270801.qxd

Oecologia (2002) 130:297–308DOI 10.1007/s004420100801 Heather Erickson · Eric A. Davidson · Michael Keller
Former land-use and tree species affect nitrogen oxide emissions from a tropical dry forest Received: 19 January 2001 / Accepted: 1 August 2001 / Published online: 28 September 2001 Springer-Verlag 2001 Abstract Species composition in successional dry for-
tively and exponentially related to litter C/N ratio for ests in the tropics varies widely, but the effect of this these dry forests and the relationship was upheld with variation on biogeochemical processes is not well the addition of data from seven wet forests in northeast- known. We examined fluxes of N oxides (nitrous and ni- ern Puerto Rico. This finding suggests that species deter- tric oxide), soil N cycling, and litter chemistry (C/N ra- mination of litter C/N ratio may partly determine N ox- tio) in four successional dry forests on similar soils in ide fluxes across widely differing tropical environments.
western Puerto Rico with differing species compositionsand land-use histories. Forests patch-cut for charcoal Keywords Dry tropical forest · Nitrogen mineralization ·
60 years ago had few legumes, high litter C/N ratios, low Nitric oxide · Nitrous oxide · Tropical legumes soil nitrate and low N oxide fluxes. In contrast, succes-sional forests from pastures abandoned several decadesago had high legume densities, low litter C/N ratios, high mean soil nitrate concentrations and high N oxide fluxes.
These post-pasture forests were dominated by the natu- Ecosystem ecologists have long known that plant species ralized legume Leuceana leucocephala, which was likely can have different effects on soils and soil processes responsible for the rapid N cycling in those forests. We (Zinke 1962). Recently, specific effects of different plant conclude that agriculturally induced successional path- species on ecosystem processes, especially those involv- ways leading to dominance by a legume serve as a mech- ing the cycling of nitrogen (N) have been explored in de- anism for increasing N oxide emissions from tropical re- tail (e.g., Pastor et al. 1984; Wedin and Tilman 1990; gions. As expected for dry regions, nitric oxide dominat- Finzi et al. 1998; García-Montiel and Binkley 1998).
ed total N oxide emissions. Nitric oxide emissions in- One of the key ways plants modify their soil environ- creased with increasing soil moisture up to about 30% ment is through litterfall and its subsequent effects on water-filled pore space then stabilized, while nitrous ox- soil processes (Pastor et al. 1984; Wedin and Pastor ide emissions, albeit low, continued to increase with in- 1993; Prescott et al. 2000). Litter high in N is associated creasing soil wetness. Inorganic N pools and net N min- with fast rates of decomposition (Melillo et al. 1982; eralization were greatest during peak rainfalls and at the Taylor et al. 1989) and high soil N turnover (Pastor et al.
post-agricultural site with the highest fluxes. Soil nitrate 1984; Hobbie 1992; Reich et al. 1997).
and the nitrate/ammonium ratio correlated positively High rates of soil N cycling, particularly in tropical with average N oxide fluxes. N oxide fluxes were nega- soils, result in increased emission of the globally im-portant N oxide gases, nitric (NO) and nitrous (N O) ox- ide (Matson and Vitousek 1990; Davidson et al. 2000; Erickson et al. 2001). NO reacts in the atmosphere to Universidad Metropolitana, P.O. Box 21150, form nitric acid, an important component of acid deposi- tion; NO is also central in the photochemical production e-mail: [email protected]: +1-787-7515386 of ozone in the lower atmosphere (Williams et al. 1992).
N O is a greenhouse gas with about 200 times the warm- ing potential as CO and destroys ozone in the strato- sphere (Prather et al. 1995). However the exact sources ofthese gases are still largely unknown (Bouwman et al.
1995; Davidson and Kingerlee 1997), and large variations USDA-FS, International Institute of Tropical Forestry, Río Piedras, PR 00928–5000, USA in N oxide emissions often exist within similar soil types.
Variation in soil N availability may explain some of high degree of coppicing currently in the forest (Murphy and Lugo the wide variations in emissions (Matson and Vitousek 1986b). Several larger-sized fragments were entirely cleared in thelate 1940s primarily for pasturing (Vélez 1996).
1990; Davidson et al. 2000; Erickson et al. 2001). Legu- We chose two stands that were “closed forest” in 1936 aerial minous tree species are often common in tropical forests; photos and two younger stands that had been cleared for agricul- the Leguminosae may be the most prevalent plant family ture by 1950. The two older stands were never cleared enough in in continental tropical dry forests (Gentry 1995). Tropi- 1936 or on subsequent photos (1950, 1963, 1971, 1983, 1989) to cal legumes typically produce litter with high N (Sprent be designated open forest, although they were among the areasthat were patch-cut for charcoal. These sites were at least 70 years et al. 1996), leading to high soil N availability and/or old by the time our study began in 1995, assuming 10 years for rates of soil N cycling (García-Montiel and Binkley canopy closure if they had been completely cut prior to 1936. The 1998; Erickson et al. 2001). Thus legumes may be im- younger sites were closed forests by 1963, making them approxi- portant drivers of soil N dynamics. Despite this general mately 45 years old at the time of sampling. Interviews with a sur-viving member of a family who farmed one of the cleared areas understanding, few studies of N oxide fluxes have con- indicated that subsistence crops, primarily corn, squash and peas sidered taxonomic structure of the plant community as a were grown and fertilizer was never used (Miguel Canals, person- al communication, biologist, Puerto Rico Department of Natural Tropical and subtropical dry forests were globally Resources and the Environment). We designate the older forestsOF1 and OF2, and the younger, post-agricultural successional for- once quite extensive (~24 million km2) (Murphy and ests, SF1 and SF2, cognizant that all sites had seen previous hu- Lugo 1990). Due to clearing and conversion to agricul- man activity. OF1 was contained within a larger area that has been ture, pastures and post-agricultural forests now cover sampled extensively for ecosystem studies of dry forest (Lugo and most of the area formerly in dry forest (Murphy and Murphy 1986; Murphy and Lugo 1986a, b). A ground fire burned Lugo 1995). Successional forests likely differ in tree through SF2 as late as 1980 (Miguel Canals, personal communica-tion).
