INTRODUCTION
Land-use change has dramatically altered tropical forest landscapes, increasing both forest fragmentation and the amount of land area in edge zones. With this change comes a need to better understand how edges affect forest ecosystems. It is well known that abiotic and biotic conditions at forest edges typically differ from those in forest interiors, and many studies of edge effects have focused on changes in variables such as forest structure, species composition and dynamics. However, few studies have examined forest functional responses to edges as a means of exploring why these characteristics may change at forest edges.
Foliar stable carbon isotope values (δ13Cleaf) can be used as an indicator of a tree's functional response to environmental gradients (Ehleringer et al. Reference EHLERINGER, FIELD, LIN and KUO1986), such as those present at forest edges. Modified growing conditions at edges, including high light, temperature, and vapour pressure deficit (VPD), are common across ecosystems (Chen et al. Reference CHEN, FRANKLIN and SPIES1995, Kapos Reference KAPOS1989, Newmark Reference NEWMARK2001, Williams-Linera et al. Reference WILLIAMS-LINERA, DOMÍNGUEZ-GASTELÚ and GARCÍA-ZURITA1998) and may influence tree growth and survival (Laurance et al. Reference LAURANCE, LOVEJOY, VASCONCELOS, BRUNA, DIDHAM, STOUFFER, GASCON, BIERREGAARD, LAURANCE and SAMPAIO2002). Specifically, these conditions can lead to increased stomatal closure and decreased ratios of intercellular to ambient CO2 concentrations (ci/ca) within the leaf, thereby limiting carbon uptake. For a given rate of net photosynthesis, declines in ci/ca can lead to an increase in δ13Cleaf as more 13CO2 is assimilated by the leaf.
Changes in δ13Cleaf can have downstream effects on plant δ13C values. Specifically, as the products of photosynthesis are translocated within a plant, some become substrates for root and rhizosphere respiration. δ13C values of soil respired CO2 (δ13CR-soil) represent a mixed signal, derived not only from sources that may rely on current photosynthate, but also from heterotrophic respiration in which soil organic carbon serves as the primary respiratory substrate. Despite the variation in inputs to soil respired CO2, climatic and environmental changes influencing the δ13C of carbon fixed in the canopy is detectable in δ13CR-soil 1–6 d following initial carbon fixation (Ekblad & Högberg Reference EKBLAD and HÖGBERG2001, Ekblad et al. Reference EKBLAD, BOSTRÖM, HOLM and COMSTEDT2005). These data indicate that both short-term environmental changes, such as seasonal increases in VPD, and long-term changes, such as those resulting from forest edge creation, will produce δ13C signals evident both above-ground and below-ground.
We chose to examine δ13C values in the ecosystem components described above: foliage, soil organic carbon and soil respired CO2 in the Sarapiquí region of Costa Rica. A shade-tolerant, leguminous tree species, Pentaclethra macroloba (Willd.) Kuntze, was selected as a focal species for this study. Reduced stomatal conductance has been observed in P. macroloba following extended exposure to full sun conditions in both field and laboratory settings (Oberbauer Reference OBERBAUER1983, Oberbauer et al. Reference OBERBAUER, STRAIN and RIECHERS1987). Because high light conditions can increase both temperature and VPD, it is probable that the stomatal sensitivity of P. macroloba under high light is linked to one or both of these factors. Therefore, it is likely that signs of environmentally induced stress on trees growing near forest edges will be apparent in P. macroloba.
With the goal of determining forest functional responses to the presence of an edge, we examined two primary hypotheses. First, δ13Cleaf will be more enriched, indicating decreased ci/ca, in canopy and understorey P. macroloba trees growing close to forest edges, relative to those in the forest interior. Second, patterns in δ13Cleaf detected in canopy trees will also be apparent in the mineral soil organic carbon δ13C value (δ13CSOC) and δ13CR-soil. Additionally, δ13Cleaf was examined relative to leaf mass per unit area (LMA, the reciprocal of SLA) and nitrogen concentration, with the expectation that differences in canopy and understorey δ13Cleaf will reflect higher photosynthetic capacity in canopy foliage.
