In contrast to growth and yield models that simulate the development of even-age stands 20-years old or more (TWIGS: Miner and others 1988; STEMS: Belcher and others 1982; and, OAKSIM: Hilt 1985), regeneration prediction models predict structure or composition of new stands during the first two decades after final harvest. This type of modeling is extremely difficult owing to the highly stochastic nature (both spatially and temporally) of regeneration events. Here wedescribesome regeneration prediction models that simulate the development of new stands.

• Predicting Reproduction in the southern Appalachian Uplands
• Prediction of Oak Regeneration in Bottomland Forests
• Regenerating Northern Red Oak in the Southern Appalachians

Regeneration models are useful silvicultural tools for assessing stand regeneration potential. Using such models usually requires an inventory of the advance reproduction and the overstory. The practical application of predictive regeneration models thus does not differ greatly from the application of growth and yield models (Johnson 1993). Some essential conceptsfor predicting and modeling oak regeneration, such as acorn, seedling, and sprout dynamics must be kept in mind.

• Essential concepts for predicting oak regeneration

For managers the application of these prediction models has one very clear imperative: a commitment to collect adequate data about existing stand conditions. The midstory and overstory inventory data are used to estimate the proportion of stump sprouts from cut trees that can reasonably be expected to produce dominant or codominant oaks at age 20. This estimate, together with the advance reproduction probabilities, is used to determine if the new stand can be expected to develop into a predominantly oak stand by age 20 (Sander and Graney 1993).

Regeneration modeling is notoriously difficult. Even when the essential characteristics of the oak reproduction are known, it is difficult to assess the contribution of oaks to the regenerated stand because of the unpredictable growth of reproduction. Much of this uncertainty is related to the difficulty of obtaining an accurate measure of root size. Growth of oak reproduction largely depends on root size, which may be much larger and older than the above-ground stem. Recurrent stem dieback and resprouting of oak reproduction beneath an overstory may go on for decades before the reproduction expresses its growth potential. This potential is finally expressed when the overstory is harvested or when it is destroyed by natural events such as fire, windthrow, and disease. Although some silviculturists can predict general regeneration trends after timber harvesting based on experience and observation, more quantitative methods are needed to accurately assess the probable contribution of oak reproduction to future stand stocking (Johnson 1993).

Unfortunately, root size is impossible to measure in practice. The next best measure of the growth potential of reproduction is probably its diameter just above the ground. However, the relation between shoot growth and basal stem diameter, although usually statistically significant, is relatively weak (Dey 1991, Johnson 1979, Sander 1971). To overcome this problem, some predictive regeneration models employ probabilistic methods to assess the importance of oaks in the future stand (Dey 1991, Loftis 1990, Sander and others 1984). Other regeneration models use other methods (Waldrop and others 1986, Marquis and others 1984). Regardless of their derivation, regeneration models are useful silvicultural tools for assessing stand regeneration potential. Using such models usually requires an inventory of the advance reproduction and the overstory. The practical application of predictive regeneration models thus does not differ greatly from the application of growth and yield models (Johnson 1993).

The structure of regeneration models is determined by the sources of reproduction and by the biotic and abiotic factors that affect population dynamics of the ecosystem being modeled. Sources of reproduction include new seedlings, advance reproduction, and sprouts from the cut stumps of overstory trees. The micro-environment (e.g., light, temperature, and soil moisture) influences the development of reproduction. Aspect, slope position, elevation, and site index are commonly used to express the integrated effect of micro-environment on reproduction. In some upland oak ecosystems, wildlife may adversely affect stand regeneration potential. Interference species such as ferns, fire cherry, red maple, and flowering dogwood may dominate the site following harvest, thereby reducing regeneration potential. The importance of these factors in influencing regeneration potential varies among regions. Thus, regeneration models must be developed for specific regions. ACORn is an example of a regional stand-level regeneration model developed for even-aged forests in the Missouri Ozarks. It can be used to predict future stand characteristics by species and diameter classes, which in turn, can be used as input for growth and yield simulators. ACORn is a model developed for conditions slightly outside the southern Appalachian region, however there are so few regeneration simulation models available that it might be used in the driersites of the region.

• Introduction
• Regeneration Models Review
• The ACORn Simulator
• Predicting Regeneration Quality
• Regeneration Surveys

Growth and yield models can simulate the development of even-age stands that are about 20 years of age and older. In the Central States, growth models such as TWIGS (Miner and others 1988), STEMS (Belcher and others 1982) and OAKSIM (Hilt 1985) can be used to project stand changes resulting from real or hypothetical management practices. However, none of these models can predict stand structure or composition of a new stand. During the first two decades after final harvest, rapidly changing tree growth and competition relations determine the character of the mature stand.

The oak reproduction that develops after clearcutting depends on the oak regeneration potential of the parent stand. Regeneration potential refers to the capacity of a species or species-group to occupy and dominate growing space at a specified time in the new stand. For oaks, it can be determined from preharvest characteristics of the parent stand, including the advance reproduction and the overstory. To quantify the regeneration potential of a stand, the various sources of reproduction thus must be considered.

Natural reproduction after final harvest may come from new seedlings, advance reproduction, and sprouts from stumps of overstory trees (Beck 1980). New oak seedlings usually grow slowly and are not a major source of reproduction in most ecosystems (Sander and Clark 1971, Sander 1972, McQuilkin 1975). Viable acorns remain in the forest floor for 6 months or less because they are rapidly consumed by wildlife, damaged by insects and pathogens, and dessicate or freeze during the winter.

Advance reproduction and stump sprouts usually are the major sources of reproduction. The competitiveness of upland oak reproduction usually depends on the development before final harvest of a large root:shoot ratio. Oak advance reproduction can develop large root:shoot ratios through repeated shoot dieback and resprouting. Shoot dieback may result from water stress, weather damage (i.e., frost and freeze), insects, pathogens, fire, and browsing. Nevertheless, oak reproduction may accumulate and develop in the understory for decades (Merz and Boyce 1956, Tryon and others 1980). Once the overstory is removed, larger stems of advance reproduction can sprout vigorously and thus capture much of the newly available growing space (Carvell 1967; Sander 1971, 1972, 1979; McQuilkin 1975; Beck and Hooper 1986; Ross and others 1986; Johnson and Sander 1988; Loftis 1988). Thus, it is the number, size, and spatial distribution of oak advance reproduction that largely determines the oak regeneration potential of a stand.

Sprouts from cut stumps of overstory trees are usually the fastest growing form of oak reproduction (Spaeth 1928, Kuenzel 1935, Sander and Clark 1971, Smith 1979, Zahner and Myers 1984). The large food reserves and absorptive capacity of the parent root system support rapid shoot growth of stump sprouts. Stump sprouts make up variable portions of the total oak reproduction and often compensate for deficiencies in oak advance reproduction (Johnson and Sander 1988).

There are several hardwood regeneration models for predicting stand development after final harvest. Predictions are based on preharvest inventories of advance reproduction, the overstory (trees larger than 1.6 in. d.b.h.), and site factors. These models thus provide land managers the opportunity to evaluate alternative prescriptions before actual harvest. Methods for evaluating the natural regeneration potential of upland oak forests have been developed by Sander and others (1976, 1984), Johnson (1977), Johnson and Sander (1988), Lowell and others (1987), Waldrop and others (1986), Loftis (1988, 1990) and Marquis and Ernst (1988). Models for evaluating the potential contribution of planted trees to future stocking also have been developed for northern red oak (Quercus rubra L.) (Johnson 1988) and other hardwoods (Johnson 1984, Johnson and Rogers 1985).

Marquis and Ernst (1988) developed an expert system for Allegheny hardwood forests called SILVAH. It generates prescriptions based on management objectives, present stand conditions, and projections of tree growth and regeneration potential (Marquis and others 1992). To evaluate even-aged stands, critical stocking values are used to predict regeneration success. In applying the system, advance reproduction is inventoried on plots 6 ft. in radius. Whether a plot is stocked depends on species, numbers of stems, and their size. For example, a plot is considered stocked with oak if there are at least 25 stems less than 4.5 ft. tall, or if there is one stem greater than 4.5 ft. tall. Oaks in the overstory also are inventoried. The number of reproduction plots expected to be stocked with oak stump sprouts at stand age 20 are estimated using models developed by Sander and others (1984). During the preharvest inventory, limiting site factors (poor drainage and stone content), amount of interference species competition, and intensity of deer browsing also are noted. Because these factors reduce the regeneration potential of a stand, the number of plots deemed stocked with desirable reproduction are accordingly reduced. The number of plots stocked with desirable tree species thus determine whether or not the regeneration potential is adequate. In general, when 70 percent of the plots are stocked with desirable advance reproduction, successful regeneration is expected to occur.

