Facilitation among woody plants establishing in an old field.
Woody plants in nonforested areas are frequently clumped (Bailey and Wroe 1974, Yarranton and Morrison 1974, Archer et al. 1988, Kellman and Kading 1992, Magee and Antos 1992, Teague and Smit 1992, Thiery et al. 1995, Milne et al. 1996). Clumps probably reflect seed dispersal patterns and vegetative spread (Payette and Filion 1985, Thiery et al. 1995, Greene and Johnson 1996), as well as the patchy distribution of suitable establishment sites (Kullman 1995). Whatever their initial cause, clumps of woody vegetation may initiate environmental changes associated with positive feedbacks (Wilson and Agnew 1992). Positive feedback occurs when a change in one part of a system influences a second part in such a way as to induce further change in the first part (DeAngelis et al. 1986). Positive feedbacks between plants and soil are frequently suggested as possible mechanisms underlying vegetation change (Berendse et al. 1989, Armand 1992, Vetaas 1992, Bertness and Callaway 1994, Wilson 1998) but, outside wetlands (Berendse et al. 1989, Bertness and Yeh 1994, Srivastava and Jefferies 1996), such suggestions have received little experimental attention.
Trees establishing in grassland in clumps may generate positive feedback by increasing soil resource availability. Possible mechanisms include the capture of fog (Wilson and Agnew 1992), rain (Weathers et al. 1995), snow (Payette and Filion 1985), or airborne mineral nutrients (Kellman 1989, Lovett 1992), sheltering nutrient-depositing animals (Wimbush and Forrester 1987, Archer et al. 1988, Belsky 1994), litter deposition (Van Cleve et al. 1983, Facelli and Pickett 1991, Hobbie 1992), or symbiotic N fixation (Chapin et al. 1994). Soils under shrubs or trees tend to have more moisture and available N than soils under adjacent grassland (Petranka and McPherson 1979, Miles 1985, Brady 1990, Jackson et al. 1990, Zak et al. 1990, Callaway et al. 1991, McPherson et al. 1991, Wesser and Armbruster 1991, Belsky 1994, Vieira et al. 1994, Kochy and Wilson 1997). Invading trees are associated with lower organic matter in northern prairie soils (Dormaar and Lutwick 1966, Bettany et al. 1973, Severson and Arneman 1973), possibly because of enhanced mineralization [TABULAR DATA FOR TABLE 1 OMITTED] rates. Environments with high rates of N mineralization may support more standing crop, causing shoot competition to increase in importance (Wilson 1993a, 1998), and allowing trees with high stem allocation to better resist competition from grasses (Werner and Harbeck 1982, Tilman 1988, Magee and Antos 1992).
The direct effects of high tree density might be expected to result in decreased growth for individual trees. Conversely, indirect effects could increase individual growth via suppression of neighboring grasses (Miller 1994). We examined these alternatives, and evaluated possible positive feedback mechanisms related to changes in light and soil conditions. Our specific objectives were to test (1) whether growing woody seedlings at high density would increase their ability to compete against grasses, and (2) whether woody plants in a simulated forest environment were less suppressed by grasses than those in an old-field environment.
We worked in an old field at the University of Regina, Regina, Saskatchewan (50 [degrees] 26 [minutes] N, 104 [degrees] 40 [minutes] W) on Dark Brown Chernozemic soil (Agriculture Canada 1992). We conducted the experiment in an old field rather than native prairie because prairie remnants are small and scarce in our region due to cultivation, The field was likely cultivated for several decades until it was planted with the introduced perennial grass Bromus inermis (nomenclature follows Looman and Best, 1987), probably in the 1960s. The grass was cut annually for hay until 1991 when it was set aside for research. The vegetation in our experiment was nearly all B. inermis (Gerry and Wilson 1995:276).
Average temperatures are -18 [degrees] C in January and 19 [degrees] C in July (Environment Canada 1982). The mean annual frostless period is 114 d. Mean monthly temperatures during the experiment differed from 30-yr mean monthly temperatures by [less than]3.5 [degrees] C (Environment Canada, unpublished data). Average annual precipitation is 384 mm. Mean monthly rainfall during the experiment was similar to the 30-yr mean, except for June 1994 and August 1995, when rain was about twice the monthly mean.
Our experiment had four factors, each with two levels: density (one or five transplants), environment (grassland or simulated-forest), grass (present or removed), and transplant species (Symphoricarpos occidentalis or Picea glauca). There were 40 replicates of each treatment combination with grass present and 20 replicates of each treatment with grass removed, for a total of 480 plots. Replication was higher for the grass-present treatment because we expected higher transplant mortality in this treatment. Each treatment was applied to a plot surrounded by a solid galvanized steel tube (15 cm diameter, 15 cm deep) hammered into the ground in early April 1994. The top of the tube was level with the soil surface. The tube ensured that plant roots interacted and it facilitated root harvest. Plots were separated by 1 m north-south and by 0.5 m east-west so that shades (see next paragraph) did not affect adjacent plots. Treatments were randomly allocated to plots. The experiment ran from spring 1994 until fall 1995.