METHODS
Site description
This study was conducted in the Sarapiquí region of north-eastern Costa Rica. Rainfall averages approximately 4000 mm annually with a mean annual temperature of 26 °C (Sanford et al. Reference SANFORD, PAABY, LUBALL, PHILLIPS, McDade, Bawa, Hespenheide and Hartshorn1994). There is little annual variation in either rainfall or temperature. The landscape is characterized by fragmented forests within an agricultural matrix comprised of active pastures and crop land (Butterfield Reference BUTTERFIELD, McDade, Bawa, Hespenheide and Hartshorn1994). Forests in Sarapiquí are primarily classified as tropical wet forest by the Holdridge life zone system (Tosi Reference TOSI1969). Between 32% and 35% of the basal area in these high-diversity forests is occupied by P. macroloba (Clark & Clark Reference CLARK and CLARK2000).
Three sites with 20–30-y-old forest-pasture borders were selected for this study, and all sites were located between 50 and 120 m asl on gently rolling terrain. Two sites, Rojomaca and Selva Verde, were located on highly weathered, acidic Ultisols derived from volcanic parent material while the third site, Tosi, was located on an Inceptisol derived from Quaternary alluvial deposits (ITCR 2004). Rojomaca was 375 ha in size, with western and south-western edge aspects, Selva Verde was 196 ha in size, with a western edge aspect, and Tosi was 120 ha in size, with a north-eastern edge aspect. The indicated site sizes represent only the size of an individual landowner's forested property; each site was embedded in a larger surrounding area of forest.
Rojomaca and Tosi were selectively harvested within 15 y of the present study, but no harvesting is known to have occurred at Selva Verde. Forest structural characteristics were similar among sites, regardless of logging history (see Schedlbauer et al. Reference SCHEDLBAUER, FINEGAN and KAVANAGH2007, Table 1, where Site 1 is Rojomaca, Site 2 is Tosi and Site 3 is Selva Verde).
Table 1. Within-leaf variation in δ13C measured for eight individual Pentaclethra macroloba trees. Trees are identified by their crown position as either canopy (C) or understorey (U). Among-pinna variation in δ13C was measured in pinnae from the bottom, middle and top third of the leaf. Within-pinna variation in δ13C was measured in basal and apical pinnules of pinnae from the middle third of each leaf.
Foliar sampling
At each site, we established three transects perpendicular to the forest edge. Transects were randomly located and spaced 22–100 m apart, depending on the length of the forest-pasture edge at each site. Samples were collected in 0.095-ha plots located at distances of 0, 25, 50, 100, 200 and 300 m from the forest edge. However, plots were not installed at two locations (25 m and 50 m in separate transects) at Rojomaca due to inundation. Each plot extended 20 m into the forest, parallel to the established transects. Two canopy and two understorey P. macroloba trees were sampled in each plot. If suitable trees could not be located within the plot, trees immediately adjacent to the plot were substituted.
A tree-climber cut one small branch from the upper third of each P. macroloba tree (dbh at 1.37 m range: 19.1–82.0 cm) sampled in the forest canopy. All leaves on these branches were in full sun and came from heights of 25–35 m above the forest floor. From each branch, we collected 10 fully expanded leaves and avoided sampling leaves with significant epiphytic growth. A similar procedure was used to collect foliage from shaded understorey P. macroloba trees (dbh range: 1.5–34.5 cm) at heights of 3–5 m above the forest floor.
A subset of the sampled trees, consisting of one canopy and one understorey tree per plot, was selected for measurements of LMA. These measurements were not made for all trees because of the difficulty associated with processing foliage from P. macroloba, a species with bipinnately compound leaves and very small pinnules. LMA was determined separately for three leaves per tree by selecting five pinnae per leaf for analysis. Pinnules were separated from the rachilla, scanned, dried at 70 °C for 48 h, and weighed to the nearest mg. The leaf area of each scanned pinnule was determined using Image-J image analysis software (v. 1.34s, National Institutes of Health, Bethesda, Maryland, USA).