composition from original forests, yet because floristicstudies are much less common in successional than inundisturbed dry forest (Murphy and Lugo 1986a) this is largely undocumented. If post-agricultural changes inforest species, particularly increased dominance by le- Additional details for the methods summarized below are found inErickson et al. (2001). Within each site, eight chamber points were gumes, lead to increases in soil N availability, N oxide located as follows. A 30-m transect was run in the direction of emissions may also increase. For example, among sever- maximum slope; at four equidistant intervals a sampling point was al tropical ecosystems in Africa, successional Acacia randomly located on each side and not exceeding 7.5 m of the forests with a history of anthropogenic disturbance had main line. PVC chamber bases (25 cm diameter) were permanent- the highest NO fluxes (Serça et al. 1998).
Gas fluxes were measured four times at each site over a We report results of a study examining relationships 10-month period from December 1995 to October 1996. Air sam- among tree species, previous land-use, litterfall C/N ra- ples for N O analyses were removed from the vented covered tios, soil N cycling and N oxide gas emissions in a dry chambers immediately, and at 10, 20 and 30 min after chamber tropical forest region of southwest Puerto Rico. We ask closure. Gases were analyzed using electron capture detection gaschromatography (Schimadzu GC-14A) within 24 h. N O fluxes whether two post-agricultural forests differ composition- were calculated from the linear regression of concentration versus ally from two forests not previously under agriculture time using the four data points from each chamber. NO was mea- within the same area (Guánica Commonwealth Forest).
sured in the field using flow-through chambers and a portable We then examine how NO and N O emissions contrast LMA3 Scintrex nitrogen oxide analyzer. Briefly, NO is converted to NO and detected by chemiluminescence. NO flux was calcu- among the forests and relate to litter chemistry, soil lated from at least a 3-min linear increase in concentration, ob- tained within the first 8 min of sampling. Using our system, NO2 was not differentiated from NO. NO emissions from soils are usually negligible (Conrad et al. 1990; Davidson et al. 1991). Dueto time constraints, 4–6 of the 8 chambers were sampled each time at a site; all sites were sampled over a 2-day period.
Air temperature, soil temperature (to 2 cm depth), and soils (see below) were sampled within 1 m of the chambers simulta-neously with gas sampling. Rainfall data were obtained from a sta- Four tropical dry forest stands located 3–8 km from the Caribbean tion within the Guánica Forest located approximately mid-way be- Ocean within the 4,000 ha Guánica Commonwealth Forest tween the sites. Hurricane Hortense passed through the area in (17°47′N, 65°52′W) in southwestern Puerto Rico were used for September 1996, causing moderate damage to the canopy.
this study. The climate is hot (MAT=25°C); a dry season typicallyoccurs between January and May. Annual precipitation averages860 mm (Murphy and Lugo 1986b), with high intra-annual vari- ability. Soils are typic Calciustolls in the Aquilita series (Lugo-López and Rivera 1976). Nearly 25% of the ground surface is ex- Soils were sampled adjacent to each chamber to 10 cm depth with posed limestone. Approximately one-third of the tree species with- an 8-cm-diameter corer. Gravimetric moisture was determined af- in Guánica are deciduous or semi-deciduous, the remainder being ter drying a subsample at 105°C for at least 48 h. Soil pH (water) broad-leaved evergreens (Murphy and Lugo 1995). The Guánica was measured on samples collected in July 1996.
Forest has been protected from human disturbance since the Bulk density was measured once at each sampling station by 1950s. Prior to the 1940s, tree-sized patches were commonly cut excavating soil from rectangular holes (approximately 10×10× (herein patch-cut) for charcoal. This is evident in the historical 10 cm) located within 1 m of the chambers. Removed soil was aerial photos (1936) of the forest showing many dispersed small sieved to 2 mm, dried and weighed. Hole volumes were measured canopy openings covering nearly 5% of the total area, and by the by filling the plastic-lined hole with water to a known volume, and corrected for removed rock volume. Water-filled pore space (WFPS) was then calculated as the proportion of gravimetricallydetermined water volume to total pore volume (Hillel 1980).
The legume Leuceana leucocephala had the greatest On three of the four sampling dates (December 1995, March 1996 basal area and stem density of any tree species in SF1 and October 1996), we measured inorganic N, net N mineraliza- and SF2, the two post-agricultural successional forests tion and net nitrification on soils. Nitrification potential was mea- (Table 1). Legume basal areas, including Acacia at SF2, sured on one date, December 1995. Inorganic N was extracted represented 29% and 66% of the total basal areas and with 2 M KCl. We used a 7-day aerobic laboratory incubation todetermine net N mineralization, following Hart et al. (1994) with 50% and 58% of the individual stems for SF1 and SF2, modifications detailed in Erickson et al. (2001). Nitrification po- respectively. In contrast, in the more floristically diverse tential was estimated from the change in NO –-N concentration and dense older forests, legumes were either much less over a 24-h period (2, 4, 18 and 24 h) in a shaken soil-slurry with dominant (OF2) or not present (OF1) (Table 1). Cactac- excess NH +-N, NO –-N and NH +-N were determined colorimet- eae genera Cereus and Pilosocereus are also among the rically on an Alpkem Rapid Flow Analyzer.