The grassland level of the environment treatment received no resource manipulation. The simulated-forest level of the environment treatment was designed to reflect the higher levels of soil moisture, available N, and shade associated with forests (Table 1). Precipitation was supplemented in both years by 300 mm/yr, an amount about equal to the mean growing-season precipitation. Water was applied weekly: 10% in June, 40% in July, 40% in August, and 10% in September. This temporal pattern reflected seasonal variability in soil moisture in nearby natural forest (Wilson 1993a). N was supplied at 5 g N[center dot][m.sup.-2][center dot][yr.sup.-1] as commercial urea (N[H.sub.2]CON[H.sub.2]). N was applied in both years, with 30% in late May, 50% in late June, and 20% in late July. Similar to water, N availability was designed to follow average seasonal patterns. N was supplied with the last weekly water addition each month. Plots were shaded with 50% shade cloth (45 cm wide, 40 cm high) held vertically by two stakes [ILLUSTRATION FOR FIGURE 1 OMITTED]. The bottom of the cloth was 10 cm above the ground, 10 cm south of each plot. This arrangement shaded plots without interfering with rain. Fog and dew are uncommon in our dry climate and were probably not influenced by the shades. Shades were removed between September and April so that they would not affect snow accumulation.
Vegetation in plots with grass present was not manipulated. Vegetation in the grass-removal plots was sprayed with glyphosate, a rapidly decomposing, systemic herbicide, in late May 1994. Vegetation within 10 cm of grass removal plots was clipped monthly to prevent shading. Grass mass in June 1994 was 797 [+ or -] 344 g/[m.sup.2] (mean [+ or -] 1 SD, N = 30); height was 21.1 cm [+ or -] 5.4.
The two transplant species were chosen for their contrasting growth forms. Symphoricarpos is a common rhizomatous shrub 50-100 cm tall that forms circular clones up to 10 m diameter in prairie. It also surrounds individual aspen stands (Coupland 1950) and is the dominant understory species in aspen stands (Kleb and Wilson 1997). Symphoricarpos rhizomes were collected in May 1994 from natural prairie 20 km east of the experiment. Picea is a dominant tree of boreal forest north of the study site and is interspersed with native prairie in parts of southern Saskatchewan (Zoltai 1975, Pielou 1991). Picea seedlings frequently occur in prairie in even-aged clumps [approximately]5 m in diameter, which appear to establish from seed (S. Wilson, personal observation). Picea seedlings were obtained from a nursery [approximately]400 km north of Regina. Initial Picea height was 12.5 [+ or -] 2.3 cm (mean [+ or -] 1 SD).
Initial transplant mass was estimated using regression equations relating mass to size, developed from 30 individuals of each species. For Symphoricarpos, initial dry rhizome mass ([R.sub.d]) was predicted from fresh rhizome ([R.sub.f]) mass (g) ([R.sub.d] = -0.0379 + 0.4879[R.sub.f], [r.sup.2] = 0.98, P [less than] 0.01). Initial Symphoricarpos mass was 1.01 [+ or -] 0.64 g (mean [+ or -] 1 SD). For Picea, initial dry shoot (S) and root (R) mass (g) was calculated from stem height (H, cm; 1/S = -0.2726 + 9.8970/H, [r.sup.2] = 0.66, P [less than] 0.01; R = 0.0435[H.sup.1.0791], [r.sup.2] = 0.60, P [less than] 0.01). Initial Picea mass was 2.66 [+ or -] 0.71 g.
Five Symphoricarpos rhizomes or Picea seedlings were transplanted into each plot of the high density treatment; one transplant was in the center of each plot and the remaining four were planted around it [ILLUSTRATION FOR FIGURE 1 OMITTED]. One transplant was planted in the center of each low density plot, and four holes surrounding the transplant were made in the same position as the extra transplants in the high density treatment. The extra holes were filled with soil. Transplants were protected from sunlight for the 1 st wk by placing shade cloth (50% shade, 40 cm high, 30 cm wide) on the south side of each plot. Seedlings were transplanted in early June 1994, just after the usual date of last frost. During the following 2 wk, dead transplants were replaced with new ones in their original positions.
Picea height was measured in late October 1994 and early April 1995 to determine whether density influenced growth over winter. We measured from ground level to the highest green needle and estimated overwinter growth using the regression equations relating mass to height.