To determine δ13Cleaf and nitrogen concentration (Nmass), 10 pinnae per sample tree were randomly selected for analysis. Pinnules were removed from the rachilla, dried at 70 °C for 48 h, and ground into a fine powder. δ13Cleaf and Nmass were determined at the Idaho Stable Isotopes Laboratory (ISIL). Stable carbon isotope data are expressed in standard delta notation as
where Rsample is the ratio of 13C to 12C in the sample of interest and Rstandard is the ratio of 13C to 12C in a standard (PDB). Twenty-two randomly selected foliar samples (10% of all samples) were analysed in duplicate, and analyses yielded a mean difference for δ13Cleaf of 0.25‰. The nitrogen concentration of each leaf, expressed as a fraction, was multiplied by LMA to calculate the quantity of nitrogen per unit leaf area (Narea).
Within-leaf δ13C variability was examined in a small subset of leaves (n = 8). For this analysis, pinnae were selected from the upper, middle and lower third of leaves from five canopy and three understorey trees. Pinnae from the middle third of these leaves were also divided into apical and basal pinnules to examine within-pinna variation in δ13C. Foliage was processed and analysed as described above.
Mineral soil organic carbon and soil respired CO2 sampling
Within each plot, three mineral soil samples from 0–10 cm depth were homogenized to form a single sample. All leaf litter was removed from the soil surface prior to sampling. Soils were air-dried to constant mass, sieved through a 2-mm mesh, and a subsample was homogenized for δ13C analysis. The mean difference in δ13CSOC for six duplicate soil samples (11% of all samples) was 0.31‰.
Soil CO2, representing both heterotrophic and autotrophic respiration, was also collected in each plot from plastic chambers fitted with rubber septa on the soil surface. Each chamber was inverted over an excavated circle in the soil, and all living plant material was removed from the area beneath the chamber, though litter was left in place. Soil from outside the study area was used to seal the edges of the chamber to minimize diffusion of atmospheric air into the chamber. Preliminary testing indicated that air beneath the chamber equilibrated with soil air within 4 d of installation (i.e. δ13C and CO2 concentration did not change with increased time) (data not shown). Therefore, chambers were left in place for at least 4 d before an air sample was collected. Soil CO2 samples were drawn from the chamber with a syringe and injected into evacuated 12-ml septum-capped Exetainer vials (Labco Ltd., High Wycombe, UK).
Air samples were shipped to the ISIL and analysed for δ13C 8–13 d after collection. δ13C values of soil CO2 samples were corrected by −4.4‰ to account for the difference in diffusion rates between 12CO2 and 13CO2 in the soil pore space (Cerling et al. Reference CERLING, SOLOMAN, QUADE and BOWMAN1991). This corrected value is hereafter referred to as the δ13C value of soil respired CO2, δ13CR-soil.
An independent test was conducted to examine the effect of air transportation on the δ13C values of CO2 stored in Exetainer vials. At the ISIL, 30 vials were evacuated and filled with CO2 gas of varying concentration (10 vials were filled with each of the following CO2 concentrations: 382.5 ppm, 3000 ppm and 10 000 ppm). The vial evacuation and filling procedure was carried out twice for each vial to reduce the possibility of contamination. Five vials per gas concentration were left in the laboratory, and the remaining five vials were sent to and returned from Costa Rica via air transportation. The δ13C values of the gas in the 30 vials were analysed 9 d after the vials were initially filled. Student's t-tests run for each CO2 concentration indicated that the δ13C of the 382.5 and the 10 000 ppm gas was not significantly changed following air transport (df = 8, 8, t = −0.763, −0.791, P = 0.47, 0.45, respectively). However, a slight isotopic shift was detected for the 3000-ppm gas. Vials shipped by air had a δ13C value enriched by a mean of 0.15‰ relative to samples that had remained in the laboratory (df = 8, t = 2.33, P = 0.048). Considered together, we interpret these results as an indication that soil CO2 samples collected in the field and sent to the ISIL for analysis were not significantly altered by air shipment.