Loftis (1990) developed a regeneration model for mixed hardwood forests of southern Appalachia. The model projects eighth-year post-harvest heights and crown classes of northern red oak advance reproduction based on preharvest measurements of advance reproduction and site quality. Heights of red oak reproduction 8 years after final overstory removal are estimated from preharvest height and basal diameter of advance reproduction and from site index. Preharvest basal diameters and site index are then used to predict the probability that a given stem of oak advance reproduction grows to dominant or codominant crown classes in the new stand at age 8. This model thus can be used to evaluate the contribution of red oak reproduction to the future stand.

Sander and others (1984) developed a regeneration model for the Missouri Ozarks. It predicts the probability of individual oaks surviving and attaining a specific future height that places them in codominant or dominant crown classes. From preharvest inventories of the overstory and advance reproduction, the model predicts the adequacy of future oak stocking. A stand is deemed to have adequate regeneration potential if projected stocking of dominant and codominant oaks at stand age 20 is at C-level or greater based on Gingrichs (1967) stocking relations. However, the model is unable to predict the diameter distribution of oaks in the new stand, and non-oaks are not considered. Although species such as blackgum (Nyssa sylvatica Marsh.), sassafras (Sassafras albidum (Nutt.) Nees, hickory (Carya spp.), and flowering dogwood (Cornus florida L.) seldom occur as dominant trees in Ozark forests once they reach 20 years of age, they do affect the development of stands during the first two decades after final harvest because of their high density and rapid growth. These species are also important to wildlife and biodiversity.

In contrast to some other deciduous forests of the eastern United States, the forests of the Missouri Ozarks regenerate primarily from sprouts of harvested overstory trees and advance reproduction. Therefore, the composition and size structure of the future stand is largely determined by the species composition and size structure of the preharvest advance reproduction and overstory. By observing these features together with measurements of site factors such as slope, aspect, and site index, model users can predict the composition and structure of the new stand.

Dey (1991) developed a regeneration model called A Comprehensive Ozark Regenerator (ACORn) to simulate the regeneration of even-aged stands in the Missouri Ozarks. It was developed from measurements of individual stems of reproduction before and after clearcutting. ACORn comprises two modules that simulate the development of the two primary sources of reproduction in that ecosystem: stump sprouts and advance reproduction. Each module, in turn, contains models for estimating future heights, diameters, and survival of individual stems by species. The probability of survival of individual stems of both advance reproduction and stump sprouts, from preharvest to a specified future stand age, is estimated from initial tree size and site factors. Species-specific models estimate future tree heights and diameters from similar predictors. The predictive models, in turn, facilitate the generation of future diameter and height distributions of surviving reproduction including stump sprouts and advance reproduction by species. Projections for hickory, flowering dogwood, blackgum, and sassafras also can be obtained.

In application, preharvest inventories of the overstory and advance reproduction provide input to survival and growth models that generate diameter distributions 21 years after clearcutting. At that age, mean stand diameter averages 3 in. d.b.h. The diameter distributions then can be used to summarize stand characteristics, such as basal area per acre, stems per acre, and percent stocking by species. From this stand summary, the adequacy of future stocking can be assessed and used to develop appropriate silvicultural prescriptions. The resulting diameter distributions also can be used as input into growth and yield simulators such as TWIGS.

Quality of reproduction is commonly defined in terms of the acceptability of growing stock. Before adequacy of regeneration potential can be assessed, it is necessary to define acceptable growing stock, which requires a consideration of species and tree characteristics. Commonly used tree characteristics are height, diameter, and crown class. Although a qualitative characteristic, crown class is a concept with which most foresters are familiar. It is a widely used descriptor of trees because it facilitates visualization of a tree's social status more easily than diameter or height measurements, per se. For example, the definition of acceptable growing stock may be limited to certain species and trees that occupy only codominant and dominant crown classes. ACORn solves that problem by projecting future diameter and survival of individual trees by diameter and crown classes to stand age 21.

Some foresters may consider only codominant and dominant trees as acceptable growing stock. Others may consider trees that are intermediate or larger as acceptable. To integrate this decision into the regeneration model, a threshold tree diameter is defined that classifies trees into one of the two user-defined crown class categories: acceptable or unacceptable. When acceptable growing stock includes only codominant and dominant trees, tree diameters equal to or greater than 3.8 in. are classified as codominant or dominant at stand age 21. When acceptable growing stock is defined as trees that are intermediate or larger, trees with diameters less than 2.6 in. are classified as suppressed, and those with larger diameters are classed as acceptable growing stock. In this way, ACORn incorporates crown class to project twenty-first-year diameter distributions.

Data from preharvest inventories of the overstory and advance reproduction provide the necessary input for ACORn projections of future stand composition and structure. To inventory a stand, the overstory (trees 1.6 in. d.b.h. and larger) is sampled separately from the advance reproduction (trees less than 1.6 in. d.b.h.). Because ACORn is an individual tree model, it projects the survival and growth of single trees and then expands this to a per-acre basis in the form of stand tables. The model thus can accommodate a variety of sampling designs.

ACORn requires an inventory of the overstory including species, d.b.h., site index (black oak, base age 50), and stand age. When the overstory is even-aged, an average stand age can be substituted for individual tree age. For advance reproduction, the required inventory includes data on species, basal diameters, and heights of reproduction, slope position (upper, middle, or lower), and aspect. A computer program is available to facilitate application of the model.

Fixed-area or variable-radius plots can be used to inventory the overstory. Sander and others (1984) recommended that the overstory be sampled on 1/20-acre plots in their regeneration guide. Small fixed-area plots were recommended for inventory of advance reproduction. The model of Sander and others (1984) requires that advance reproduction be inventoried on 1/735-acre plots. The choice of sample plot size and number should be done on a stand-by-stand basis. There are numerous sampling techniques that can be used to design efficient and effective inventories (e.g., Freese 1962).

In general, the more variation there is in species composition, tree size, and the spatial distribution of trees, the more sample plots that are needed. It is usually best to sample a large number of small plots where stand variation is great than it is to sample fewer large plots. A common rule-of-thumb is to inventory at least 30 plots regardless of stand size. The upper limit to the number of plots depends on (1) constraints such as budget, crew size, and timeframe; (2) intended use of simulation, i.e., forest planning or stand prescription development; (3) amount of stand variation; and (4) desired level of precision. Plots should be distributed randomly or systematically throughout the stand.

Johnson (1980) developed a hardwood regeneration prediction model for bottomland hardwoods. The model evaluates regeneration potential by assigning points based on number and size (height class) of advance regeneration and sprout potential of stumps from severed trees. Research is still underway to test the method and make refinements if necessary. Current data indicate that 78 percent of plots evaluated as adequately stocked before harvest were still adequately stocked with desirable regeneration following harvest. Preliminary results--describing seedling mortality, seedling growth by size (height and root collar), and logging losses--indicate that improvements to the method can be made through modifications of point assignments.

• Introduction
• The Prediction Model
• Testing the Prediction Model
• Results and Conclusions
• Recommendations
  
  

Through continuing and widespread research efforts, there now exist predictive models to evaluate the potential for oak regeneration success prior to harvest of upland hardwood stands (Dey 1991, Loftis 1990, Marquis and Bjorkbom 1982, Sander and others 1976, Waldrop and others 1986). The foundation underlying the use of these evaluation techniques is that, to successfully regenerate oaks, there must exist adequate numbers and/or size of advance oak regeneration prior to harvest. These models were designed for use within specific physiographic regions and application elsewhere should be approached with caution.

The harvest and regeneration of bottomland hardwoods has been a fascinating subject for many years. Hardwood regeneration in the bottoms is often easy, but almost as often, the successful regeneration of bottomland hardwoods, especially oaks, can be very difficult. Forest land managers have for many years needed to predict when, where, and how their hardwood regeneration efforts would be successful. In 1980, a hardwood regeneration prediction model for bottomland hardwoods was devised (Johnson 1980). The purpose of this paper is to examine how the Johnson method can be used to predict oak regeneration, and based upon preliminary testing, how the procedure can be improved.