We established 30 additional plots with grass present but without woody targets, in order to measure the performance of grass in the absence of targets. Half the plots were assigned to the simulated-forest treatment to allow us to test the hypothesis that grass is more suppressed by woody plants in this environment.
Soil moisture and available N were sampled on 23 October 1994, and 2 May, 12 June, and 11 August 1995. Sample dates were at least 5 d after rainfall or water addition. One soil sample (2.5 cm diameter, 10 cm deep) was taken from each of 10 randomly chosen plots in each treatment combination on each sampling date. A 10-g subsample was extracted in 50 mL 0.02 mol/L KCl for 6 h and the supernatant was frozen until analysis for available N (sum of ammonium and nitrate) using an ion selective electrode (Orion, Boston, Massachusetts). The remainder of each soil sample was dried to determine water content.
Transplants and grass inside tubes were harvested in early September 1995. Belowground parts were washed and transplants were separated from grass. We measured the root mass (including rhizomes) and shoot mass of transplants and grass in each plot after drying plant material to constant mass at 70 [degrees] C. Mean transplant mass was determined for the high density plots. Root: shoot ratios were determined for transplants and grass. Transplant relative growth rates (RGR) were calculated as: RGR = [ln([M.sub.2]/[M.sub.1])]/d where [M.sub.1] is initial biomass, [M.sub.2] is final biomass, and d is the length of the growth period in days.
We used three-way contingency table analyses to determine whether survivorship varied with density, environment, or grass removal for each transplanted species. Analysis of variance (ANOVA) was used to examine the effects of density, environment, grass, and transplant species on transplant growth, root mass, shoot mass, and root: shoot ratios. We analyzed mean values from the high density treatments. Because growth accounts for initial differences in size among transplants, we used growth to test our hypotheses.
ANOVA involving all four factors was also used to examine soil water and N content on each sample date. Data were examined for normality and heteroscedasticity and transformed where necessary.
The response of grass to transplant density, environment, and transplant species was examined with three-way ANOVA. It was not possible to compare grass performance in these treatments with grass performance with no transplants present because this case had only one level of transplant density (0). Therefore, the response of grass biomass and root: shoot ratios to the presence of transplants was examined with two-way ANOVA with environment and target species (none, Symphoricarpos, or Picea) as main effects. This analysis used data pooled from the two density treatments because the previous analysis showed that transplant density had no significant effect on grass performance.
Survivorship of Symphoricarpos did not vary significantly with any treatment ([[Chi].sup.2] = 4.45, P [greater than] 0.1), but tended to be higher in the simulated-forest environment and with grass removed [ILLUSTRATION FOR FIGURE 2 OMITTED]. Survivorship of Picea was generally high but did not vary significantly with any factor ([ILLUSTRATION FOR FIGURE 2 OMITTED]; [[Chi].sup.2] = 8.98, 0.05 [less than] P [less than] 0.1). The results for Picea were probably near significant because survivorship was much lower in the grassland environment with grass present than in any other treatment [ILLUSTRATION FOR FIGURE 2 OMITTED]. The data suggest that grass decreased Picea survivorship in the grassland environment more than in the simulated-forest environment.
Transplant growth rates varied significantly with all main effects (Table 2; see Table 3 for the number of replicates used in analyses). Growth rates were generally lower at high transplant density, higher in the simulated-forest environment, higher in plots with grass removed, and generally higher for Picea than Symphoricarpos [ILLUSTRATION FOR FIGURE 3 OMITTED].
Interactions were of particular interest for testing our hypotheses (Table 2). The significant three-way interaction among density, grass removal, and species allowed us to test the hypothesis that the response of transplants to grass varied with density and between transplant species. We compared means between density treatments at each level of grass treatment for each transplant species [ILLUSTRATION FOR FIGURE 4A OMITTED]. Means were compared across environments since there was no four-way interaction involving environment (Table 2). With grass removed, growth rates were significantly reduced at high density for both transplanted species [ILLUSTRATION FOR FIGURE 4A OMITTED]. Picea growth was also decreased at high density with grass present. Symphoricarpos growth rates, however, were significantly greater at high density with grass present. Thus, the growth of one woody species in the presence of grass was significantly enhanced at high density.
Growth rates also varied significantly with the three-way interaction among density, environment, and grass removal (Table 2). This allowed us to test the hypothesis [TABULAR DATA FOR TABLE 2 OMITTED] that, for both transplant species considered together, transplants were more suppressed in the grassland environment than in the simulated-forest environment. Growth was significantly reduced by grass at both densities in both environments [ILLUSTRATION FOR FIGURE 4B OMITTED], suggesting that the ability of woody plants to resist suppression from grasses was not enhanced in the simulated-forest environment. The significant interaction was probably attributable to the highest growth rates occurring with grass removed in the simulated-forest environment at low density.