Statistical analyses
The statistical language R (version 2.0.1, R Development Core Team) was used for all data processing and analyses. Linear mixed-effects models (Pinheiro & Bates Reference PINHEIRO and BATES2000) were used to examine differences in δ13C in pinnae and pinnules from varying leaf positions. ‘Leaf position’ was used as each model's fixed effect while each ‘leaf’ was used as the random effect. Analysis of variance (ANOVA) was used to test for differences in δ13C among the different leaf positions. LMA, Nmass, Narea, δ13Cleaf, δ13CSOC and δ13CR-soil data were also analysed with linear mixed-effects models. ‘Distance to forest edge’ was analysed as a continuous variable as each model's fixed effect. Random effects for these models were ‘transect’ nested within ‘fragment’. Prior to analysis, the LMA data for understorey foliage were reciprocal-transformed to normalize data, as suggested by Box–Cox tests (Box & Cox Reference BOX and COX1964). Individual ANOVAs were performed for each variable to test for differences along the forest edge to interior gradient.
RESULTS
Within-leaf δ13C variation
No significant within-leaf variability was detected in the δ13C value of pinnae from the bottom, middle and top of P. macroloba leaves (ANOVA, df = 14, F = 0.456, P = 0.64) (Table 1). However, a small but significant difference in pinnule δ13C values was detected (ANOVA, df = 7, F = 5.63, P = 0.0494), with basal pinnules more depleted than apical pinnules (mean difference: −0.16‰) (Table 1). It should be noted that this difference was below the level of precision in analysed duplicate foliar samples (0.25‰). The small variation in δ13C detected in these analyses indicate that a random sampling of pinnae sufficiently captures the overall bulk δ13Cleaf.
Forest edge to interior gradients
δ13Cleaf was invariant along the forest edge to interior gradient for both canopy (ANOVA, df = 95, F = 0.06, P = 0.80) and understorey foliage (ANOVA, df = 96, F = 1.20, P = 0.28) (Figure 1a). The mean ± 1 SE δ13Cleaf for canopy foliage was −29.6 ± 0.07‰ and was −32.5 ± 0.08‰ for understorey foliage.
Figure 1. Above- and below-ground δ13C values (mean ± SE) expressed in relation to the distance of each plot to the forest edge. Foliar δ13C values of Pentaclethra macroloba canopy and understorey foliage (δ13Cleaf) (a). Soil organic carbon δ13C values in the top 10 cm of mineral soil (δ13CSOC) and soil respired CO2 δ13C values (δ13CR-soil) (b).
No differences in LMA were detected for either canopy (ANOVA, df = 43, F = 1.33, P = 0.25) or understorey (ANOVA, df = 43, F = 0.64, P = 0.43) foliage along the edge to interior gradient (Table 2). Similarly, no significant differences in Nmass or Narea were observed along this gradient for either canopy (ANOVA, df = 43, 43, F = 0.62, 0.28, P = 0.44, 0.60, respectively) or understorey foliage (ANOVA, df = 43, 43, F = 0.14, 1.35, P = 0.71, 0.25, respectively) (Table 2). LMA in canopy foliage averaged 66.6 ± 0.97 g m−2 and was nearly double the mean LMA of understorey foliage, 35.2 ±0.69 g m−2 (Table 2). Narea was consistently higher in canopy foliage, relative to understorey foliage, while the opposite was true of Nmass.
Table 2. Mean leaf mass per unit area (LMA), nitrogen concentration (Nmass), and nitrogen per unit leaf area (Narea) ± 1 SE for canopy and understorey Pentaclethra macroloba foliage in relation to the distance of each plot to the forest edge.
δ13CSOC for samples collected from the top 10 cm of mineral soil did not vary along the forest edge to interior gradient (ANOVA, df = 42, F = 2.07, P = 0.16) (Figure 1b), and the mean value across all distance classes was −28.0 ± 0.05‰. However, δ13CR-soil did vary significantly across the edge to interior gradient (ANOVA, df = 42, F = 4.76, P = 0.0347) (Figure 1b). Values were more depleted at the forest edge than in the interior and ranged from −29.2 ± 0.39‰ at 0 m to −28.6 ± 0.28‰ at 300 m.