Total C and N were measured by combustion on a LECO dominants in the successional forests, but were absent CNS-2000 on a subsample of soils collected in July 1996. Air- from our plots in the older forests. The OF sites had a dried soils were ground to pass a 20-mesh sieve. The sample for greater number of stems, most with diameters less than total C was rinsed with sufficient HCl prior to analysis to remove 10 cm (data not shown), when compared to the post-agri- all carbonates; N was analyzed on untreated samples. An addition-al subsample was dried at 105°C to correct results for soil mois- cultural successional forests and this is most likely due to stump sprouting from trees cut for charcoal production(Murphy and Lugo 1995).
A nested circular plot technique was used to sample forest vegeta- Aboveground litterfall inputs and chemistry tion. We recorded diameter at breast height (dbh) and speciesidentification for all trees equal to or greater than 1 cm dbh within Daily rates of litterfall ranged from 0.5 to 3.0 g m–2 (data a 500- m2 circular plot encompassing our gas sampling chambers.
not shown) during most months except immediately after Within a concentric 1,000-m2 circular plot all trees larger than Hurricane Hortense, when daily rates at all sites exceed- 10 cm dbh were recorded. Nomenclature follows Liogier andMartorell (2000).
ed 7.0 g m–2. On average, aboveground total litterfall We collected litterfall from March 1996 to March 1997. Nine was 30% greater in the older forests (OF1 and OF2) than litterfall traps (0.25 m2) were spaced 40 m apart in an 80×80 m in the post-agricultural successional forests (SF1 and grid encompassing the gas sampling areas. Litter was collected SF2), although due to high within-site variation, differ- monthly, oven-dried at 60°C, sorted into leaves and other classesof litter and weighed to the nearest 0.01 g. For each site/date com- ences among sites were not statistically significant bination, leaves from the nine baskets were pooled into a single (Table 2, P>0.05). Leaf litterfall was also 30% lower in sample. Subsamples were removed, ground to pass a 20-mesh the younger compared with older forests (Table 2), but sieve, and analyzed for total C and N by combustion on a LECO CNS-2000. For all analyses, results are expressed on an oven-dry(soil or litter) basis.
Leaf litter C/N ratio differed significantly among all four sites (Table 2), ranging from a low of 22.8 at SF1 toa high of 49.0 at OF1. Leaf litter N concentrations dif- fered in a like manner among the sites, ranging from ahigh of 2.23% at SF1 to a low of 1.05% at OF1. The A repeated measures ANOVA was used to compare the effects ofsite, sampling date and their interactions on N O flux and soils da- concentration at OF1 compares favorably with a 1.01% ta. Differences in NO flux due to the same effects were assessed leaf litter N reported from the well-studied area encom- with a two-way, rather than a repeated measures ANOVA because passing our plot (Lugo and Murphy 1986). Total leaf lit- different chambers were sampled for NO on different dates. We ter N inputs were, on average, greater for successional used a univariate approach for the repeated measures analysis(SAS 1987). Gas flux and most soil variables were log-trans- (mean=4.97 g N m–2 year–1) than for older forests formed prior to analysis to meet the assumptions of ANOVA, ex- (mean=4.23 g N m–2 year–1) though N inputs from OF2 cept for percentage data, which were arcsin-transformed (Sokal could not be statistically distinguished from either suc- and Rohlf 1995). We assessed differences among sites within a date using a one-way ANOVA followed by a Tukey’s multiplecomparison. Where a parameter was measured only once (e.g.,soil pH and nitrification potential) only differences among siteswere evaluated. Linear and exponential models were used to ex- amine relationships between total N oxide fluxes and N availabili-ty indices; in each case, the model with the higher R2 value is re- Soil pH values for the upper 10 cm of mineral soil were lowest at SF1 (mean=6.6) compared to the other threesites where means ranged from 7.7 to 7.9 (Table 3). Soiltemperature ranged from 25°C to 32°C (data not shown)but showed no consistent pattern of differences amongthe sites. Bulk density was higher in the successional ) in the two post-agricultural successional forests ( (cotinued)
1
Table 2 Annual means (SE in parentheses) for total aboveground
for each basket at a site (n=9). Means for C/N ratio and N (%) are (AG) litterfall, total leaf litterfall, and leaf litter N chemistry for two based on 14 different collections (n=14). Different superscripts with- mid-successional and two older tropical dry forests in Guánica For- in columns indicate significant differences (P< 0.05) determined by est, P.R. Means for litterfall are based on annual estimates calculated an a posteriori Tukey multiple comparisons test after ANOVA Table 3 Selected site and soil
(0–10 cm) characteristics for
ues are means (SE in parenthe-ses, n=8) except for porosity, significant differences (P<0.05)determined by an a posteriori forests (mean=1.00 g cm–3) versus the older forests mineralization, ranging from about 3.5 to 8.8 µg N g (0.85 g cm–3). Although SF2, the successional forest that soil–1 7 day–1, was usually similar among the sites burned 16 years earlier, had nearly 25% lower total soil (Fig. 1d). The one exception was during March 1996 C and N in the upper 10 cm than the other three sites when the highest rate measured for any site ( ~21 µg N g (Table 3), the difference was not significant (P>0.05).
soil–1 7 day–1) was measured for SF1. Nitrification po- Soil C/N ratios ranged from 10.0 to 10.2 and did not tential was greatest at SF1 (2.77 mg N kg–1 h–1) but vary among the sites (P>0.05).
overlapped with rates from SF2 and OF2 (Table 5).