The regeneration prediction method when used within the constraints discussed later has been found to be both useful and reliable. However, we should emphasize that the expected results for any given situation are based upon a combination of empirical results and the senior authors experience. The authors believe that with on-the-ground experience many users can adapt the procedure to more closely fit their needs and conditions.

For bottomland oaks, our target species in this paper (see Table), we know, for example, that acorn production varies considerably among species, years, and phenotypes. Good acorn crops may be 3 to 5 years apart. Acorns are often utilized by wildlife after seedfall and a good seed crop may not produce a good seedling crop. In some years and on some sites, acorns of bottomland oaks may be protected from birds and animals when they drop into or are soon covered by water that stays in the forest until early in the growing season. Duration of the water cover is important. If the water leaves before germination, acorns will be exposed and eaten; if the water stays into the summer, acorns will rot. Water that recedes during the germination period usually results in several days of moist seedbed that allows germinating acorns to establish seedlings. Without surface water and given that radicles cannot penetrate a dry, hard surface soil, acorns that roll into soil cracks or that are buried by animals or birds (Deen and Hodges 1991) have the best chance to establish a new seedling.

(Table:Ranking of Desirable Species; i.e. Oak and Ash)Oak seedling populations may be large in number beneath the crowns of mature oak trees; e.g., Johnson (1975) found up to 100,000+ nuttall oak (Quercus michauxii Nutt.) seedlings per acre in the Mississippi Delta. Red oak seedlings in full shade will normally grow approximately a foot in height during the first year, and most die within 3 years. Those that receive 2 hours of direct sunlight daily may survive beyond 3-5 years, but grow only an inch or so in height annually. Oaks that die back and resprout almost always start in small openings where there is enough sunlight to allow for 5-10 ft. of height growth before the overstory closes and direct sunlight is shut off. These larger trees have a well-developed root system that will support rapid shoot growth when the overstory is removed.

Once released, mortality of oak seedlings is most likely to occur the first year. Flooding is a likely cause of mortality, but exposure to frost may result in dieback or mortality (McGee 1988). Thereafter, at least through year 10, three of four oaks should survive even with overtopping competition. With only side competition, oak regeneration will continue growing upward and have a good chance to eventually become dominant.

Height growth of released seedlings <1 ft. tall is slow for the first 5 years, whereas taller advanced reproduction and stump sprouts grow rapidly from the start. For bottomland oaks, stumps up to 12 in. in diameter have a good chance of producing sprouts that are acceptable regeneration; larger stumps usually do not produce sprouts, or if they do, the sprouts are likely to die. Because of the length of time necessary to produce quality oak lumber, oak stems will grow beyond a threshold-diameter capable of producing many, if any, stump sprouts. Also, the total number of sawlog-sized oak stems per acre in a mature, bottomland hardwood stand would be low enough to discount these stumps as providing adequate stocking of oak sprouts for regeneration purposes. Sawlog-sized oak, then, can be mostly discounted as an adequate source of sprouts from which to restock the newly developing stand (Johnson 1975).

The method proposed by Johnson (1980) is based on the silvical characteristics of the species as discussed above. It is a numerical evaluation of regeneration potential which emphasizes the size (height class) and numbers of advance regeneration and the contribution of stump sprouts to regeneration potential (see Figure). Weights and points given various regeneration components are based on the senior authors 35 plus years of research experience. The following procedures are used in the evaluation process:

1. The sample plot is a circular 1/100th acre. This plot size will accommodate large trees (because stump and root sprouts are considered as reproduction) as well as seedlings and saplings. The selected size also provides an easy blow-up factor.
2. One point for each tree <1.0 ft. tall. Mortality is high among such trees because they are only a year or two old and are not well established. In fact, all trees <1.0 ft. tall may die if overstory removal is delayed for more than a year. However, a few small trees may be older, well established, and likely to survive several years with or without overstory release. The one point per tree is an attempt to balance the survival potential of young and older, established trees. Young or old, small trees are not very competitive.
3. Two points for each tree 1.1 to 2.9 ft. tall. Trees of this size are older than 1 year and have a well-developed root system that will aid survival and provide for height growth. Trees in this size class are more competitive than trees <1.0 ft. tall, but less competitive than those >3.0 ft. tall.
4. Three points for each tree >3.0 ft. tall but less than 5.5 in. d.b.h. Trees are able to grow into this size class while in small openings but not in full shade. Thus, the trees are several years old, have large root systems, are almost certain to survive an overstory harvest even if their tops are broken off, are good stump sprouters, and should have rapid post-harvest height growth. Trees in this class make good competitors.
5. Two points for each tree 5.6 to 10.5 in. d.b.h. A high percentage of stumps from trees in this size class will produce competitive sprouts that survive and grow very well. Usually no more than three stumps of this size will occur in a 1/100th- acre plot.
6. One point for each tree 10.6 to 15.5 in. d.b.h. Not many of these relatively large stumps will produce acceptable, competitive sprouts. No more than one or two trees of this size class will occur in a 1/100th-acre plot.
7. Twelve points per plot is considered high enough to ensure adequate regeneration following harvest. The points could represent 12 trees per plot or 1,200 trees per acre that are a foot or less in height, 4 trees per plot (400 per acre) 3.0 ft. tall to 5.4 in. d.b.h., or from any combination of trees equalling 12 points. Although natural stands may start with 20,000 plus trees per acre, they will thin to fewer than 400 crop trees per acre at minimum commercial size (6 in. d.b.h.). The 12 points is a judgment call, believed conservative, and should be considered only as a general guide.

(Table:Ranking of Desirable Species; i.e. Oak and Ash)A possible limitation of the guide is that it does not recognize seedlings which germinate after harvest. Oak seedlings may become established after harvest and become an important part of the regenerated stand, but, this model does not provide for such a happening. If advance regeneration is lacking and if there are no oak stumps of sprout-producing size, this model predicts zero oaks in the regeneration stand. Also, the total number of stocked plots needed for successful regeneration of a given tract is still questionable. It appears that at least 60 percent of the total plots must be stocked (>12 points) and well distributed over the tract to ensure successful regeneration.

Research is currently underway to test and evaluate the regeneration prediction method described herein.

1. Study areas 10 acres and larger were located mainly in the floodplain of minor streambottoms.
2. Permanent 1/100th-acre circular plots were systematically located within each study area. Woody vegetation was placed into one of three size classes and identified as either desirable (oaks and ash) or undesirable (all other woody species).
3. To develop a priority rating for predicting regeneration potential of oaks and ash, desirable stems placed in seedling or sapling categories were identified and tagged. Stems could be identified from year-to-year, enabling development patterns to be detected and related to preharvest size.
4. Desirable and undesirable vegetation was identified by species and placed in the appropriate height class. In addition, permanent trees were selected and measured for root collar diameter. All woody vegetation was designated as to origin class; i.e., sprout or seedlings. Desirable and permanent sample trees were noted as to occurrence of top die-back, insect or disease damage, damage due to harvesting operation, and subjectively rated as to competitive status, i.e., free-to-grow, medium competition from surrounding vegetation, or overtopped.
5. Regeneration was evaluated for each plot according to Johnsons (1980) technique (see Figure) before harvest and after each remeasurement to determine the adequacy of the technique and, if needed, make adjustments to the guide to more accurately predict regeneration success.(Table:Regeneration Potential)
6. Following harvest, a resurvey of the plots was conducted to quantify logging damage, if any, to permanent sample trees and to the site that would adversely affect subsequent vegetative composition and growth and development of the future stand. At this time, a survey was taken of the number of light-seeded species occurring within 200 ft. of each plot. Seedlings from these fast-growing, intolerant species could ultimately suppress the development of desirable species, oaks and ash, in the future stand.
7. Other factors affecting stand development (i.e., high water, severe vegetative competition, etc.) were recorded at each plot visit. Also, newly germinating oak or ash species were marked with a combination of colored expansion rings, designating a particular height class and a particular year of germination, so further development can be followed.

