Our first hypothesis predicted that low soil moisture would have a large influence on two-year survival and growth of C. florida and Q. alba seedlings. Even though soil moisture was an important factor, this hypothesis was rejected, because water did not have the large impact expected for first or second year survival, or for growth. Analysis at the whole plot scale indicated that there were broad differences in water availability across sites and treatments, but the analyses of seedling responses within plots revealed that water availability explained seedling responses very infrequently. At the scale of whole plots, both species had greater survival at the two xeric sites than at the mesic site , even though the difference between xeric and mesic sites was significant only for C. florida . Height, diameter and biomass growth were affected by site differences; however, there was no consistent pattern in the differences attributable to site moisture . Since the lowest mean and extreme soil moisture measures occurred at the xeric sites, roughly half those for the mesic site , we concluded that the effects of low water availability on survival and growth did not exceed the influence of other plot-scale factors. Further evidence at the scale of whole plots was shown in the comparison of cutting treatments. The clearcut plots had significantly greater water availability than the understory plots , but there were no significant differences in seedling survival for either species between the two treatments. All growth responses were significantly different between cutting treatments; however, it could not be stated unequivocally that soil moisture availability was the lone reason for these differences, because clearcut plots also had greater availability of nutrients and light than the understory plots. At the within-plot scale, water was seldom a significant factor explaining variability of survival in the clearcut plots, although it was one of the most frequently identified variables affecting first year survival in the understory plots (Table 9). Water significantly affected growth in only 1 out of the 24 tests performed ; however, that one significant growth response to microsite variation in water availability was positive (i.e., positive regression coefficient). It was barely positive, illustrating that increased water had only a slight positive effect on Q. alba biomass growth , not the large positive relationship we expected. Although not a dominant factor, water was clearly important. In fact, the presence of excess water (flooding) was likely the most important factor for survival in both C. florida and Q. alba seedlings. Numerous puddles caused by poor drainage were present at the Hog Barn site and were the principal cause for the significant water availability differences between sites (Figure 4; Table 4). Virtually all seedlings, most notably C. florida, located in these puddles died. In effect, we may have been looking at the water availability from the wrong direction. The seedlings may not have been able to respond to added resources, because they could not tolerate high water availability, implying that water may truly have been more important than our data led us to believe. In addition to water, nutrient availability and light were correlated with seedling performance, but none of these variables stood out as the principal limitation to seedling survival or growth. At the whole plot scale, sites and treatments differed significantly in nitrogen and phosphorus availability (Table 4), but the largest differences were between the clearcut and understory treatments. Furthermore, light was obviously much greater in clearcuts than in the understory (full sunlight in clearcut versus a mean of 25.13 ± 1.01% of full sunlight in the understory). As reported earlier, most growth parameters for both species in the clearcut plots were also significantly greater than in the understory plots. Because of this, the facilitative effects of the understory did not show much of an impact on these seedlings. Facilitation may have occurred, but it was not rapid enough to compensate for the losses from increased shade and decreased resources to compare the understory seedlings with their clearcut counterparts. Since water, nutrients and light were all least in the understory, we were unable to determine which factor, at the scale of whole plots, was the catalyst for such differences. Analysis within plots was less ambiguous. Nitrogen availability was significant two times (out of a possible 12 regressions) for survival in the clearcut plots, and phosphorus was significant one time (Table 9). In the understory, phosphorus and nitrogen availability each significantly impacted survival only one time (out of a possible 12 regressions). All five of these significant situations occurred at the Hog Barn site. For growth, phosphorus availability was significant in three regressions, and nitrogen in two (Table 10). Nitrogen was only significant for growth in the understory, possibly because nitrogen availability was much more abundant in the clearcut plots. Phosphorus was a limiting factor at the Hog Barn site, because that site had the lowest mean phosphorus levels . Total carbon content at the whole plot scale was greatest at the Hog Barn site where survival was poor and growth was modest or low . As expected, total carbon was also significantly greater in the understory plots than the clearcut plots . When analyzed at the whole plot scale, total carbon content had an apparent negative effect, because increased levels of total carbon content were associated with decreased survival and growth. An exception to this rule was the positive relationship of the C. florida Green Pond clearcut combination , indicating a strong association between increased growth response and increased total carbon content. Within plots, however, the apparent effect of total carbon content was complex. Survival was significantly affected by total carbon content only twice (out of a possible 24 regressions; Table 9). In both cases, which occurred at the Hog Barn site, increased carbon was associated with decreased survival. Growth was significantly related to total carbon content in three cases (out of a possible 24 regressions) spread across both species, both treatments and two sites (Green Pond and Beaufort). Only in one case, however, was there an obvious positive relationship between carbon and biomass growth; i.e., for C. florida at the Green Pond clearcut plot. This was the result we expected, but it happened rarely. We concluded that carbon was relatively unimportant, but it could also be concluded that points of lesser total carbon content could be places of increased mineralization leading to increased nutrient content. From this perspective, seedlings should respond better to lesser total carbon content and have a negative relationship, making total carbon content an important variable.