Inorganic N pools and soil N transformations For soil extractable NO –-N and NH +-N and net nitrifi- Site, sampling date and their interaction influenced the cation, the effects of sampling date, site and date × site fluxes of NO (P<0.001). The effect of site depended on interactions were statistically significant (Table 4). Soil date. SF1 had the highest NO fluxes on all four dates, NO –-N ranged widely, from ~7 to 48 µg N g soil–1, with though its fluxes could not be distinguished from SF2’s the largest concentrations measured in March 1996 in December 1995 (Fig. 2). The two highest mean NO (Fig. 1, Table 4). Similarly, extractable NH +-N showed fluxes (>23.0 ng N cm–2 h–1) were measured in March large variation, ranging from ~4 to 52 µg N g soil–1, with 1996 and June 1996. Mean NO fluxes from the OF sites the largest pools also measured in March 1996 (Fig. 1).
were generally low (<2 ng N cm–2 h–1), and differed from Soil NO –-N pools at SF1 were 1.7–4 times greater than at the other sites. NH +-N pools tended to be similar SF1 had higher N O fluxes in March 1996 (P<0.05) among the sites, except during March 1996 when pools and June 1996 (P=0.07) than the other sites (Fig. 2, were high at all sites (Fig. 1). Then, NH +-N pools at Table 4). Fluxes from the other sites did not differ OF1 were twice those of the other sites (Fig. 1). Net ni- during any of the four sampling dates. Only two mean trification was highest for all sites in March 1996 N O fluxes were greater than 0.50 ng N cm–2 h–1 (3.25 and 0.57, measured during March 1996 for SF1 and The effect of sampling date on net N mineralization SF2 respectively). While average fluxes of N O (n=8) depended on site (interaction P=0.0017, Table 4). Net N were negative at three of the sites, indicating N O up- Table 4 Repeated measures ANOVA results for soil NO –-N, NH +-N, net N mineralization, net nitrification and N O emissions
Table 5 Mean inorganic N, soil N cycling rates and N oxide flux-
sampling point (n=8), except for nitrification potential which was es (SE in parentheses) for the four study areas within the Guánica Forest Preserve. Means are based on means over time for each Fig. 1 KCl (2 M) extractable
NO –-N (a) and NH +-N (b)
and net rates of nitrification
(c) and N mineralization (d)
for the four sites at Guánica.
OF “Older forest”, SF “post-
agricultural successional for-
est”. Data are means (n=8) and
standard errors. For each vari-
able, different lower-case let-
ters within a given sampling
date indicate significantly dif-
ferent means (P< 0.05) using a
Tukey multiple comparisons
test after one-way ANOVA.
Statistical analyses were done
on log-transformed data; means
and standard errors are based
on non-transformed data to fa-
cilitate comparison with other
studies
take, these means were not significantly different from (ANOVA, P<0.05 for the main effects; the effect for in- teraction was NS). March 1996 was the wettest month;during which time WFPS nearly doubled compared tothe other months (Fig. 2C). Overall, WFPS was about Rainfall, soil moisture and N oxide emissions 25% lower at OF1 compared to the other sites (Fig. 2C).
The highest fluxes of NO for three of the sites (OF1, We sampled N oxide gases during both wet and dry peri- OF2 and SF2) coincided with the high March WFPS val- ods (Fig. 3). WFPS ranged from a low of just over 16% ues (Fig. 4a). For SF1, the highest fluxes (22–25 ng N to a high of 55%, varying by site and date of sampling cm–2 h–1) occurred during the relatively wet months of Fig. 4 Fluxes of NO (a) and N O (b) versus % WFPS for each of
the four sites at the Guánica Forest, P.R. Note differences in scalesbetween the two panels. Site codes are explained in Fig. 1 Fig. 2 Fluxes of NO (a), N O (b), and percent WFPS measured
from the four sites at Guánica. Within a given date, different low-er-case letters indicate significantly different means (P< 0.05) us-ing a Tukey multiple comparisons test after one-way ANOVA.
Note differences in scales in all three panels. Site codes are ex-plained in Fig. 1 Fig. 5 The sum of N oxide fluxes (N O + NO) versus a the ratio
of NO –-N/NH +-N and b leaf litter C/N ratio. For b, two separate
curves are plotted for the humid and dry forests; the equation andmodel R2 value are from the combined data sets. For fluxes and in-organic N pools, each data point is the mean of the 8 samplingpoints for a site: fluxes were measured 4 times in the dry forestand 12 times in the humid forest sites; inorganic N pools were Fig. 3 Daily rainfall and dates of N oxide sampling (arrows) at
measured 3 times in the dry forest. Litterfall C/N ratios are based Guánica from 12 January 1995 to 12 January 1996 on means of 14 collections made at each site March and June when WFPS was 25–55%. For the two (>40%; Fig. 4b). N O:NO ratios, calculated only if mean sites with the highest fluxes, SF1 and SF2, NO fluxes N O fluxes were significantly different from zero, showed no significant increase with WFPS values be- ranged from 0.105 to 0.140, indicating the predominance The fluxes of N O for three of the sites (SF1, SF2, and OF2) peaked when WFPS was greatest for each site Lower N oxide fluxes from successional versus pri- mary forests had been widely reported (Robertson and Across all sites the sum of the N oxide fluxes was posi- Tiedje 1988; Keller and Reiners 1994; Verchot et al.
tively related to mean soil NO –-N, and the ratio of 1999), but without detailed information on species com- NO –-N/NH +-N and negatively related to leaf litter C/N position. In humid northeastern Puerto Rico, the greatest ratio (P<0.05; Fig. 5). Average N oxide fluxes were not N oxide fluxes (primarily N O) were from a succession- related to averages of net N mineralization, net nitrifica- al forest that had more legumes than other nearby suc- tion, nitrification potential or to NH +-N. Exponential cessional and old growth forests (Erickson et al. 2001).
model fits had higher R2 values than linear models for Similarly, successional Acacia forests had greater NO NO –-N/NH +-N and to leaf litter C/N ratio (Fig. 5). The fluxes compared with other non-legume forests in relationship between N oxide fluxes and leaf litter C/N Africa (Serça et al. 1998). These findings, together with ratio for the four Guánica sites was remarkably similar to our Guánica results, suggest that some successional for- that obtained for seven forests in the humid NE (Fig. 5 ests may contribute more N oxides to the atmosphere than previously thought (cf., Robertson and Tiedje1988). While rarely documented, species composition,especially the abundance of N-fixing genera, may help explain the variation in N oxide fluxes among tropicalforests.