There were 118 plots established on nine bottomland hardwood tracts in east central and south Mississippi. The tracts used in the study were logged between August and September of 1989 and 1990. This Table 1 gives the total number of desirable stems, by species, that were initially sampled in this study.(Table:Ranking of Desirable Species; i.e. Oak and Ash)

Based on this Prediction Model, the sampled stands shown in Table 2 exhibit differing trends of stocking success following harvest. The decline in percentage stocking of those stands shown in Table 2 can be attributed to the high degree of mortality, due to logging, of oak regeneration <1.0 ft. tall (Table 3). Seedlings <1.0 ft. tall accounted for the bulk of points assigned to plots within these stands before harvest, and the loss of these seedlings following harvest resulted in lower point totals for these plots. Accordingly, the Prediction Model may need to be revised to reflect the need for greater numbers of regeneration <1.0 ft. tall for adequate regeneration when this is the only source of regeneration. (Table:Percentage of Plots Adequately Stocked, with Oak and Ash)

The one stand (U4KTR, Table 2) that showed an increase in stocking percentage following harvest contained a majority of desirable stems >1.0 ft. tall before harvest. Desirable stems, 1.1-2.9 ft. tall and >3.0 ft. tall, have a much greater chance for survival(Table 3). Also, in this stand individual stems have progressed in height class, increasing the points allocated to that particular seedling, and likewise increasing the point total for that plot. Four stands remained constant as far as percentage stocking is concerned. Two stands appear adequately stocked, one stand is borderline, and one stand is considered nonstocked. (Table:Percent Morality and Survival of Desirable Species Seedlings, i.e. Oak and Ash)

One point we are trying to determine is the cut-off point where a stand should be classified as suitable or not for regeneration based on initial stocking. Twelve points is now considered the threshold. However, some plots that do not make 12 points contain several stems that are large enough to successfully compete for growing space and some day become a component of the overstory. If that is the case, plots containing less than 12 points but containing larger regeneration may be adequately stocked. Observation of stand development over time is needed to clarify this point.

Other significant results include the following:

1. Logging damage following harvest was assessed using the permanent sample trees located within each plot. Saplings and/or seedlings were categorized as to the extent of logging damage, i.e., intact, bent, broken, missing, or dead. Stems that were classified with severe logging damage (i.e., broken, missing, or dead) which were dead at the next remeasurement were assumed to have died as a result of logging damage (Table 3). Some stems in the sapling class, 2.0-4.4 in. d.b.h., were pushed completely out of the ground exposing their rootstocks and ultimately resulting in death. Loss of this type of regeneration can be reduced by chainsawing of saplings prior to harvest.
2. Development of oak regeneration after clearcutting followed similar patterns to those reported in other regions (Loftis 1983, Sander 1972) where large advance regeneration exhibited better survival than small advance regeneration. Survival data (Table 3) indicate that desirable seedlings <1.0 ft. in height cannot be counted on to provide many stems to the newly regenerated stand. Also, field data and observation showed that these seedlings <1.0 ft. tall which did survive had large root collar diameters, indicating that such stems had probably died back and resprouted several times. The larger rootstocks enabled these smaller stems to produce a vigorous, competitive sprout. Stems in the 1.1-2.9 ft. and >3.0 ft. height classes exhibited much better survival rates following harvest. It appears these two size classes will contribute the bulk of the stems to the new regeneration stand.
3. One growing season after harvest, 162 new oak and ash germinants, < 1.0 ft. tall, and seven new ash germinants, 1.1-2.9 ft. tall, were found. Forty percent of the seedlings <1.0 ft. tall survived their first growing season while six of the seven ash germinants 1.1-2.9 ft. tall survived. In the second year following harvest there were 22 new germinants of oak and ash which were <1.0 ft. tall, and six new ash seedlings, 1.1-2.9 ft. tall. They have yet to be evaluated after a full growing season so their fate is undetermined. However, most research indicates that contribution of the new seedlings will be insignificant and that advance regeneration will provide the bulk of the stems to the new stand.
4. Seedling height growth was as follows: seedlings <1.0 ft. tall averaged 7.7 in. tall preharvest. One year after harvest, average height for these seedlings was 10.9 in., and after 2 years average height was 17.7 in. Seedlings 1.1-2.9 ft. tall, similarly, realized a 10.0-in. increase in average height over 2 years, growing from 20.6 to 30.6 in. However, the desirable stems in the <1.0 ft. and 1.1-2.9 ft. height classes are competing with desirable stems from the >3.0 ft. height class, as well as undesirable stems, for growing space in the newly developing stands. A number of stems in these height classes are free-to-grow and may eventually achieve a dominant position.
5. Seedlings >3.0 ft. tall before harvest showed a decrease in average height after the first growing season, from 82.2 in. to 47.3 in. This was primarily a result of logging damage, but there is evidence that exposure to low temperatures in the spring following release can result in top die-back (McGee 1988). After the second year, average height increased to 60.1 in. This is a result, mainly, of the vigorous sprout growth of these larger stems of advance regeneration. Some stems were also able to respond rapidly to the increased light conditions created by clearcutting. A majority of these stems are already in a free-to-grow position and have assumed a dominant position in the new stand. Their vigorous, rapid height growth may assure a dominant crown position as the stands develop.
6. Preliminary data show that stump sprouting for trees >1.5 in. d.b.h. were as follows: 45.6 percent for the 2.0-5.0 in. d.b.h. class, 42.9 percent for the 6.0-10.0 in. d.b.h. class, 33.3 percent for the 11.0-15.0 in. d.b.h. class, 50.0 percent for the 16.0-20.0 in. d.b.h. class (only two trees in this class), and the >20.0 in. d.b.h. class had no sprouting. The trend seems to be a decrease in sprouting with an increase in tree (stump) size as shown in other studies (Johnson 1977, McQuilkin 1975). Also, as trees increase in size their numbers decrease on a per-acre basis and, coupled with their reduced sprouting ability, further reduces their contribution to the regeneration stand.

A model that predicts regeneration success was developed for southern bottomland hardwood forests, but should be suitable throughout the southeastern United States. It is applicable only where complete overstory harvest is intended, meaning that all trees 2 in. d.b.h. and larger are cut from an area sufficient in size to allow full sunlight to reach the ground. Because the number and vigor of understory oaks change rapidly, prediction by the model is useful for 1 year only. If harvesting is delayed, the stand should be resampled.

Based on experience, it appears that one sample plot per acre might be needed for tracts under 50 acres, whereas for larger areas one plot per 2 or 3 acres may do. Stratified sampling is suggested where different stand conditions can be recognized and delineated. A systematic sampling scheme is practical and adequate.

The model is not refined enough to predict which tree, among the many mixed species, may eventually dominate. But, in the same growth environment (same growing space and micro-site) a tree with three points is much more likely to survive and compete than a 1-point tree. Growth environments differ, however, even over a foot or two in space, so there may be considerable variation in the way one 3-point tree compares to three 1-point trees in long-term development among competitors.

A point total of less than 12 per plot doesn't necessarily mean the sample area will not regenerate satisfactorily, but it does signal caution. For example, it seems likely that if oaks contribute only two points per plot, then they will not be a major part of the new stand. On the other hand, large advance regeneration appears highly likely to survive the rigors of harvesting to become a component of the new stand. Possibly, only two large stems per plot may be necessary to ensure successful regeneration. A plot with more than 12 points, made up mostly of seedlings <1 ft. in height, may not be adequate for regeneration due to heavy mortality among seedlings in this height class. More seedlings <1 ft. in height may be needed to qualify plots as adequately stocked.

David A. Marquis, USDA Forest Service, Retired, Homosassa, FL 34446
Mark J. Twery, Northeastern Forest Experiment Station, USDA Forest Service, Morgantown, WV 26505

Failure to obtain prompt regeneration of desired species after a harvest cut can leave a stand unproductive for many decades, cost excessive amounts to reclaim through artificial means, and severely limit the suitability of the stand for a wide range of forest values. But prescribing silvicultural treatments that ensure successful regeneration most of the time is a difficult task. Forest managers need to consider all the many factors that could affect regeneration success, weigh the many available silvicultural techniques available to accomplish the task, consider the characteristics of the particular stand under consideration, and make an informed decision that is most likely to achieve the desired conditions. In this paper, decision-making procedures are described that provide a systematic way to analyze stand potential and prescribe regeneration treatments. In the Allegheny region where the system has been used extensively, successful regeneration has been obtained in over 90 percent of the stands harvested, compared to only about 50 percent prior to development of this system. The procedures can be adapted for use in any geographic region or forest type.

• Introduction
• Decision Charts
• Decision Criteria
• Data Requirements

Introduction

Natural regeneration of oaks is difficult to obtain in the northern forest ecosystem, just as it often is in most other parts of the oak range. To the common problems resulting from infrequent acorn crops, acorn depredations, slow juvenile oak seedling growth, and intense competition from other fast-growing plants, northern ecosystems add additional obstacles. oak stands in the north are found on sites better adapted to northern hardwoods, and the natural tendency is for them to revert to northern species since fire and grazing have been removed as factors favoring the oaks. In addition, many northern ecosystems support excessive deer populations that make regeneration of any woody species difficult.