Overwinter growth of Picea was significantly higher in plots without grass [ILLUSTRATION FOR FIGURE 5 OMITTED] but did not vary significantly with any other factor. Overwinter growth was not consistently higher at high density, so there was no [TABULAR DATA FOR TABLE 3 OMITTED] evidence that conspecific neighbors improved overwinter performance.
Transplant shoot and root mass followed patterns similar to that of growth (Table 3), being greater for Picea than Symphoricarpos, greater with grass removed, greater in the simulated-forest environment, and greater at low density than high. The significant interaction among density, grass removal, and target species that occurred for growth did not occur for mass (Table 2), probably because initial differences in transplant size tended to increase variance and decreased our ability to detect interactions.
Transplant root:shoot ratios were slightly but significantly lower at high density than at low ([ILLUSTRATION FOR FIGURE 6 OMITTED]; Table 2). Root:shoot ratios did not vary significantly between environments. Root: shoot ratios were significantly higher in the presence of grass, and significantly higher for Symphoricarpos than Picea (note the different scales for each species in [ILLUSTRATION FOR FIGURE 6 OMITTED]). The highly significant interaction between grass removal and transplant species reflected the greater sensitivity of Symphoricarpos to the presence of grass; Symphoricarpos root:shoot ratios were more than doubled by the presence of grass, while Picea was unaffected.
Mass of grass neighbors in plots with surviving transplants was significantly higher in the simulated-forest environment [ILLUSTRATION FOR FIGURE 7A OMITTED] but did not vary significantly with transplant density or species. Similarly, grass biomass was not significantly reduced by the presence of transplants [ILLUSTRATION FOR FIGURE 7B OMITTED], but was higher in the simulated-forest environment. Grass root: shoot ratios did not vary significantly with any factor [ILLUSTRATION FOR FIGURE 7C OMITTED].
Soil moisture varied significantly with all four main effects on at least one sample date (Table 2). Soil moisture was significantly lower at high transplant density on three dates [ILLUSTRATION FOR FIGURE 8 OMITTED]. Moisture was significantly higher in the simulated-forest environment on all dates. Moisture was significantly higher in plots with grass in May but significantly lower in June. Moisture was significantly higher under Picea in October 1994 but significantly lower in May 1995. In summary, soil moisture was consistently lower at high transplant density, and higher in the simulated-forest environment.
Available N varied significantly with density only in May (Table 2), when N was lower at high density [ILLUSTRATION FOR FIGURE 9 OMITTED]. Simulated-forest plots had significantly more N than control plots in October 1994 and June 1995. Grass removal significantly increased N on each date. Transplant species did not differ significantly in their effects on available N.
High transplant density significantly enhanced the ability of Symphoricarpos to compete against grasses [ILLUSTRATION FOR FIGURE 4A OMITTED]. In the absence of grass, Symphoricarpos grew better at low than high density. In contrast, with grass present, Symphoricarpos growth rates at high density were twice those at low density. The results suggest that the direct effect of high Symphoricarpos density, measured in plots without grass, was to decrease Symphoricarpos growth, whereas the indirect effect of high density in the presence of grass was to increase Symphoricarpos growth. Symphoricarpos growth was enhanced by conspecific neighbors in the presence of grass [ILLUSTRATION FOR FIGURE 3 OMITTED], but final mass was lower with conspecific neighbors (Table 3). The reason for this apparent discrepancy is the lower survival rate with conspecific neighbors [ILLUSTRATION FOR FIGURE 2 OMITTED] which caused us to replace transplants early in the experiment. Replacement transplants were smaller than those used initially. Indirect positive effects were also found for perennial old-field species in a removal experiment by Miller (1994), but he found that positive effects only mitigated negative effects, so that net effects were negative. We may have found positive net effects because of the difference in growth forms between Symphoricarpos and its grass neighbors, or because we measured the response of transplants instead of populations.
In contrast, Picea growth was not facilitated at high density, possibly because of intense intraspecific competition. Total Picea mass at high density was [approximately]1500 g/[m.sup.2] whereas grass mass was -900 g/[m.sup.2] [ILLUSTRATION FOR FIGURE 7A OMITTED]. Symphoricarpos mass at high density was [approximately]250 g/[m.sup.2]. Similarly, Picea shoots may have influenced each other from the start of the experiment, whereas Symphoricarpos rhizomes without shoots were probably most affected by grass. Thus Symphoricarpos was probably most affected by competition from grass whereas Picea was most affected by intraspecific competition, and this may account for decreased Picea growth at high density regardless of the presence of grass.