Forest species, feedback and N oxide fluxes Despite lower leaf litterfall, the L. leucocephala suc- cessional forests at Guánica had, on average, greater in- Land-use and land-use history are known to affect flux- puts of leaf litter N than the older forests due to greater es of important N oxide gases from the tropics. Despite leaf litter N concentrations. High foliar N in N-fixing this understanding, estimates of terrestrial sources of species is common, and N in legume foliage may be high these gases remain uncertain (Bouwman et al. 1995; in both nodulated and non-nodulated genera (Sprent etDavidson and Kingerlee 1997). Forest clearing (Luizão al. 1996). Nonetheless, symbiotic N fixation by L. leuc- et al. 1989; Keller et al. 1993), burning (Levine et al.
ocephala appears to be widespread (Högberg and 1988; Neff et al. 1995; Serça et al. 1998) and fertiliza- Kvarnström 1982; Parrotta et al. 1994, 1996; Sprent et tion of agricultural lands (Matson et al. 1996; Mosier al. 1996). One of the successional forests, SF1, had and Delgado 1997; Veldkamp and Keller 1997) have greater N oxide fluxes and foliar N concentrations than been identified as causes of increasing N O or NO emis- the previously burned successional forest, SF2, though sions from tropical regions. In the dry forests of both forests had similar densities (50% and 58% respec-Guánica, total N oxide fluxes were greater from le- tively) of legumes. Ground fire can increase inorganic N gume- (mostly Leuceana leucocephala) dominated post- pools (cf., Ellingson et al. 2000) which, in turn, may de- agricultural successional forests than from the nearly le- crease N fixation (Marschner 1995), providing a possible gume-free non-agricultural older forests (Table 5).
explanation for the discrepancy in N oxide fluxes be- These findings indicate that agricultural abandonment leading to successional trajectories favoring legumes Litterfall C/N ratios often relate negatively to decom- can significantly affect N oxide fluxes.
position rates (Mellilo et al. 1982) and to net N mineral- The low level of canopy removal and subsequent ization (Pastor et al. 1984; García-Montiel and Binkley stump sprouting at the OF sites likely maintained the 1998; Erickson et al. 2001). SF1 had the lowest leaf litter original species complement (Murphy and Lugo 1995).
C/N ratio, the highest rates of net N mineralization, and Nearly complete canopy removal and fewer sprouts in the highest combined N oxide fluxes. Although one of the former agricultural sites suggest that invasion by new the older forests, OF2, had high total inputs of litterfall species was the mechanism for forest recovery. L. leuc- N (this forest also had minimal representation by le- ocephala, the dominant species at the successional sites, gumes), the litter was, on average, of higher C/N and N is known to invade pastures and recently disturbed areas oxide fluxes were low. Thus for the Guánica sites litter elsewhere in Guánica (Farnsworth 1993; Molina and chemistry, or C/N ratio, relates more directly to soil pro- Lugo, unpublished data). Our floristic data show that L. cesses than does the quantity of N from litterfall.
leucocephala is minimally present in the less disturbed Despite greater inputs of N in the successional for- forests, and likely became dominant in the post-agricul- ests, total pools of soil N did not differ among the sites.
tural successional forests after agricultural abandonment.
While not statistically significant, the most recently L. leucocephala has been naturalized in Puerto Rico and burned successional site tended to have less soil C and N the Caribbean (F. Wadsworth, personal communication) and is considered relatively short-lived. Moreover, it is The high bulk density in the post-agricultural sites is widely distributed elsewhere in the neotropics, where it most likely due to compaction by cattle (Rhoades et al.
may also be important in post-agricultural succession. If 1999). High bulk density leads to increased WFPS, abandonment of additional agricultural lands leads to which may affect N oxide fluxes, although among these forests dominated by L. leucocephala, N oxide emissions sites N availability appears to be a more important driver N oxide emissions in drought-deciduous forests: dict N oxide fluxes, the 2.23% N for SF1 is among the highest reported for any tropical forest (Jaramillo andSanford 1995), suggesting our flux estimates are reason- The positive relationship between N oxide fluxes and various measures of soil N availability is well document- NO is absorbed by forest canopies. Therefore, NO ed (Matson and Vitousek 1987, 1990; Keller and Reiners emissions from soils may be particularly important from 1994; Verchot et al. 1999; Davidson et al. 2000; Erickson drought deciduous forests because absorption is presum- et al. 2001). Many indices of N availability (e.g., soil ni- ably less during periods of low leaf mass (Matson and trate, net N mineralization and net nitrification) have Vitousek 1995). Even the relatively low mean NO flux been evaluated, and the strength and character of their from the OF sites at Guánica (0.62 kg ha–1 year–1) was relationships to N oxide fluxes vary somewhat among from 2 to 100 times greater than annual NO fluxes from sites. For example we found positive relationships with sites in humid northeastern Puerto Rico (Erickson et al.
soil nitrate, the NO –-N/NH +-N ratio and leaf litter C/N 2001). Only the site most dominated by legumes from ratio (Fig. 5). High soil nitrate relative to ammonium in- that study (a successional forest) matched the mean NO dicates excess N availability relative to plant demand (Keller and Reiners 1994; Neill et al. 1997; Davidson et Because the relative proportion of N O versus NO al. 2000; Erickson et al. 2001). Erickson et al. (2001) re- fluxes is related to soil moisture (Davidson 1991), higher ported an exponential decline in N oxide fluxes with in- fluxes of NO would be expected in seasonally dry cli- creasing leaf litter C/N for forests in humid eastern Puer- mates. Rainfall in northeastern Puerto Rico is nearly five to Rico; adding the results from this study resulted in a times the average rainfall at Guánica. N O:NO ratios remarkably similar relationship (Fig. 5). Erickson et al.