DISCUSSION
Above-ground responses to forest edge effects
In contrast to our expectations, we did not observe enriched δ13Cleaf in canopy or understorey P. macroloba trees growing near forest edges. These data indicate that P. macroloba did not exhibit a decline in ci/ca related to environmental stress at forest edges. Studies that have investigated changes in δ13Cleaf over environmental gradients have found irradiance and VPD to be significant factors in determining δ13Cleaf (Ehleringer et al. Reference EHLERINGER, FIELD, LIN and KUO1986, Hanba et al. Reference HANBA, MORI, LEI, KOIKE and WADA1997). However, neither of these factors appears to be influential along forest-edge-to-interior gradients in these tropical forests.
Conclusions regarding δ13Cleaf and ci/ca, must be considered in light of canopy boundary layer dynamics. Total conductance from the leaf to the atmosphere can be reduced by the presence of a large boundary layer in a forest canopy. Therefore, δ13Cleaf and ci/ca can be influenced by poor coupling of the canopy and the atmosphere. Forest canopies in Sarapiquí are typically uneven, ranging from 22–38 m in height, with canopy emergents reaching 50 m or greater (Clark & Clark Reference CLARK and CLARK2001, Lieberman et al. Reference LIEBERMAN, LIEBERMAN, PERRALTA and HARTSHORN1996). Further unevenness is contributed by frequent gap formation in these forests (Denslow & Hartshorn Reference DENSLOW, HARTSHORN, McDade, Bawa, Hespenheide and Hartshorn1994). Because aerodynamically rough canopies are typically well coupled to the atmosphere, it is unlikely that poor coupling and canopy boundary layer build-up played significant roles in determining total conductance, δ13Cleaf, and ci/ca in these canopies.
A study of edge effects conducted 5 y after forest edge creation in the Brazilian Amazon also failed to find variation in the δ13Cleaf of two canopy tree species along forest-edge-to-interior transects (Kapos et al. Reference KAPOS, GANADE, MATSUI and VICTORIA1993). However, enrichment in δ13Cleaf was observed in an understorey species at these forest edges and this was linked to both increased canopy openness at these young forest edges and enriched δ13C values in understorey air (Kapos et al. Reference KAPOS, GANADE, MATSUI and VICTORIA1993). Fragmented forests in Sarapiquí develop edges that seal with dense vegetation in the 20–30 y following edge creation, and no difference in canopy openness has been found between edge and interior environments (Schedlbauer et al. Reference SCHEDLBAUER, FINEGAN and KAVANAGH2007). The presence of a sealed edge decreases the likelihood that well-mixed air from outside the forest will penetrate edges and influence understorey δ13Cleaf. Our understorey δ13Cleaf data are consistent with this idea.
Although we did not detect evidence of edge effects in our measures of δ13Cleaf, there exists the possibility that P. macroloba trees growing directly adjacent to forest edges do exhibit enriched δ13Cleaf. Our sampling design was such that edge plots included trees growing anywhere between 0 and 20 m from the forest edge. Trees at the immediate edge of the forest are likely subject to a more extreme environment than trees growing close to an edge. The influence of hotter, drier conditions in adjacent pastures, as well as increased crown irradiance at edges could affect δ13Cleaf on a scale smaller than that measured in the present study. While this possibility exists, it is unlikely that it would have any significant effect on the forest as a whole. The development of dense vegetation at forest edges in Sarapiquí appears to stabilize edges (Schedlbauer et al. Reference SCHEDLBAUER, FINEGAN and KAVANAGH2007), and the potential for greater environmental stress on trees growing at immediate forest edges is unlikely to threaten this stability.