Research now underway is providing much information that will help us develop silvicultural procedures to perpetuate oaks. We hope that this research will also define the range of sites where the battle to maintain oaks--rather than allowing conversion to northern species--is economically and ecologically feasible. There is still a long way to go before reliable procedures to regenerate oak on good northern sites are available, but progress is being made.

One point is already quite clear, however: there will be no simple or universal treatment that guarantees consistent oak regeneration across the wide range of stand and site conditions found throughout the oak region. Instead, success will depend upon the careful prescription of treatments tailored to each individual situation. Stand and site conditions and all potential obstacles to oak regeneration will need to be evaluated systematically before a specific treatment is recommended. This decision-making process will require considerable knowledge and judgment on the part of the forest manager.

About 12 years ago, a systematic procedure for making silvicultural prescriptions for individual cherry-maple, northern hardwood, and oak stands in the Allegheny region was developed. That procedure has been refined as new knowledge has accumulated. This stand evaluation and prescription process, known as the SILVAH system (Marquis and others 1992, Marquis and Ernst 1992), has been widely used in the Allegheny region, where it has helped forest managers consistently prescribe successful treatments in areas where deer browsing and other factors had traditionally prevented regeneration. That basic decision-making procedure is now being expanded to cover other forest types and geographic regions of the northeastern United States (Marquis 1991a). In this paper, decision-making procedures weve developed for regeneration prescriptions in the Northeast are described. The process could be applied anywhere in the oak region if decision criteria appropriate to the geographic area are substituted for the Northeastern criteria. These decision-making procedures are also being expanded within the Northeast Decision Model (NED) for purposes beyond timber production. The successful establishment of new young trees after disturbance is important for all values derived from forests. Although the decision criteria may differ, the same data and decision procedures may be used in making regeneration decisions where wildlife, aesthetics, water, or other forest values are the management goal.

Decision Criteria

When using the decision charts, one must answer each decision question either "Yes" or "No." Although the questions and answers appear subjective, it is the intent of the system that these determinations be made as objectively as possible, on the basis of specific parameters that are measured in the stand under question. This BROKEN-LINK BROKEN-LINK table illustrates these parameters, and the levels that qualify for either a "Yes" or "No" answer (from Marquis 1991b). The decision questions are listed in the approximate sequence they are encountered in the decision charts. The example table is from the Allegheny region. Similar tables are being developed for other geographic regions, and for situations where success requires the regeneration of specific target species (such as oaks).

The question "Is sunlight limiting to seedling establishment?" illustrates how specific stand parameters are used to arrive at an objective answer. The stand parameter used to evaluate the density of the overstory and thus the amount of sunlight reaching the forest floor is relative stand density or stocking percent. This measure works well in stands of any species composition, is easily calculated from standard cruise data, and requires no subjective judgment. In stands with more than 75 percent relative density, the overstory is dense enough that sunlight may limit seedling establishment. At lower densities, seedlings can usually become established, although shade-intolerant species may not grow much there. So, 75 percent relative density is the breaking point; at densities above 75, sunlight is considered limiting and shelterwood cutting becomes a possible treatment. At densities below 75, light is probably adequate for establishment and any lack of advance seedlings is probably due to other limiting factors.

Some of the decision questions require much more complex criteria to evaluate adequately. The amount of advance regeneration that is adequate to ensure successful reproduction following harvest is a good example. In the Allegheny region, adequate advance regeneration requires that at least minimum numbers of seedlings be present over at least 70 percent of the stand area. The minimum numbers of seedlings depends on the species, their size, and the amount of deer browsing expected. The table, below (from Marquis and others 1992), shows the weighted minimum numbers of seedlings required on each plot, as a function of species and deer impact index. Weighted numbers are derived by counting (or estimating) seedling numbers in each of three size classes, and applying weights to each size class to arrive at a weighted total number. Larger seedlings have a better chance of survival, and thus carry a higher weight than small seedlings.(Table:Number of Seedlings required for regenration plots to be stocked)

The deer impact index is a value determined from a chart showing the estimated deer population (in deer per square mile) for various habitat conditions. Deer population estimates may be obtained from the state game agency, or from a sample census conducted for this purpose. Habitat condition is determined from the proportions of the 1-square-mile area surrounding the stand in question (the deer home range roughly) in such conditions as: agricultural fields, open land, recent harvest cuts, thinned and unthinned forest, etc. All of these vegetative conditions affect the amount of deer food available in the vicinity to reduce browsing pressure on regeneration in any new harvest cuts.

Thus, the evaluation of advance regeneration adequacy is based on the percentage of understory that meets certain required minimum conditions. These conditions vary tremendously depending on such factors as tree species, seedling size, and deer pressure. To illustrate: in the Allegheny region, the minimum number of maple or ash seedlings per 6-ft. radius plot is 100, if deer pressure is high (level 4) and all seedlings are 2 in. to 1 ft. tall; but, the minimum number is only 7 or 8 if deer pressure is low and all seedlings exceed 1 ft. in height. Since the decision question in the chart considers both advance regeneration and sprouts, the sprouting potential of all trees to be cut must also be evaluated. Then, estimates of the proportion of the area that will be regenerated by sprouts must be added to the advance regeneration estimates.

Thus, a simple "Yes" or "No" answer in this case involves a whole host of factors that must be carefully evaluated to make effective use of the decision charts. Other decision criteria, such as the density of interfering plants and seed supply adequacy, are evaluated in a similar manner. Several factors must be included in determining whether or not these factors are limiting to or are adequate for regeneration. Details on these and other factors for the Allegheny region are presented in the SILVAH literature (Marquis and others 1992), and are not repeated here.

Developing suitable decision criteria for a specific forest type and geographic region is a major task. In the Allegheny region, it took nearly 15 years of research aimed specifically at that goal to complete the SILVAH system. Information needed to develop decision criteria for other regions and forest types that have received less focused research effort will not always be available. Nevertheless, many of the general principles and procedures from the Allegheny region can be used elsewhere, and experienced silviculturists can probably make informed estimates of the critical parameters in their areas. Such estimates provide a starting point that is certainly better than no system at all. At the very least, it forces the decision maker to systematically consider all of the many factors that affect regeneration success, and to make a judgment on each based upon the best information currently available.

Furthermore, the system itself serves as a model or framework on which to evaluate research needs. Those factors or criteria on which knowledge is inadequate become quickly apparent as one attempts to define the decision questions and criteria. The relative importance of each research need also can be assessed in terms of its relative importance and frequency of occurrence in the overall oak regeneration process.

The decision charts presented here are intended for general use over a wide range of forest types and geographic areas. As a result, some of the decision points and decision criteria will be of little importance in some types or regions. Rather than creating a whole series of specialized charts that eliminate these factors entirely for the charts for that region, it is best to simply set the criteria at a level that will effectively ignore the factor. For example, deer browsing may not be a factor at all where deer populations are low across a wide geographic area. In these cases, the criteria table can simply suggest that deer impact index be set to 1 (very low) or 2 (low) anywhere in that region.

Additional decision criteria are being formulated to extend this system of decision making to a wide variety of resource values. For example, visual qualities within a forest stand are highly dependent on numbers and sizes of woody stems close to the ground, as is habitat quality for many wildlife species. Criteria that ensure the desired stems densities for these objectives fit very comfortably in this system.

Data Requirements

To function consistently, any stand analysis and prescription system must have reliable data on which to answer the decision questions. In SILVAH and the NE Decision Model, those data come from a cruise in the stand under consideration. Data on overstory and understory vegetation and site factors are collected and summarized for use in the decision charts. Computer programs ease the calculation job and permit extensive analyses to be performed as needed. The amount of detail collected in the cruise can be varied to meet the needs of a variety of organizations and individuals, although there are certain essential data items that must be collected in all cases. The recommended cruise procedure involves:

1. A variable-radius (prism) cruise of the overstory, recording trees by species and diameter at the minimum, with other observations on individual trees as desired (such as merchantable height, timber quality, grade, defect, wildlife value, etc.).
2. A fixed-plot sample of understory conditions, with several options in the amount of detail recorded. At the minimum, a simple checkmark is used to indicate the presence or absence of 15 to 20 specified conditions (such as the presence of various categories of advance seedlings and interfering plants). Estimated numbers or percent of coverage may also be recorded for each of these classes. Or, complete enumerations may be made of all vegetation by species and size classes.
3. A fixed-plot sample of site conditions. At the minimum, this involves a checkmark to indicate the presence of certain site factors known to limit regeneration. It may also include site index or site class determination and recording of other important site factors.