Competition from grass significantly decreased transplant growth [ILLUSTRATION FOR FIGURE 3 OMITTED], but transplants had no significant effect on grass mass [ILLUSTRATION FOR FIGURE 7B OMITTED], reflecting the typical asymmetrical effects of established vegetation on seedlings (Goldberg 1990). In contrast, Symphoricarpos at high density had positive effects on the growth of conspecifics in the presence of grass, suggesting that high-density Symphoricarpos had some kind of negative effect on grass, even though grass mass was not significantly reduced. Spatial variation may have made the small reductions of grass mass in the presence of woody transplants [ILLUSTRATION FOR FIGURE 7B OMITTED] nonsignificant. Alternatively, grass may have responded to increased transplant density by decreasing physiological activity, benefiting Symphoricarpos without decreasing grass biomass.
Woody plants might be expected to be less suppressed in the simulated-forest environment if competitive abilities vary among environments (Walter 1985, Tilman 1988). Survivorship data supported this idea, with Picea survivorship lowest in the grassland environment with grass present but high in the simulated-forest environment with grass present [ILLUSTRATION FOR FIGURE 2 OMITTED]. Although the simulated-forest environment appeared to enhance the ability of Picea to survive in the presence of grasses, growth data did not support variation in competitive abilities. Growth of Picea was suppressed by grasses in both environments [ILLUSTRATION FOR FIGURE 3 OMITTED], as was the growth and survivorship of Symphoricarpos. Similarly, there was no evidence that the ability of transplants to affect grass varied with environment [ILLUSTRATION FOR FIGURE 7A OMITTED]. Changes in competitive abilities among environments may occur only for some species at some stages of their life history.
Winter winds have been suggested as a factor excluding trees from treeless areas (Kullman 1995), so trees at high density might provide some shelter and enhance individual growth. Density, however, had no effect on overwinter growth [ILLUSTRATION FOR FIGURE 5 OMITTED]. Grasses suppressed trees even in winter, as was found above an alpine tree line (Egerton and Wilson 1993).
Removal experiments suggest that competition in our old field is entirely below ground (Gerry and Wilson 1995, Peltzer et al. 1998). Experiments with herbaceous transplant species indicate that competition shifts from roots to shoots as nutrients increase and light decreases (Wilson and Tilman 1995). As a consequence, the total effect of competition varies little along such gradients. We observed this in our experiment: transplant growth rate showed no significant interaction between environment and grass removal (Table 2), suggesting that competition suppressed transplants to about the same extent in both environmental treatments. Transplant root:shoot ratios may also have reflected changes in root and shoot competition. Root:shoot ratios were significantly higher with grass present (Table 2, [ILLUSTRATION FOR FIGURE 6 OMITTED]), reflecting intense root competition. In contrast, root: shoot ratios were significantly lower at high transplant density (Table 2, [ILLUSTRATION FOR FIGURE 6 OMITTED]), reflecting increased stem allocation due to light competition.
Competitive effects of grasses on Picea survivorship appeared to be most intense in the grassland environment: survivorship was high in the forest environment regardless of grass neighbors, but was cut in half in the grassland environment when grass neighbors were present [ILLUSTRATION FOR FIGURE 2 OMITTED]. In contrast, the survivorship of Symphoricarpos, with a much higher root: shoot ratio [ILLUSTRATION FOR FIGURE 6 OMITTED], was unaffected by grass neighbors. Similar patterns emerged from comparisons of species with contrasting root: shoot ratios grown with and without neighbors in natural grassland and woody vegetation in experiments nearby (Wilson 1993a) and close to an alpine treeline (Wilson 1993b): plants with low root: shoot ratios are most susceptible to competition from grass.
Transplant survivorship was never promoted by the presence of grass [ILLUSTRATION FOR FIGURE 2 OMITTED], in contrast to many studies which show that neighbors increase survivorship while decreasing subsequent growth (Callaway 1995). Positive neighbor effects on survivorship are often attributable to protection from grazers. There may have been few grazers of Picea and Symphoricarpos at our experimental site, so that the net effect of neighbors on transplant survival was negative. Survivorship might have been lower if we had used smaller transplants.
Resource measurements indicated that our environmental treatment succeeded in creating simulated-forest conditions. Soil water was significantly higher on all sample dates [ILLUSTRATION FOR FIGURE 8 OMITTED], consistent with observed higher soil moisture in forest than prairie (Table 1). Soil moisture was higher in our experiment than in nearby prairie, probably because the prairie is a remnant on sandy soil unsuitable for agriculture. Soil available N was significantly higher in the simulated-forest treatment on two sample dates [ILLUSTRATION FOR FIGURE 9 OMITTED], consistent with higher soil N in forest than in prairie (Table 1). Soil N was lower in the experimental old-field than in nearby prairie, probably because of cultivation (Burke et al. 1995).
