, aboveground biomass, fine roots, and nodules) and potential N inputs via turnover of these tissues. Significance values reported below were calculated using simple effect slices to partition treatment by species interaction. The interaction ( *** P < 0.001) was significant for legume tissue inputs and legume contribution of fixed N 2 via annual tissue inputs
This pattern of tissue turnover, in turn, influenced estimated individual legume N contribution (Table 2). Treatment (P < 0.05) and species (P < 0.0001) were significantly different with respect to inputs of symbiotically fixed N2. Across all three species, March burns (mean = 3.03 g N per plant) significantly increased (F1,36 = 8.61, P < 0.01) the potential contribution of legume N via tissue turnover compared to June burn treatments (mean = 1.25 g N per plant). Species differed, however, in response to burn treatments (P < 0.05), with C. virginianum burned in June contributing more N than unburned individuals of that species and individuals of T. virginiana burned in June contributing the least legume N for that species.
Controls on NFUE
Dominant legumes native to longleaf pine savannas vary widely in their ability to fix N2. For the three native legumes in this study, fire regime did not predictably alter fixed N2 use efficiency. Contrary to prediction, carbon supply did not limit N2 fixation in large perennial legumes, except at very small sizes. Fire season did, however, influence NFUE rates, apparently through changes in C allocation, but the response was species-specific, depending on growth and reproductive status. Despite the differences in species response to fire regime, general patterns emerged that may explain the relationships observed between reproduction, growth, and NFUE of native legumes in longleaf pine savannas.
For both C. virginianum and T. virginiana, this study revealed an asymptotic relationship between NFUE and above-ground plant biomass. At small sizes, fixed N2 use efficiency increased rapidly with increased growth. This relationship quickly reached a threshold of NFUE, beyond which these processes became decoupled. The significant relationship with above-ground biomass suggests that NFUE may be limited by C supply when canopies are small and/or growth rates are high. Although NFUE decreased curvilinearly with increased flower production in C. virginianum (Fig. 3c), the clipping experiment revealed no differences in biomass or NFUE in plants that were not allowed to reproduce. Moreover, individuals consistently derived > 80% of their N from N2 fixation. These results suggest that there is no trade-off between fixed NFUE and reproduction for C. virginianum.
For T. virginiana, however, data suggest complex trade-offs with respect to C allocation following fires in different seasons. Fire-stimulated flowering in T. virginiana occurred in May following March burns (Hiers et al., 2000). Total number of flowers showed a strong positive relationship to NFUE with no apparent relationship to above-ground biomass. This pattern is supported by the clipping experiment, which detected significant increases in tissue N and NFUE in flowering individuals. It appears that reproduction in T. virginiana is N-limited, rather than C-limited, as flower production increased with higher N2 fixation. In some perennials, reproductive structures are self-supporting in terms of C demand (Thompson & Stewart, 1981), potentially explaining how T. virginiana can simultaneously invest in greater N2 fixation and reproduction. Moreover, leaves of reproductive plants may have enhanced photosynthetic rates (Reekie & Bazzaz, 1987), and retranslocation of nutrients from reproductive structures may defray the costs of reproduction (Ashman, 1994). Retranslocation may also account for the significantly higher tissue N that was observed following March burns.
Results from the clipping experiment showed that simultaneous increases in N2-fixation and reproduction were accompanied by significant decreases in fine root biomass. In June burns, correlations pointed to a significant role of roots in C allocation among plant functions. Following growing season fires, flowering is delayed until mid-summer, and flowering response of T. virginiana at this time can vary dramatically from year-to-year (Hiers et al., 2000). It is unclear from this study whether this relationship represents the supply of stored C by roots or increased surface area for water uptake. The importance of below-ground biomass to reproduction following June fires may also be related to observed patterns of biomass allocation in unburned individuals. Unburned T. virginiana plants showed limited investment in reproduction, but they significantly increased root:shoot ratios by allocating greater biomass below-ground. Increased investment in storage during no-burn years has been observed in species that resprout following fire (Pate et al., 1990).
Little work has been done on herbaceous, perennial legumes; thus, hypothesized trade-offs between reproduction and N2-fixation have been based primarily on annual crop species. Annuals maximize reproductive effort by allocating a greater proportion of their net assimilation of CO2 to reproduction (Harper, 1977). Rhychonsia reniformis is thought to be a short-lived perennial and may frequently recruit new individuals from seeds (J. K. Hiers, pers. obs.). High mortality of R. reniformis in this study prevented more extensive analysis of trade-offs among plant functions, but uniformly low rates of N2 fixation were observed throughout the study for this species.
The effects of fire regime on N2 fixation patterns were species-specific. Nevertheless, we suggest that at least two general patterns emerge that offer functional explanations for patterns of N2 fixation in herbaceous legumes. If increased N2 fixation is associated with decreased growth rates, then legumes will likely pursue a strategy of limited N2 fixation. Alternatively, if greater N2 fixation is related to increased growth rates, then such legumes may be likely to pursue a strategy similar to that of C. virginianum. C limitations to N2 fixation are likely to occur only in small individuals or during periods of rapid growth. Reproduction can influence N2 fixation but, as in the case of T. virginiana, the relationship may often be positive. This is opposite to patterns indicative of C trade-offs between these functions, which are documented in annual legumes.
Although it appears likely that species of long-lived perennial legumes are capable of maintaining high NFUE rates and play a significant role in maintaining N in longleaf pine ecosystems, the three dominant species varied significantly in NFUE rates. Additional studies are needed to assess N2 fixation rates of other common legume species and to examine the effects of environmental constraints, such as light, soil moisture, and soil fertility (N and P) on N2 fixation rates. This study represents an important first step in understanding the role of legumes in replenishing N in frequently burned longleaf pine savannas. To understand the role of legumes in the maintenance of N in longleaf pine ecosystems more fully, future studies must address the interactions among fire regimes on above-ground vs below-ground processes of tissue turnover and decomposition.
We thank Brian Van Eerden, John McGuire, Carlos Wilson, Stephanie Davis, Tom Hay, and the staff at the Joseph W. Jones Ecological Research Center. Funding for this research was provided by the Woodruff Foundation and the Joseph W. Jones Ecological Research Center.