Effect of light
Our second hypothesis involved the effect of light on growth during the duration of the study. We cannot reject or accept this hypothesis; however, we suspect that it is incorrect. Growth at the clearcut sites was clearly greater than at the understory sites where light was much greater as compared to the understory. As stated above, however, clearcut plots also had more available water and nutrients than understory plots, possibly confounding the effect light may have on growth and vice versa. In the understory, where variation in light levels might be expected to play a role in growth, the role of light was weak or undetectable. There was a significant difference in understory light availability among sites , but growth parameters did not parallel these differences. For example, the Beaufort site had the greatest mean understory light availability, yet seedling growth there was not distinctly greater than either of the other two sites . Within each understory site, light was identified by multiple regressions as a factor affecting growth, but it was not very influential when compared to the other environmental factors . Light was significant for only 4 out of 24 (16.7%) tests. Leaf area was the growth parameter most affected by light; the two sites with the greatest light availability in the understory (Beaufort and Green Pond) had the greatest total leaf area. Light availability only affected biomass growth in 1 out of 6 site/treatment combinations (Figure 12). Although light was not the principal limiting factor affecting seedling performance during the first two growing seasons, it is assumed that it will have a much greater impact in the near future.
Effects of herbivory
Our third hypothesis stated that herbivory will have a significant impact on seedling survival and growth. At the whole plot scale, evidence to support this hypothesis was weak. First, no significant differences in stem herbivory were detected among sites in either species . Second, in the sites where leaf herbivory was particularly heavy, survival was not particularly low. For example, leaf herbivory of Q. alba was greatest in the Beaufort understory site, but Q. alba survival was very good at this site . The same pattern was seen for C. florida at the Green Pond understory site; i.e., high leaf herbivory yet good survival. At the Beaufort understory site, Q. alba growth was low while leaf herbivory was at its greatest among all site/treatment combinations. This relationship between leaf herbivory and growth was not as evident for C. florida. Leaf herbivory was greatest for C. florida at the Hog Barn understory plot , but growth was not consistently the smallest at this site. Within plots, evidence for herbivory effects was much stronger. Leaf herbivory was a significant factor in the survival analyses for just 3 out of 24 tests and stem herbivory was significant only once (Table 9); however, both leaf and stem herbivory were significant factors in many of the growth analyses. In fact, leaf and stem herbivory were the two most frequently identified variables affecting responses within plots . Leaf herbivory was significant 9 times and stem herbivory 7 times, each out of a possible 24 regressions . In models where they were significant, both forms of herbivory were inhibitors to total leaf area growth (average r2 for 1st year = 0.190; average r2 for 2nd year = 0.205) as well as total biomass growth (average r2 = 0.189). They explained the most variation in both of these growth responses among all of the environmental variables, including water and light. They were also present in every case where there were two or more variables included, and occurred together in four separate regressions. Because of this and the fact that herbivores consumed 40-60% of the leaf area or stem in some plants (Figure 9), herbivory was a strong factor. However, the importance of herbivory may have been overstated by the data. The herbivores may have selected either slow-growing or weak plants, indicating a greater effect on survival than would generally be present. Generally, deciduous species can overcome modest degrees of defoliation (Krause and Raffa 1996). Krause and Raffa found that deciduous larch (Larix deciduas) defoliated by 33% and 66% recovered their aboveground biomass growth to near control values within one year. Recovery, however, varies among species. Many species have been known to compensate for losses to herbivores up to about 75% (Risley 1993) while 25% defoliation may fatally inhibit plant function in other species (Verkaar 1987). Beyond these threshold values, growth losses due to early abscission are likely (Risley 1993). Hardwood growth after defoliation, or recovery, is accomplished by returning to as nearly normal a photosynthetic machinery as possible, taking advantage of its ability of multiple flushes and allowing the plant to survive until more favorable times (Hodkinson and Hughes 1982). Therefore, much of this recovery ability is speciesdriven. Many of these herbivory studies were performed by mechanical defoliation or in greenhouse studies. Plants grown in natural environments where stress factors are present may be much less capable of overcoming leaf tissue losses as compared to greenhouse seedlings. One field experiment used a random sample of 50 leaves of 52 plant species, giving a mean defoliation of 10.46%, much higher than our results (de la Cruz and Dirzo 1987), but recovery was not measured. Linit et al. (1986) also had field experiment leaf area losses of 20-24%, but the impact on survival and growth was not measured; it was only assumed that the defoliation was not detrimental. This is very common among herbivory experiments. Either the biology of recovery via mechanical defoliation studies or the quantification of leaf herbivory via field studies is the main goal of these studies, not both.