Both transplant growth and grass biomass were significantly higher in the simulated-forest environment [ILLUSTRATION FOR FIGURES 3, 7A OMITTED]. Survivorship also tended to be higher in the simulated-forest environment [ILLUSTRATION FOR FIGURE 2 OMITTED]. In contrast, the root: shoot ratios of both transplants and grass were not affected by the environment treatment (Table 2, [ILLUSTRATION FOR FIGURES 6, 7C OMITTED]). Root: shoot ratios usually decrease as soil resources increase (Wilson 1988). This may not have occurred in our experiment because the environment treatment was relatively modest in scale.
We hypothesized that woody plants at high density might initiate positive feedback by enhancing soil moisture and N mineralization, but the results showed the opposite pattern. Soil moisture was always lower at high transplant density [ILLUSTRATION FOR FIGURE 8 OMITTED] and soil available N was lower at high density on the single date when it varied significantly between density levels [ILLUSTRATION FOR FIGURE 9 OMITTED]. The transplants in our experiment had simple uptake effects on resources. More frequent measurements of soil water and more integrative measurements of available N (e.g., mineralization rates) might produce a more accurate picture of woody effects. Alternatively, woody seedlings may have had no positive effect on available N in our experiment because there is relatively little organic matter remaining to be mineralized in previously cultivated soils (Burke et al. 1995). The experiment might detect effects if it were conducted on undisturbed prairie soil with more organic matter. Lastly, more time might be required.
The lack of positive effects on soil resources caused by our transplants contrasts with results from large trees surrounded by grasses and with trees invading grasslands, both of which tend to be associated with higher soil moisture and N availability (Vetaas 1992, Ko and Reich 1993, Wilson 1993a, Belsky 1994). There may be a gradient of effects of woody plants on soil resources, with simple uptake (negative) effects caused by young, small woody plants and positive effects caused by mature trees. Our seedlings were clearly closer to the small end of the gradient. Maximum total transplant mass in our experiment was [approximately]1500 g/[m.sup.2], whereas tree mass in nearby forest is [approximately]7500 g/[m.sup.2] (Table 1). The mass of woody plants, and the time, required for effects on soil resources to change from negative to positive remains to be determined.
Regardless of short-term resource effects, our experiment suggests that indirect positive effects of conspecific neighbors can enhance the ability of woody plants to establish in grasslands. The results also support models invoking positive feedbacks during the invasion of grasslands by woody plants. Implications of positive feedbacks for natural area management include the possibility that invasions will be irreversible while feedbacks continue.
We thank R. Callaway, M. Huston, M. Kochy, R. Mitchell, D. Peltzer, and an anonymous reviewer for improving earlier drafts of this paper, L. Ambrose for technical help, R. Douslin of Saskatchewan Environment and Resource Management for allowing Symphoricarpos collection, J. Thompson of the same organization for supplying Picea, the University of Regina for support, and NSERC for a postdoctoral fellowship to X. Li and a research grant to S. Wilson.
Agriculture Canada. 1992. Soil landscapes of Canada: Saskatchewan. Centre for Land and Biological Resources Research, Research Branch, Agriculture Canada, Contribution 87-45, Publication 5243/B, Ottawa, Canada.
Archer, S., A. Scifres, C. R. Bassham, and R. Maggio. 1988. Autogenic succession in a subtropical savanna: conversion of a grassland to thorn woodland. Ecological Monographs 58:111-127.
Armand, A.D. 1992. Sharp and gradual mountain timberlines as a result of species interactions. Pages 360-378 in A. J. Hanson and F. di Castri, editors. Landscape boundaries. Springer-Verlag, New York, New York, USA.
Bailey, A. W., and R. A. Wroe. 1974. Aspen invasion in a portion of the Alberta Parklands. Journal of Range Management 27:263-266.
Belsky, A. J. 1994. Influences of trees on savanna productivity: tests of shade, nutrients, and tree-grass competition. Ecology 75:922-932.
Berendse, F., R. Bobbink, and G. Rouwenhorst. 1989. A comparative study on nutrient cycling in wet heathland ecosystems. II. Litter decomposition and nutrient mineralization. Oecologia 78:338-348.
Bertness, M.D., and R. Callaway. 1994. Positive interactions in communities. Trends in Ecology and Evolution 9:191-193.
Bertness, M. D., and S. M. Yeh. 1994. Cooperative and competitive interactions in the recruitment of marsh alders. Ecology 75:2416-2429.
Bettany, J. R., J. W. B. Stewart, and E. H. Halstead. 1973. Sulfur fractions and carbon, nitrogen, and sulfur relationships in grassland, forest and associated transitional soils. Soil Science Society of America Proceedings 37:915-918.
Brady, N. C. 1990. The nature and properties of soils. MacMillan, New York, New York, USA.
Burke, I. C., W. K. Lauenroth, and D. P. Coffin. 1995. Soil organic matter recovery in semiarid grasslands: implications for the Conservation Reserve Program. Ecological Applications 5:793-801.