The overall lack of variation in δ13Cleaf within both the canopy and understorey suggests that the δ13C value of carbon available when leaves were developing was relatively constant. This is not surprising, given the relative aseasonality of temperature and precipitation in the Sarapiquí region (Sanford et al. Reference SANFORD, PAABY, LUBALL, PHILLIPS, McDade, Bawa, Hespenheide and Hartshorn1994). Studies in tropical regions with distinct seasonality also report little to no change in δ13Cleaf in canopy species between wet and dry seasons (Buchmann et al. Reference BUCHMANN, GUEHL, BARIGAH and EHLERINGER1997, Terwilliger Reference TERWILLIGER1997).
Comparison of canopy and understorey foliage
Photosynthetic capacity may influence δ13Cleaf (Duursma & Marshall Reference DUURSMA and MARSHALL2006, Hanba et al. Reference HANBA, MORI, LEI, KOIKE and WADA1997), and was assessed here indirectly via measures of LMA, Nmass and Narea (Field & Mooney Reference FIELD, MOONEY and Givnish1986, Reich et al. Reference REICH, WALTERS and ELLSWORTH1992, Reference REICH, WALTERS and ELLSWORTH1997). None of these parameters exhibited significant changes with increased distance to forest edges for either canopy or understorey foliage, indicating that differences in photosynthetic capacity were not influential in determining δ13Cleaf. However, the variation in these parameters between canopy and understorey P. macroloba trees was important in explaining the differences in δ13Cleaf from the canopy to the understorey.
A mean difference in δ13Cleaf of 2.86‰ was observed between canopy and understorey P. macroloba foliage, with canopy foliage exhibiting a more enriched δ13Cleaf than understorey foliage. This difference is slightly lower than is typically reported for tropical forests (Buchmann et al. Reference BUCHMANN, GUEHL, BARIGAH and EHLERINGER1997, Medina & Minchin Reference MEDINA and MINCHIN1980, Sternberg et al. Reference STERNBERG, MULKEY and WRIGHT1989), perhaps because other studies introduce interspecific variation to measures of δ13Cleaf and we report values for one species only. Buchmann et al. (Reference BUCHMANN, BROOKS and EHLERINGER2002), in a global analysis, determined that approximately 70% of the variation in δ13Cleaf within forest canopies is related to changes in isotopic discrimination, while the remaining 30% is attributed to variation in the δ13C of source air available for photosynthesis. Of the variation related to discrimination, there is a lack of consensus regarding the strongest drivers of the gradient in δ13Cleaf within forest canopies. However, variation in light availability as well as differences in photosynthetic capacity between canopy and understorey foliage is often influential in determining δ13Cleaf (Duursma & Marshall Reference DUURSMA and MARSHALL2006, Hanba et al. Reference HANBA, MORI, LEI, KOIKE and WADA1997).
LMA nearly doubled in canopy foliage, relative to understorey foliage. This pattern is characteristic of sun and shade foliage (Lambers et al. Reference LAMBERS, Chapin and PONS1998) and reflects the difference in light availability between the canopy and understorey. The high Narea observed in canopy foliage is also a manifestation of high light availability in the forest canopy, as most leaf nitrogen is associated with the photosynthetic apparatus of a leaf (Evans & Seemann Reference EVANS, SEEMANN and Briggs1989, Hanba et al. Reference HANBA, MIYAZAWA and TERASHIMA1999). Although we detected higher Nmass in understorey foliage than canopy foliage, this pattern is likely related to the thinness of shade leaves. An enrichment in canopy tree δ13Cleaf relative to understorey tree δ13Cleaf is partially the result of high light availability and photosynthetic capacity, both of which lead to draw-downs in ci. However, differences in water conducting path length, boundary layer conductance, leaf temperature, and leaf-to-air vapor pressure difference between foliage in canopy and understorey trees can also influence ci by inducing earlier or more frequent stomatal closure in canopy foliage.