were >1 from the sites in northeastern Puerto Rico (2001) also suggested that as plant and aerobic heterotro- (Erickson et al. 2001) and <1 from the sites at Guánica.
phic microbial sinks for N became satisfied, the propor- Based on published annual fluxes (Davidson et al. 1991; tion of inorganic N available for nitrification and denitri- García-Méndez et al. 1991) the ratios for a dry forest at fication, the two processes responsible for N oxide pro- Chamela, Mexico, were also generally at or below 1.
duction, would increase, resulting in the non-linear rela- These data support the idea that regional differences in tionship. Although an exponential relationship described relative N oxide emissions may be accounted for by the relationship between total N oxide flux and the ratio variation in precipitation and effects on soil moisture of NO –-N/NH +-N for the dry forest sites (Fig. 5a), a (Davidson 1991; Davidson et al. 2000). Davidson (1991) linear model better described the relationship for the wet also proposed a unimodal relationship between soil forest sites (data not shown). Leaf litter C/N ratio ap- moisture and N oxide gas emissions, in which the opti- pears to be a more consistent predictor of total N oxide mal water content for maximum emissions occurs at in- fluxes than soil inorganic-N in the wet and dry tropical termediate values and emissions decline as moisture environments encompassed in these studies.
increases or decreases. Thus, at dry sites, higher NO The mean NO flux for SF1 (11.9 kg N ha–1 year–1) fluxes are often found during the wet season (this study, represents, to our knowledge, the largest mean soil-at- Davidson et al. 1991; Serça et al. 1998) whereas, at more mosphere NO flux measured for a tropical forest. In their humid sites, higher fluxes are found during the dry sea- review, Davidson and Kingerlee (1997) list the highest son (e.g., Verchot et al. 1999). At Guánica, NO increases mean or median NO flux as 4.3 kg ha–1 year–1 from an at lower percent WFPS, while at higher WFPS, NO lev- Amazonian tropical evergreen forest (Kaplan et al. 1988) els out, and N O clearly increases (Fig. 4), suggesting and 0.7 kg ha–1 year–1 from a tropical deciduous forest of the optimal water content for NO emissions had been Mexico (Davidson et al. 1991). (The mean flux from the Guánica old forest sites – 0.62 kg ha–1 year–1 – is re- The exact effect of moisture on N oxide fluxes ap- markably similar to the Mexican dry forest estimate.) We pears constrained primarily by N availability. The high- averaged the four time measurements of NO fluxes for est N oxide fluxes occurred in March 1996 when soils comparative purposes but recognize the limitation in ex- were the wettest (Fig. 2), inorganic N pools and net nitri- trapolating to annual fluxes. The March 1996 measure- fication highest (Fig. 1), and coincided with a rainfall ments were made immediately after a rainfall event, and event that had been preceded by a long dry period likely represent extremely high fluxes. Experimental wet- (Fig. 3). Generally higher rates of net N mineralization ting has been shown to stimulate NO fluxes (Davidson during wet seasons have been documented for other 1991. 1993; Poth et al. 1995; Levine et al. 1996), espe- tropical dry or seasonal forests (García-Méndez 1991; cially after long periods of drought. June fluxes, howev- Babbar and Zak 1994; Rhoades et al. 1999). Yet despite er, made 8 days after measurable precipitation, were also WFPS above 35% at all of the sites in March 1996 (see high. We also measured the fluxes from areas not cov- also Fig. 4), fluxes from the two older forests, with lower ered by rock. Reducing the mean fluxes from SF1 and soil NO – pools, were not nearly as high as fluxes from SF2 by 25% to reflect the proportion of the ground cov- the successional forests, with greater soil NO – pools.
ered by rock, we still estimate a relatively high mean NOflux of 5.4 kg N ha–1 year–1 for these legume-dominatedforests. Furthermore, if leaf N proves to adequately pre- Davidson EA, Vitousek PM, Matson PA, Riley R, Garcia-Mendez G, Maas JM (1991) Soil emissions of nitric oxide in a season-ally dry tropical forest of Mexico. J Geophys Res 96:15439– In our study, a legume capable of high rates of symbiotic N fixation (L. leucocephala) dominated the post-agricul- Davidson EA, Matson PA, Vitousek PM, Riley R, Dunkin K, tural successional forests after nearly 25 years or more García-Mendez G, Maas JM (1993) Processes regulating soil since the last known disturbance. These successional for- emissions of NO and N O in a seasonally dry forest. Ecology ests showed substantially greater fluxes of N oxides (pri- Davidson EA, Keller M, Erickson HE, Verchot LV, Veldkamp E marily NO) than two nearby older forests that had never (2000) A cross-site test of a conceptual model of nitrous oxide been cleared for agriculture. The former land-use and and nitric oxide emissions from soils. BioScience 50:667–680 dominance of L. leucocephala in the post-pasture succes- Ellingson LE, Kauffman JB, Cummings DL, Sanford RL Jr, sional forests suggests that when agricultural abandon- Jaramillo DL (2000) Soil N dynamics associated with defores-tation, biomass burning, and pasture conversion in a Mexican ment leads to dominance by legumes, fluxes of NO may tropical dry forest. For Ecol Manage 137:41–51 also increase. Thus dry tropical forest clearing and suc- Erickson HE, Keller M, Davidson E (2001) Nitrogen oxide fluxes cession leading to legumes is identified as a previously and nitrogen cycling during secondary succession and forest fertilization in the humid tropics. Ecosystems 4:67–84 Farnsworth EJ (1993) Ecology of semi-evergreen plant assemblag- The high N oxide fluxes were directly related to es in the Guánica Dry Forest, Puerto Rico. Caribb J Sci 29: low leaf litter C/N and high soil NO –-N/NH +-N ratios.