Callaway, R. M. 1995. Positive interactions among plants. Botanical Review 61:306-349.
Callaway, R. M., N. M. Nadkarni, and B. E. Mahall. 1991. Facilitation and interference of Quercus douglasii on understory productivity in central California. Ecology 72: 1484-1499.
Chapin, F. S., L. R. Walker, C. L. Fastie, and L. C. Sharman. 1994. Mechanisms of primary succession following deglaciation at Glacier Bay, Alaska. Ecological Monographs 64:149-175.
Coupland, R. T 1950. Ecology of mixed prairie in Canada. Ecological Monographs 20:271-315.
DeAngelis, D. L., W. M. Post, and C. C. Travis 1986. Positive feedback in natural systems. Springer, Berlin, Germany.
Dormaar, J. T., and L. E. Lutwick. 1966. A biosequence of soils of the Rough Fescue prairie-poplar transition in southwestern Alberta. Canadian Journal of Earth Sciences 3:457-471.
Egerton, J., and S. D. Wilson. 1993. Overwinter competition in alpine plant communities. Arctic and Alpine Research 25:124-129.
Environment Canada. 1982. Canadian climate normals: temperature and precipitation 1951-1980: prairie provinces. Environment Canada, Ottawa, Canada.
Facelli, J. M., and S. T. A. Pickett. 1991. Indirect effects of litter on woody seedlings subject to herb competition. Oikos 62:129-138.
Gerry, A. K., and S. D. Wilson. 1995. The influence of initial size on the competitive responses of six plant species. Ecology 76:272-279.
Goldberg, D. E. 1990. Components of resource competition in plant communities. Pages 27-49 in J. B. Grace and D. Tilman, editors. Perspectives on plant competition. Academic Press, San Diego, California, USA.
Greene, D. F., and E. A. Johnson. 1996. Wind dispersal of seeds from a forest into a clearing. Ecology 77:595-609.
Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Trends in Ecology and Evolution 7:336-339.
Jackson, L. E., R. B. Strauss, M. K. Firestone, and J. W. Bartolome. 1990. Influence of tree canopies on grassland productivity and nitrogen dynamics in deciduous oak savanna. Agricultural Ecosystems and Environments 32:89-105.
Kellman, M. 1989. Mineral nutrient dynamics during savanna-forest transformation in Central America. Pages 137-151 in J. Proctor, editor. Mineral nutrients in tropical forest and savanna ecosystems. Blackwell, London, UK.
Kellman, M., and M. Kading. 1992. Facilitation of tree seedling establishment in a sand dune succession. Journal of Vegetation Science 3:679-688.
Kleb, H. R., and S. D. Wilson. 1997. Vegetation effects on soil resource heterogeneity in prairie and forest. American Naturalist 150:283-298.
Ko, L. J., and P. B. Reich. 1993. Oak tree effects on soil and herbaceous vegetation in savannas and pastures in Wisconsin. American Midland Naturalist 130:31-42.
Kochy, M., and S. D. Wilson. 1997. Litter decomposition and nitrogen dynamics in aspen forest and mixed-grass prairie. Ecology 78:732-739.
Kullman, L. 1995. Holocene tree-limit and climate history from the Scandes Mountains, Sweden. Ecology 76:2490-2502.
Looman, J., and K. K. Best 1987. Budd's flora of the Canadian prairie provinces. Publication 1662, Agriculture Canada, Research Branch, Hull, Canada.
Lovett, G. M. 1992. Atmospheric deposition and canopy interactions of nitrogen. Pages 152-166 in D. W. Johnson and S. E. Lindberg, editors. Atmospheric deposition and forest nutrient cycling. Springer-Verlag, New York, New York, USA.
Magee, T. K., and J. A. Antos. 1992. Tree invasion into a mountain-top meadow in the Oregon Coast Range, USA. Journal of Vegetation Science 3:485-494.
McPherson, G. R., G. A. Rasmussen, D. B. Wester, and R. A. Masters. 1991. Vegetation and soil zonation associated with Juniperus pinchotii Sudw. trees. Great Basin Naturalist 51:316-324.
Miles, J. 1985. The pedogenic effects of different species and vegetation types and the implications of succession. Journal of Soil Science 36:571-584.
Miller, T. E. 1994. Direct and indirect species interactions in an early old-field plant community. American Naturalist 143:1007-1025.
Milne, B. T., A. R. Johnson, T. H. Keitt, C. A. Hatfield, J. David, and P. T. Hraber. 1996. Detection of critical densities associated with pinon-juniper woodland ecotones. Ecology 77:805-821.
Payette, S., and L. Filion. 1985. White spruce expansion at the tree line and recent climatic change Canadian Journal of Forest Research 15:241-251.