Below-ground responses to forest edge effects
Consistent with our hypothesis that canopy-level patterns in δ13Cleaf would be reflected below-ground, δ13CSOC was found not to vary with distance to forest edges. However, a significant depletion in δ13CR-soil was detected in plots close to the forest edge. The magnitude of this depletion was approximately 0.5‰ and, though small, was consistent for all plots within 0–50 m of the forest edge. Evidence presented above decreases the likelihood that alterations in canopy processes such as photosynthesis and transpiration are responsible for these changes at the forest edge. Further, relatively constant δ13CSOC along the edge to interior gradient makes it unlikely that differences in the δ13C of soil organic matter explain these changes.
As described previously, the edges of forest fragments in Sarapiquí seal with vegetation in the 20–30 y following edge creation (Schedlbauer et al. Reference SCHEDLBAUER, FINEGAN and KAVANAGH2007), a process that is related to an increase in pioneer species at edges (Forero & Finegan Reference FORERO and FINEGAN2002). Pioneer species and shade-tolerant or late-successional species in tropical forests do not vary consistently in δ13Cleaf (Bonal et al. Reference BONAL, SABATIER, MONTPIED, TREMEAUX and GUEHL2000a, Huc et al. Reference HUC, FERHI and GUEHL1994), although the direction of isotopic shifts among tree functional types often varies with the ecophysiological traits of individual species (Bonal et al. Reference BONAL, BARIGAH, GRANIER and GUEHL2000b, Huc et al. Reference HUC, FERHI and GUEHL1994). In the forests of Sarapiquí, it is possible that the carbon available for root and rhizosphere respiration in pioneer species is more depleted than that of late-successional species. This depleted signal would then be evident in δ13CR-soil in areas of the forest dominated by early successional species, such as the forest edge. These possible shifts in δ13Cleaf may not have been detected in δ13CSOC because these forest edges are not occupied solely by early successional species. Further research is needed to assess whether variation in δ13Cleaf among tree functional groups is a potential driver of the pattern in δ13CR-soil observed at forest edges in Sarapiquí.
In the present study we were unable to accurately determine the CO2 concentration of soil air samples. As such, it was not possible to correct δ13CR-soil for the effect of atmospheric CO2 intrusion into the soil profile (Cerling 1991), though this can be done (Cernusak et al. Reference CERNUSAK, FARQUHAR, WONG and STUART-WILLIAMS2004). Because the δ13CR-soil data presented here are uncorrected, there remains the possibility that variation in soil CO2 concentration may have contributed to the reported variation in δ13CR-soil.
Conclusions
Functional changes in P. macroloba trees growing close to forest edges were not detectable in the fragmented forests of Sarapiquí. Along edge to interior gradients, we did not observe significant changes in δ13Cleaf, thus ci/ca, or any measured leaf characteristics in canopy or understorey foliage. Therefore, we conclude that P. macroloba trees growing close to forest edges are robust and functionally similar to trees in the forest interior. Variation in δ13Cleaf and other leaf characteristics between canopy and understorey foliage was typical of patterns commonly observed in forest canopies. Consistent with our above-ground data, no differences in δ13CSOC were found along edge to interior gradients. However, we did detect significantly greater depletion of δ13CR-soil at forest edges relative to interiors, but this pattern may be tied to shifts in species composition at edges rather than physiological responses to edge effects. Overall, these results are consistent with other research on edge effects in Sarapiquí reporting the development of a dense wall of vegetation at these forest edges (Forero & Finegan Reference FORERO and FINEGAN2002, Schedlbauer et al. Reference SCHEDLBAUER, FINEGAN and KAVANAGH2007). This dense vegetation appears to be effective in sealing the forest edge, as we detected few forest functional changes in edge zones.
ACKNOWLEDGEMENTS
Thanks to Leonel Coto for assistance in climbing trees and also to Bob Brander, Stacy Sesnie, Benjamin Miller, Erika Barrientos and Bertha Gonzalez for laboratory assistance. Germán Obando and Andrés Sanchún from FUNDECOR provided logistical support for fieldwork. Thanks to the landowners in Sarapiquí who graciously permitted access to their land. Helpful comments on earlier versions of this manuscript were provided by Bryan Finegan, John Marshall, Paul McDaniel and two anonymous reviewers. This research was supported by NSF-IGERT grant no. 0114304.