Because understory and site conditions vary considerably within a stand where overstory conditions are uniform, we recommend that twice as many understory plots be sampled as overstory plots. Understory and site data can easily be collected at the same time as overstory data. Understory and site plots are taken at each overstory plot, and additional understory/site plots are located half way between each overstory plot.

When the SILVAH system was first introduced, there was some reluctance on the part of forest managers to collect understory and site data. Traditional cruises included only the overstory (often only the merchantable size trees), and it was assumed that the extra data collection would add considerably to cruise costs. However, studies by both the Allegheny National Forest and Hammermill Paper Co. revealed that collecting the minimum understory and site data did not add much to the time required. Recording simple checkmarks at each understory plot took only a few minutes per plot, and was insignificant compared to the time required to lay out cruise lines, walk from plot to plot, and collect prism data.

Of course, more detailed understory data collection would add significantly to cruise time and costs. Although the minimum data collection provides sufficient information for prescription preparation, extra data may be desirable for a wide variety of reasons. Each organization must tailor data collection to its own needs and budget. The system is designed in such a way that it can utilize the added data if available, but can still function for writing prescriptions with the minimum data.

Decision Charts

NE Decision Chart 8.1z

In these charts, specific symbols are used to denote their function in the decision chart. Circles denote the start of the process. Five-sided boxes are connecting symbols that route you to (or from) other charts. Arrows denote the sequence of travel within the chart. Horizontal lines with a short question above the line represent the main decision points in the chart. Rectangular boxes at either end of the decision question represent the possible answers in this case, either "Yes" or "No." Ovals represent prescriptions, or recommended treatments; once you reach an oval, you have completed the decision process for that stand. Note in the legend, below, that some prescription ovals are shaded or cross-hatched, indicating different degrees of confidence in the recommendation and/or various levels of investment required.

Target Species

Portions of the decision process that led to Chart 8 are omitted because they do not involve regeneration considerations. However, to arrive at Chart 8, you would have had to determine that even-age silviculture was the appropriate silvicultural system to use in this management unit. In the NE Decision Model, eight separate silvicultural "systems" are recognized, of which traditional even-age silviculture is number eight. This decision on the overall silvicultural system to be used is based primarily on the specific management goals selected, plus the forest type.

The eight systems in the NE Decision Model are viewed as a continuum on two different directions: intensity of disturbance and frequency of disturbance. The even-age system described in Chart 8 is at one end of each continuum of intense but infrequent disturbance. Of course, the system selected is very important in determining which of the potentially desirable species are likely to succeed under a prescription made by the model. Tracing a few prescriptions through the decision chart will illustrate its use.

Prescription Example

The first chart in this series, Chart 8, has only one zone, and its purpose is to determine the stage of development of this stand. If the stand is mature and ready to be regenerated, you will be routed to the 8.1 section of the chart. If regeneration treatments have already been applied, but the stand is still in the process of regenerating, you will be routed to the 8.2 section of the chart where the need for weeding, protection against animal depredations, and similar matters are considered. If the stand is in the sapling or larger stage but not yet mature, you will be routed to section 8.3 of the chart where the need for intermediate thinning and similar treatments will be evaluated.

NE Decision Chart 8

Starting at the top of Chart 8, the first decision point we reach contains the question: "Is this stand mature?" If we answer "Yes," we are routed to Chart 8.1, which is the beginning of the section on regeneration prescriptions.

NE Decision Chart 8.1

Remember that decision charts are simply an outline for the decision process. To decide whether to answer the question about stand maturity "Yes" or "No," well need specific criteria on what constitutes stand maturity. Those criteria will be covered later.

At the top of Chart 8.1, the five-sided box completes the connection from the previous chart. As the title indicates, it will consider the possibility that a removal cut is appropriate in this stand. The first zone deals with the adequacy of existing advance regeneration plus expected sprouts to regenerate the stand if harvested. There are three specific decision points or questions in this zone.

The simplest case involves those stands that already have regeneration of sapling size established in adequate amounts to form a new stand. Such a stand might have resulted from a heavy cut 10 to 25 years earlier, perhaps a diameter-limit cut or the second cut of a three-cut shelterwood. All that needs to be done is to remove the remaining overstory of the old stand to release the new stand. This prescription is reached if you answer "Yes" to the first decision question, "Is Sapling Regen Adequate?" Again, the criteria by which you decide whether or not sapling regeneration is adequate will be discussed later.

If sapling regeneration is not adequate, follow the route through the "No" box to the question "Is Advance Regen + Sprouts of Target Species Adequate?"

A "Yes" answer here moves us through the first zone to additional zones that deal with interfering plants and site limitations. A final removal cut, or clearcut, will be recommended only if interfering plants and site are not limiting. If interfering plants are dense, removal cutting is not likely to provide satisfactory regeneration of the desired species, and you are routed to Chart 8.1b where alternative techniques are considered. Interfering plants here include any plants that may interfere with successful regeneration of desired or target species. These may be herbaceous plants like fern and grass, woody plants of no commercial timber value like striped maple, dogwood, and sassafras, or commercial tree species such as yellow-poplar and red maple that outgrow desired oaks species under full sunlight. The criteria for evaluating these interfering plants deal with both those plants present in the understory before cutting and those present in the forest floor as dormant seed banks.

NE Decision Chart 8.1b

If certain site factors alone are limiting to regeneration, the recommendation will be for the removal harvest to be done in a sequence of two cuts that ameliorate the site limitations (wet or extremely rocky surface soil), with the first removal cut to be made now. Retaining 35 to 40 percent overstory density until the advance seedlings and sprouts grow to small sapling size and have established roots in deeper soil layers has been found effective in maintaining a transpiration pump. This avoids saturated conditions on wet sites, and stabilizes surface organic soil on rocky areas, as opposed to single-cut removals which accent the soil limitations.

Going back to the first zone in Chart 8.1, a "No" answer will lead to a repeat of the question about adequacy of advance regen + sprouts, assuming the area is fenced against deer. In areas of high deer population, the numbers of advance seedlings and sprouts needed to assure successful regeneration is very high, since deer browsing destroys many of them. A fence to exclude deer has the effect of reducing the numbers of advance seedlings and sprouts needed, and may permit removal cutting to occur where it would otherwise not be feasible. This section of Chart 8.1 is identical to that discussed above, except that any prescription for a removal cut now includes erection of a deer fence as well.

If advance regen + sprouts of target species are not adequate even if fenced, you are routed to Chart 8.1b where alternative techniques to encourage advance regeneration establishment are considered.

The remaining charts work in much the same way, but deal with other circumstances.

NE Decision Chart 8.1a

Chart 8.1a considers the possibility that a shelterwood sequence will permit advance seedlings to develop. Decision points here include evaluations of interfering plant density, overstory density or shade level, seed source availability, and deer browsing. Note that the prescriptions here include both natural and artificial regeneration, that shelterwood seed cuts may be recommended alone or in combination with either artificial regeneration and deer protection (which could include tree shelters in the case of oaks if ongoing research proves them effective). Note also that a "wait" (wait a few more years for advance seedlings to develop) prescription occurs when the desired level of shading or shelter already exists; this prescription also may be combined with planting and/or protection.

Chart 8.1z is to be used in combination with all of the prescriptions obtained earlier, and includes supplementary treatments to be applied along with the primary prescription. These include the need for grapevine control, the need to treat unmerchantable saplings and poles, and the need or desirability of selecting individual stems from the existing stand to be retained after final harvest for specific purposes. If residuals are to be retained, it is important to identify them during any cuts prior to final harvest, so that they are not removed inadvertently at the earlier cut.

NE Decision Chart 8.1z

Thus, the decision charts provide an outline for the decision process. Major factors such as the adequacy of advance regeneration, expected sprouts, seed source availability, interfering plants, deer browsing, site limitations, and overstory shade conditions are evaluated, and the questions lead you quickly through the decision process.

Some factors of known importance may appear to be missing from the decision charts. For example, we know that acorn insects can be a major factor in oak regeneration. Although not included individually in the decision charts, losses of acorns are included in the criteria used to determine whether or not seed supply is adequate. Other pest problems, and many other factors are or could be included in a similar manner, as part of the decision criteria.