Peltzer, D. A., S. D. Wilson, and A. K. Gerry 1998. Competition intensity along a productivity gradient in a low-diversity grassland. American Naturalist 151:465-476.
Petranka, J. W., and J. K. McPherson. 1979. The role of Rhus copallina in the dynamics of the forest-prairie ecotone in north-central Oklahoma Ecology 60:956-965.
Pielou, E. C. 1991. After the ice age: the return of life to glaciated North America. University of Chicago Press, Chicago, Illinois, USA.
Severson, R. C., and H. F. Arneman. 1973. Soil characteristics of the forest-prairie ecotone in northwestern Minnesota. Soil Science Society of America Proceedings 37: 593-599.
Srivastava, D. S., and R. L. Jefferies. 1996. A positive feedback: herbivory, plant growth, salinity, and the desertification of an Arctic salt-marsh. Journal of Ecology 84:31-42.
Teague, W. R., and G. N. Smit. 1992. Relations between woody and herbaceous components and the effects of bush-clearing in southern African savannas. Tydskrif van die Weidingsvereniging van Suidelike Afrika 9:60-71.
Thiery, J. M., J.-M. D'Herbes, and C. Valentin. 1995. A model simulating the genesis of banded vegetation patterns in Niger. Journal of Ecology 83:497-507.
Tilman, D. 1988. Plant strategies and the dynamics and structure of plant communities Princeton University Press, Princeton, New Jersey, USA.
Van Cleve, K., C. T Dryness, L. A. Viereck, J. Fox, F. S. Chapin, and W. Oechel. 1983. Taiga ecosystems in interior Alaska. BioScience 33:39-44.
Vetaas, O. R. 1992. Micro-site effects of trees and shrubs in dry savannas. Journal of Vegetation Science 3:337-344
Vieira, I. C. G., C. Uhl, and D. Nepstad. 1994. The role of the shrub Cordia multispicata Cham. as a 'succession facilitator' in an abandoned pasture, Paragominas, Amazonia. Vegetatio 115:91-99.
Walter, H. 1985. Vegetation of the earth Springer-Verlag, New York, New York, USA.
Weathers, K. C., G. M. Lovett, and G. E. Likens. 1995. Cloud deposition to a spruce forest edge. Atmospheric Environment 29:665-672.
Werner, P. A., and A. L. Harbeck. 1982. The pattern of tree seedling establishment relative to staghorn sumac cover in Michigan old fields. American Midland Naturalist 108: 124-132.
Wesser, S. D., and W. S. Armbruster. 1991. Species distribution controls across a forest-steppe transition: a causal model and experimental test. Ecological Monographs 61: 323-342.
Wilson, J. B. 1988. A review of evidence on the control of root: shoot ratio, in relation to models. Annals of Botany 61:433-449.
Wilson, J. B., and A. D. Q. Agnew. 1992. Positive feedback switches in plant communities. Advances in Ecological Research 23:263-336.
Wilson, S. D. 1993a. Belowground competition in forest and prairie. Oikos 68:146-150.
-----. 1993b. Plant competition and resource availability in alpine heath and grassland in the Snowy Mountains of southeast Australia Journal of Ecology 81:445-451.
-----. 1998. Competition between grasses and woody plants Pages 231-254 in G. Cheplick, editor Population biology of grasses. Cambridge University Press, Cambridge, UK.
Wilson, S. D., and H. R. Kleb. 1996. The influence of prairie and forest vegetation on soil moisture and available nitrogen. American Midland Naturalist 136:222-231.
Wilson, S. D., and D. Tilman. 1995. Competitive responses of eight old-field plant species in four environments. Ecology 76:1169-1180.
Wimbush, D. J., and R. I. Forrester. 1987. Effects of rabbit grazing and fire on a subalpine environment. II. Tree vegetation. Australian Journal of Botany 36:287-298.
Yarranton, G. A., and R. G. Morrison. 1974. Spatial dynamics of a primary succession: nucleation. Journal of Ecology 62:417-428.
Zak, D. R., D. F. Grigal, S. Gleeson, and D. Tilman. 1990. Carbon and nitrogen cycling during old-field succession: constraints on plant and microbial biomass. Biogeochemistry 11:111-129.
Zoltai, S.C. 1975. Southern limit of coniferous trees on the Canadian prairies. Information Report NOR-X-128, Canadian Forestry Service, Edmonton, Alberta, Canada.
|Printer friendly Cite/link Email Feedback|
|Author:||Li, Xingdong; Wilson, Scott D.|
|Date:||Dec 1, 1998|
|Previous Article:||An 85-year study of Saguaro (Carnegiea gigantea) demography.|
|Next Article:||Regulation of fine root dynamics by mammalian browsers in early successional Alaskan Taiga forests.|