In fact, the exponential relationship between total N Finzi AC, Van Breeman N, Canham CD (1998) Canopy tree-soiloxide fluxes and leaf litter C/N ratio for all sites was interactions within temperate forests: species effects on soil nearly identical to one obtained for seven forest sites of carbon and nitrogen. Ecol Appl 8:440–446 García-Méndez G, Maass JM, Matson PA, Vitousek PM (1991) varying ages in a wet region of northeastern Puerto Rico Nitrogen transformations and nitrous oxide flux in a tropical (Erickson et al. 2001). This argues for coupling mea- deciduous forest in Mexico. Oecologia 88:362–366 sures of litter chemistry and N oxide gases where possi- García-Montiel DC, Binkley D (1998) Effect of Eucalyptus ble. Litter C/N ratio appears to be a robust predictor of saligna and Albizia falcataria on soil processes and nitrogensupply in Hawaii. Oecologia 113:547–556 N oxide fluxes across very different tropical environ- Gentry AH (1995) Diversity and floristic composition of neotropi- cal dry forests. In: Bullock SH, Mooney HA, Medina E (eds) In all cases in Guánica, NO emission surpassed those Seasonally dry tropical forests. Cambridge University Press, of N O. Rainfall increased N oxide emissions at all sites, Hart SC, Stark JM, Davidson EA, Firestone MK (1994) Nitrogen but not nearly as much in the older forests. This suggests mineralization, immobilization, and nitrification. In: Weaver R that N availability is the primary control of N oxide (ed) Methods of soil analysis, part 2. Microbiological and bio- emission, and moisture a secondary control.
chemical properties, SSSA Book Series, no. 5. American Soci-ety of Agronomy, Madison, pp 985–1019 Acknowledgements We thank María Rivera, Carlos Ortiz, Brynne
Hillel D (1980) Fundamentals of soil physics. Academic Press, Bryan, Carmen Marrero, Mary Jean Sánchez and Edwin López for top-notch assistance in the field and laboratory. Sandra Molina Hobbie SE (1992) Effects of plant species on nutrient cycling.
and Miguel Canals are thanked for assisting with field sites and M.C. also for logistical support, rainfall data and contacting the Högberg P, Kvarnström M (1982) Nitrogen fixation by the woody Gutteriez family for oral history. Jill Thompson, John Pastor and legume Leuceana leucocephala in Tanzania. Plant Soil 66: Fred Scatena provided insightful reviews of earlier drafts. This re- search was supported by the NASA Terrestrial Ecology Program Jaramillo VJ, Sanford RL (1995) Nutrient cycling in tropical de- under NASA grants NAGW-3772 and NAG5–6277. We acknowl- ciduous forests. In: Bullock SH, Mooney HA, Medina E (eds) edge support from the NASA-IRA program to the University of Seasonally dry tropical forests. Cambridge University Press, Puerto Rico, and from the National Science Foundation’s Model Institutions for Excellence program.
Kaplan WA, Wolfsy SC, Keller M, DaCosta JM (1988). Emission of NO and deposition of O3 in a tropical forest system. J Geo-phys Res 93:1389–1395 Keller M, Reiners WA (1994) Soil-atmosphere exchange of ni- trous oxide, nitric oxide, and methane under secondary succes-sion of pasture to forest in the Atlantic lowlands of Costa Babbar LI, Zak DR (1994) Nitrogen cycling in coffee agroecosys- tems: net N mineralization and nitrification in the presence Keller M, Veldkamp E, Weitz AM, Reiners WA (1993) Effect of and absence of shade trees. Agric Ecosyst Environ 48:107–113 pasture age on soil trace-gas emissions from a deforested area Bouwman AF, Van der Hoek KW, Olivier JGJ (1995) Uncertain- ties in the global source distribution of nitrous oxide. J Geo- Levine JS, Cofer WR, Sebacher DI, Winstead EL, Sebacher S and Boston PJ (1988) The effects of fire on biogenic soil emissions Conrad R, Kramer AR, Baumgartner M, Nagele W (1990) Mecha- of nitric oxide and nitrous oxide. Global Biogeochem Cycl 2: nisms of NO exchange between soil and atmosphere. EUR- OTRAC Symposium. SPB Academic, The Hague, pp 123–126 Levine JS, Winstead EL, Parsons DAB, Scholes MC, Scholes RJ, Davidson EA (1991) Fluxes of nitrous oxide and nitric oxide from Cofer WR III, Cahoon DR Jr, Sebacher DI (1996) Biogenic terrestrial ecosystems. In: Rogers JE, Whitman WB (eds) Mi- soil emissions of nitric acid (NO) and nitrous oxide (N O) crobial production and consumption of greenhouse gases: from savannas in South Africa: The impact of wetting and methane, nitrogen oxides and halomethanes. American Society for Microbiology, Washington, D.C. pp 219–235 Liogier, HA, Martorell LF (2000) Flora of Puerto Rico and adja- Davidson EA, Kingerlee W (1997) A global inventory of nitric cent islands: a systematic synopsis, 2nd edn revised. Universi- oxide emissions from soils. Nutr Cycl Agroecosyst 48:91–104 Lugo AE, Murphy PG (1986) Nutrient dynamics of a Puerto Rican Poth M, Anderson IC, Miranda HS, Miranda AC, Riggan PJ subtropical dry forest. J Trop Ecol 2:55–72 (1995) The magnitude and persistence of soil NO, N O, CH Lugo-López MA, Rivera LH (1976) Taxonomic classification of and CO fluxes from burned tropical savanna in Brazil. Global the soils of Puerto Rico, 1975. Bulletin 245. UPR, Agriculture Experiment Station, Río Piedras, P.R.