On high-quality sites in the Southern Appalachians, dominant and codominant oak stems will be present in regenerated stands only when adequate regeneration sources exist in the mature stands to be harvested. These regeneration sources may be present in mature stands as advance reproduction or as stump sprout potential. Since oak trees tend to be larger and fewer in number on high-quality sites than on lower quality sites, and since large stumps are less likely to produce viable sprouts than are smaller stumps, advance reproduction is the most important oak regeneration source on high quality sites. The potential for advance oak seedlings to become dominant or codominant stems in the next stand increases with seedling size. Thus, the expected contribution of trees (as stump sprouts) and advance reproduction are both related to size. Methods are currently available to assess oak regeneration potential on high-quality sites in the southern Appalachians. If the oak regeneration potential (and the potential for other species) in an existing stand is sufficient to meet management objectives, the overstory can be removed in one or more harvest cuts. If, however, the oak regeneration potential is not sufficient to meet management objectives, oak regeneration potential must be increased. Since very little can be done to increase the stump sprout contribution in the short run, the focus of activities must be to increase the contribution of advance reproduction. A technique currently recommended reduces stand basal area of the mature stand from below with herbicides, leaving a main canopy with no canopy gaps. This treatment allows existing, small oak seedlings to grow, but does not encourage yellow-poplar establishment and growth. The herbicide treatment prevents tolerant, mid-story stems from sprouting and removes them as a source of competition both before and after overwood removal. About 10 years after treatment, oak seedlings should be large enough to compete, and overwood removal can begin.

• Introduction
• The Oak Regeneratoin Problem
• Assessing Red Oak Regeneration Potential
• Increasing Red Oak Regeneration Potential
• Limits of Application

A predictive model for estimating "dominance probabilities" facilitates quantitative expression of the regeneration potential of red oak reproduction under shelterwood in the Appalachian region (Loftis 1990a). A dominance probability is the probability of an advance reproduction stem being dominant or codominant 20 years after shelterwood removal. Probabilities increase with increasing basal (ground line) diameter and decreasing site index. On site Index 70, probabilities range from 0.01 for reproduction 0.1 inch in basal diameter to 0.46 for reproduction 1.6 to 2 inches in diameter; on site index 90, probabilities range from 0 to 0.34 for the same range of diameters. In application, the expected future numbers of dominant and codominant red oaks per unit area can be calculated by multiplying the observed number of seedlings in each of several basal diameter classes by the dominance probability for each class and summing those products across all diameter classes. (Isebrands and Dickson, 1994)

Northern red oak is a very important tree species in the southern Appalachians. The stands in which it occurs range from those dominated by red oak and other upland oaks to those dominated by deciduous species other than oaks. High-quality wood and relatively rapid diameter growth make red oak one of the most desirable sawtimber species, and its good, but infrequent, acorn crops are an important wildlife food. The problem of regenerating northern red oak on high-quality sites in the southern Appalachians is well-documented and has been the subject of much research over the past several decades. In this paper I will discuss the overall research and the development of a successful regeneration technique.

Thirty years ago, Charles E. McGee and Don Beck began a series of regeneration studies on the Bent Creek Experimental Forest that (1) identified the problem and (2) provided important direction for future research (McGee 1967, Beck 1970, McGee 1975, Beck and Hooper 1986, Della-Bianca and Beck 1983, Loftis 1983). The findings from these studies indicated that (1) oaks regenerated well after clearcutting on lower quality, xeric sites (where northern red oak seldom occurs); (2) oaks, and particularly northern red oak, performed poorly in competition with other species, particularly yellow-poplar and sprouts of shade-tolerant subcanopy species, on high-quality sites after clearcutting; and (3) regeneration under shelterwoods, over a broad range of residual basal areas, was essentially the same as would be expected after clearcutting. Concurrent research, in the Central States by Ivan Sander and Bryan Clark emphasized the importance of pre-existing vegetative structures, advance reproduction and stump-spouts, for oak regeneration (Sander and Clark 1971). In addition, Sander reported that growth of advance reproduction following harvest was positively correlated with size of advance reproduction prior to harvest (Sander 1971, Sander 1972), a relationship we subsequently found important for red oak in the southern Appalachians (Loftis 1990a).

McGee (1967) observed that under mature mixed oak stands, "seed germinate, the seedlings live for a few years, die, and are replaced by new ones. Cutting the overstory interrupts this cycle and stimulates growth." A study installed by Beck (1970) provides a quantitative description of the dynamics of red oak seedling populations in the absence of disturbance. New seedlings become established whenever there is a good acorn crop, but after 10 years fewer than 10 percent of the seedlings have survived. And those survivors have grown very little (Loftis 1983).

It is now clear that regenerating red oak on high-quality sites is a problem in large part because the large advance reproduction necessary for maintaining a red oak component in the next stand is seldom present at the time of heavy regeneration cuts. Further, large red oak advance reproduction will not develop in the absence of disturbance sometime prior to harvest.

Assessing Red Oak Regeneration Potential

In a stand where regeneration is being considered, and where maintaining a component of red oak in the regenerated stand is a management objective, the first step in meeting this objective is an assessment of existing red oak regeneration potential. As Johnson (this Proceedings) has indicated, regeneration potential is the expected contribution of existing advance reproduction and stump sprouts to the next stand.(Table:Dominance Probabilities for Northern Red Oak)

These probability values can be used to estimate the contribution of existing advance reproduction and sprouts produced by stumps if the stand were regenerated immediately by:

N = the expected number of dominant and codominant red oaks per acre at age 20 in the new stand

n_ij = the number of red oak per acre in the i^th size class on the j^th site index class

p_ij = the probability of red oak in the i^th size class on the j^th site index class becoming dominant or codominant at age 20.

For example, consider a mature stand on a site index of 90 with the following distribution of red oak:

Obviously, the stand under examination must be sampled to provide estimates of n_i for advance reproduction (stems 2 in. or less in diameter at groundline) and for stems 2 in. d.b.h. and larger. Site index (j) must also be determined. Given the age of the study on which the probability model is based, the results are tentative. However, the models should allow the silviculturist to recognize, at the very least, those stands where the current red oak regeneration potential is inadequate to meet management goals.

If the current red oak regeneration potential is deemed adequate, the stand can be harvested in a single cut. However, if management objectives indicate the need to retain some overstory trees for a period of time, oak regeneration can still be successful if more than one cut is used to harvest the stand.

If current red oak regeneration potential is deemed inadequate to meet management objectives, silvicultural treatment must be used to increase the regeneration potential. Shelterwood methods have long been recommended for regenerating oaks. However, as stated earlier, our experience with shelterwood cuts was disappointing. In shelterwoods with residual basal areas ranging from 25 sq. ft. to 66 sq. ft., oak seedlings did grow. But oak seedlings were soon overtopped by yellow-poplar seedlings and by sprouts of tolerant subcanopy species. Oaks did not benefit, relative to other species, from higher residual basal areas or longer periods of overstory retention. These results suggest that a shelterwood method to encourage red oak regeneration must (1) provide for the development of large advance red oak reproduction while preventing the establishment and growth of yellow-poplar, and (2) control competition from shade-tolerant, midstory species.

To address this question, we installed plots in well-stocked, mature stands with basal area reductions ranging from 0 to 40 percent of initial basal area. The basal area reduction was accomplished from below, with herbicides and deals in the most straight-forward manner with sprouting of understory species. A complete description of the methodology used in these studies and a presentation of early results can be found in Loftis (1990b).

The results of these studies after 13 years are very promising. We have defined residual stand densities that impede the establishment and growth of yellow-poplar while still providing enough light for the growth of small red oak seedlings that exist at the time of treatment. We did not find in these studies that shelterwoods enhance the establishment of new red oak seedlings.

If advance reproduction exists in the stand, but is not adequate to provide the desired oak component in the next stand (based on the model noted previously), basal area reductions in the ranges of 25-30 percent on SI=90, 30-35 percent on SI=80, and 35-40 percent on SI=70 sites will result in development of larger advance reproduction. This treatment, should leave a stand with no gaps in the main canopy. Rarely would this initial treatment result in the removal of commercial material. The cost of application of this method is likely to be only marginally greater than the cost incurred when the same material is removed by other methods used to create well-stocked young hardwood stands. The initial treatment controls essentially the same vegetation that must be controled regardless of the choice of regeneration method. The difference is primarily one of timing.