One important goal was to develop effective and cost efficient methods to artificially regenerate large-seeded hardwoods in pine-dominated landscapes of the coastal plain. In a preliminary experiment at Savannah River Site, we found that direct seeding of Q. alba and C. florida into pine understories or recently clearcut pine forests resulted in poor survival (less than 10% of planted C. florida seeds and approximately 25-30% of planted Q. alba seeds) and very poor growth (generally less than 10 cm of height growth) four years after planting. In this study, we evaluated the planting of oneyear-old, bare-root seedlings directly into the understory, or in clearcut plots during the first dormant season after harvest and site preparation. We conclude that planting in the clearcut plots is the most effective method, principally because there was no significant difference in survival between cutting treatments (Table 4); however, in most cases, growth in the clearcut treatments was significantly greater than in the understory (Table 5). The large differences in growth between the treatments also proved that the two-year duration of the study was long enough for us to detect effects of our proposed environmental factors on seedling performance. Our recommendation contrasts with general guidelines for hardwood regeneration that emphasize establishment of advance regeneration in the understory followed by release. These guidelines are based on studies that show positive effects of clearcutting on herbaceous and non-crop woody species, and subsequent strong impacts of these competitors on crop species (Sander et al. 1976, Wright et al. 1984, Beck and Hooper 1986, Crow 1988, Brose and Van Lear 1998). However, these studies also show that competition problems are much less in the understory than in the clearcuts, and regeneration is easier on poor quality versus high quality sites. Our two driest sites certainly qualify as unproductive relative to most eastern US hardwood-dominated forests, and survival was much greater at these sites than at our more productive Hog Barn Site. Standing water may have complicated the Hog Barn site’s productivity, but in general, our results seem consistent with the literature. On the other hand, our sites had few hardwoods in the understory prior to harvest. Site preparation (including burning and herbicide) was also used to reduce competing vegetation. These caveats should be considered before upland hardwood seedlings are planted into recently clearcut forests.
Our intent in this study was to determine which of several environmental factors have the greatest influence on survival and growth of Quercus alba and Cornus florida seedlings. Since the environmental factors varied both among and within plots, teasing apart the influence of individual factors was difficult. However, some important key points emerged. First, variation among seedlings was quite high, and very few of our models were capable of explaining more than 40% of the variation in survival and growth. Thus, there are clearly other variables, such as genetic differences among the seedlings that affect seedling performance not accounted for in our study. This is not an unusual finding; virtually all studies of seedlings planted in natural conditions have found a substantial background variability that is not accounted for by measured variables. A greenhouse study might better quantify our limiting factors, but it would eliminate the stochastic factors present among the natural environment. Analysis of allometric relationships, e.g., between leaf weight and shoot weight, might also better show how the environmental variables, specifically light and water, impacted the seedlings.
2. Objectives and hypotheses
3. Literature Review
3.1 Establishment and Persistence
3.2 Light or Water?
3.3 Environmental Stressors
3.4 Other factors affecting seedling establishment and long-term survival
4. Results from Previous Work
5.1 Site Description
5.2 Plot Selection
5.3 Seedling Planting
5.4 Water Measurements
6.1 Site and cutting treatment effects on environmental conditions and herbivory
6.2 Site and cutting treatment effects on seedling survival and growth
6.3 Within-site seedling responses
7.1 Resource availability
7.2 Effect of light
7.3 Effects of herbivory
7.4 Practical implications
GET THE COMPLETE PROJECT
Factors limiting the regeneration of largeseeded hardwoods in the Upper Coastal Plain of South Carolina