Prather M, Derwent R, Ehhalt D, Fraser P, Sanhueza E, Zhou X Luizão F, Matson P, Livingston G, Luizão R, Vitousek P (1989) (1995) Other trace gases and atmospheric chemistry. In: Nitrous oxide flux following tropical land clearing. Global Houghton JT, et al (eds) Climate change 1994: radiative forc- ing of climate change and an evaluation of the IPCC IS92 Marschner H (1995) Mineral nutrition of higher plants. Academic emission scenarios. Cambridge University Press, New York, Matson PA, Vitousek PM (1987) Cross-ecosystem comparisons of Prescott CE, Chappell HN, Vesterdal L (2000) Nitrogen turnover soil nitrogen and nitrous oxide flux in tropical ecosystems.
in forest floors of coastal Douglas-fir at sites differing in soil Matson PA, Vitousek PM (1990) Ecosystem approach to a global Reich PB, Grigal DF, Aber JD, Gower ST (1997) Nitrogen miner- nitrous oxide budget. BioScience 40:667–672 alization and productivity in 50 hardwood and conifer stands Matson PA, Vitousek PM (1995) Nitrogen trace gas emissions in a tropical dry forest ecosystem. In: Bullock SH, Mooney HA, Rhoades CC, Coleman DC (1999) Nitrogen mineralization and ni- Medina E (eds) Seasonally dry tropical forests. Cambridge trification following land conversion in montane Ecuador. Soil Matson PA, Billow C, Hall S, Zachariassen J (1996) Fertilization Robertson GP, Tiedje JM (1988) Deforestation alters denitrifica- practices and soil variations control nitrogen oxide emissions tion in a lowland tropical forest. Nature 336:756–759 from tropical sugar cane. J Geophys Res 101:18533–18545 SAS (1987) SAS/STAT guide for personal computers, version 6.
Mellilo JM, Aber JD, Muratore JM (1982) Nitrogen and lignin control of hardwood leaf litter decomposition dynamics. Ecol- Serça D, Delmas R, LeRoux X, Parsons DAB, Scholes MC, Abbadie L, Lensi R, Ronce O, Labroue L (1998) Comparison Mosier AR, Delgado JA (1997) Methane and nitrous oxide fluxes in of nitrogen monoxide emissions from several African tropical grasslands in western Puerto Rico. Chemosphere 35:2059–2082 ecosystems and influence of season and fire. Global Bioge- Murphy PG, Lugo AE (1986a) Ecology of tropical dry forest.
Sokal RR, Rohlf FJ (1995) Biometry, 3rd edn. Freeman, New Murphy PG, Lugo AE (1986b) Structure and biomass of a sub- tropical dry forest in Puerto Rico. Biotropica 18:89–96 Sprent JI, Geoghegan IE, Whitty PW, James EK (1996) Natural Murphy PG, Lugo AE (1990) Dry forests of the tropics and sub- abundance of 15N and 13C in nodulated legumes and other tropics: Guánica forest in context. Acta Cient 4:15–24 plants and the cerrado and neighbouring regions of Brazil.
Murphy PG, Lugo AE (1995) Dry forests of Central America and the Caribbean. In: Bullock SH, Mooney HA, Medina E (eds) Taylor BR, Parkinson D, Parsons WF (1989) Nitrogen and lignin Seasonally dry tropical forests. Cambridge University Press, content as predictors of litter decay rates: a microcosm test.
Neff JC, Keller M, Holland EA, Weitz AW, Veldkamp E (1995) Veldkamp E, Keller M (1997) Nitrogen oxide emissions from a Fluxes of nitric oxide from soils following the clearing and burn- banana plantation in the humid tropics. J Geophys Res 102: ing of a tropical rain forest. J Geophys Res 100:25913–25922 Neill C, Piccolo MC, Cerri CC, Steudler PA, Melillo JM, Brito M Vélez, LL (1996) Land use and land cover maps series of Guánica (1997) Net nitrogen mineralization and net nitrification rates Commonwealth Forest. USDA Forest Service, IITF, Rio Pied- in soils following deforestation for pasture across the south- western Brazilian Amazon Basin landscape. Oecologia 110: Verchot LV, Davidson EA, Cattanio JH, Ackerman IL, Erickson HE, Keller M (1999) Land-use change and biogeochemical Parrotta JA, Baker DD, Fried M (1994) Estimation of nitrogen fix- controls of nitrogen oxide emissions from soils in Eastern Am- ation in Leucaena leucocephala using 15N-enrichment meth- azonia. Global Biogeochem Cycl 13:31–46 odologies. In: Sprent JI, McKey D (eds) Advances in legume Wedin DA, Pastor J (1993) Nitrogen mineralization dynamics in systematics 5: the nitrogen factor. Royal Botanic Gardens, grass monocultures. Oecologia 96:186–192 Wedin DA, Tilman D (1990) Species effects on nitrogen cycling: a Parrotta JA, Baker DD, Fried M (1996) Changes in dinitrogen fix- test with perennial grasses. Oecologia 84:433–441 ation in maturing stands of Casuarina equisetifolia and Leuca- Williams EJ, Hutchinson GL, Fehsenfeld FC (1992) NOx and N O ena leucocephala. Can J For Res 26:1684–1691 emissions from soil. Global Biogeochem Cycl 6:351–388 Pastor J, Aber JD, McClaugherty CA, Mellilo JM (1984) Above- Zinke PJ (1962) The pattern of influence of individual forest trees ground production and N and P cycling along a nitrogen min- eralization gradient on Blackhawk Island, Wisconsin. Ecology65:256–268

Source: http://www.whrc.org/resources/publications/pdf/EricksonetalOecologia.02.pdf