Mean basal diameter growth of red oak seedlings resulting from the above treatment can be predicted (Figure), and a future basal diameter distribution can be projected. For example, in a mature stand on a site index of 90 in which basal area is reduced by 27.5 percent, the predicted mean basal diameter growth would be 0.42 in. (Figure). The data suggest that growth about this mean is normally distributed.

By recovering the standard deviation σ from the equation:

a vector of transition probabilities based on the normal distribution can be generated. These transition probabilities are the probabilities that a seedling will grow 0.0 in., 0.1 in., 0.2 in., etc., in 10 years.

For a given population of seedlings, say:

adjusted for survival:

a future basal diameter distribution of red oak advance reproduction can be projected:

Then, by applying the dominance probabilities to this projected distribution we can predict how much oak we would expect if this treatment is applied:

By collecting information about the size distribution of red oak from small seedlings to large trees in an existing mature stand, the forester can explore the consequences of two potential prescriptions. First, the dominance probabilities (Table) can be applied to the existing size distribution to predict the red oak component expected if the overstory removal in one or more cuts is begun immediately, with appropriate treatment of competing vegetation. Second, the existing size distribution can be used in the projection method presented above to predict the expected red oak component in the next stand if overstory removal is preceded by a basal area reduction from below followed by 10 years of advance reproduction development, as described above.

The results are based on studies primarily of northern red oak in the southern Appalachians. The stands in which this study were conducted would be classified as northern red oak-white oak-yellow-poplar, by far the most common mesic association in the southern Appalachains. In this association, yellow-poplar is a very aggressive competitor. The association lacks a truly shade-tolerant canopy species, but does contain a number of shade-tolerant subcanopy species. Further, these studies were conducted in stands that were initially well-stocked, with complete closure of the main canopy and significant density of subordinate crown layers. While similar results have been reported in bottomland hardwoods (Janzen and Hodges 1985) and in more xeric oak systems (Graney 1989, Schlesinger and others 1993), the details of the method and almost certainly the models will be different.

Based on observations and measurements in some of the study plots (Loftis, unpublished data), black oak, scarlet oak, and chestnut oak respond to release similarly to red oak. White oak, however, does not. (Loftis, 1993b)

Predicting oak regeneration requires predicting the dynamics of acorn production, oak seedlings, seedling sprouts, and stump sprouts. The following sections discuss the various factors that influence the natural variation in these four regeneration components.

• Predicting Acorn Production
• Predicting Seedling Cohort Dynamics
• Predicting Seedling Sprout Dynamics
• Predicting Stump Sprout Dynamics

The unpredictable nature of flowering in oaks results in the irregular occurrence of acorn crops and thus new seedlings (Cecich 1991). On the average, most species produce a good acorn crop once every 3 or 4 years (Olson 1974). Numerous biotic and abiotic factors influence acorn viability, germination, initial seedling establishment, and survival. For example, dry weather, droughty soils, and freezing temperatures can reduce acorn viability and germination (Korstian 1927). Acorn crops also are frequently destroyed by unpredictable but frequent infestations of acorn weevils (Christisen and Kearby 1984). Most of the remaining acorns may be consumed by rodents, deer, birds, and other animals (Marquis and others 1976, Sork and others 1983). So, even after a bumper acorn crop, few acorns may be available for seedling production. Among the few remaining viable acorns, many fall into microsites unsuitable for germination and seedling establishment (Johnson 1993).

Significant numbers of new oak seedlings thus occur as unpredictable population waves associated with bumper acorn crops and a spatial distribution resulting from patchy germination suitability and seedling survival. The relatively infrequent occurrence of large seedling populations originating from one acorn crop (cohort) usually coincides with a bumper acorn crop combined with other fortuitous events, such as weather, that favor the preservation of acorn viability through fall and spring germination periods and low populations of acorn consumers. (Johnson 1993)

The acorn producing capacity of a stand, and thus the rate of seedling input into oak forests, changes with time. Although variation in stand structure and age can account for some of the variation in both the temporal and spatial variation in oak seedling establishment, inherent variation in acorn production among trees introduces an essentially random element into predicting seed and seedling inputs into oak forests (Johnson 1993). Some of the variability noted:

1. Large trees usually produce more acorns than smaller trees because, other factors being equal, acorn production increases with crown area. In turn, crown area is correlated with bole diameter (Goodrum and others 1971).
2. However, in some species there is a threshold diameter above and below which acorn production decreases (Downs 1944). Large, senescent oaks are poor acorn producers (Huntley 1983).
3. The production of acorns per unit of crown area is also greater in open-grown trees than in forest-grown trees of the same size (Gysel 1956, Sharp 1958).
4. Some trees are better acorn producers than others even when tree size and environmental factors are the same (Sharp and Sprague 1967). For example, in mature white oaks in Pennsylvania, only 30 percent produced any acorns even in good seed years (Sharp 1958), and an even smaller percent produced a good crop in those years (Sharp and Sprague 1967).

Several studies have shown that survival of northern red oak seedlings originating from a single cohort is influenced by overstory density and other stand characteristics (see survival curves). A cohort of oak seedlings are the survivors of a single acorn crop.  In general, reducing overstory density increases seedling survival and growth (Beck 1970; Crow 1992; Loftis 1988, 1990). A dense layer of lower story trees, shrubs, or ground cover also can reduce seedling survival and growth (Beck 1970, Loftis 1990, Scholz 1955). Other factors that can reduce oak seedling survival include animal browsing, insect defoliation, droughty soils, inadequate light, and frost (Crow 1992, Gottschalk 1988, Hanson and others 1987, Korstian 1927, McGee 1988; Johnson 1993).

Survival Curves for Northern Redoak

During their first 9 years, numbers of northern red oak seedlings from a single cohort in North Carolina declined exponentially beneath the parent stand (survival curve, Part A). Unfortunately, there are few detailed reports of oak seedling survival of similar duration. Short-term studies nevertheless point out the great variation in survival rate among seedlings of the same species growing in various regions representing different stand densities and different light and competition environments. They also establish the range over which we might reasonably expect the survival rates of oak advance reproduction to recur. For example, after 5 years, survival of individual cohorts of northern red oak seedlings ranged from about 0.16 to 0.86, depending on overstory density or understory competition (q.v., survival curves) (Johnson 1993).

Despite the seemingly complex problem of predicting the establishment of oak seedlings, more than half the variation in the density of black oak and white oak advance reproduction in xeric forests in northern Lower Michigan was explained by relatively simple measures of overstory density and structure (Johnson 1992). For both species, 55 percent of the variation in reproduction density was explained by total overstory basal area and the basal area of "large" trees presumed to be the primary seed producers. Large trees were defined as those at least 14 in. d.b.h. for black oak and those at least 12 in. d.b.h. for white oak (see figure, below). The related models also showed that, per unit basal area of large trees, white oak was more efficient at producing seedlings than black oak. Moreover, high densities of black oak reproduction were favored under low density stands, whereas the reverse was true for white oak. Other studies have shown that topographic factors, stand history, and site quality also influence oak reproduction density (Arend and Scholz 1969, Carvell and Tryon 1961, Nowacki and others 1990, Ross and others 1986, Walters 1990; Johnson 1993).

Density of Black and White Oak

Oak reproduction often may be absent or scarce on mesic or hydric sites in the absence of disturbance (Carvell and Tryon 1961, R. L. Johnson 1975, Will-Wolf 1991). Nevertheless, oak reproduction densities in these forests may at times exceed 50,000 seedlings per acre in mesic forests (Tryon and Carvell 1958) and 100,000 per acre in bottomland forests (R. L. Johnson 1975). When oak seedlings do occur, they may represent only one or two acorn crops. Because of low survival rates, most of the seedlings from a single cohort may die before the next good acorn crop occurs. The rapid rate of seedling disappearance results largely from the shade-intolerance of oak reproduction and the low light levels on the forest floor (Hanson and others 1987). In the absence of disturbance, these forests typically possess high overstory basal areas and multiple subcanopy layers (Braun 1967, Loftis 1990). Such vertical stratification occurs in mesic and hydric oak forests throughout the deciduous forest region (Johnson 1993).

Because the seedling stage is usually brief, seedling sprouts are the predominant form of oak reproduction in many, if not most, oak forests. Seedlings can sprout from dormant buds anywhere along the stem between the root collar and the terminal bud cluster. Dieback and resprouting seem to be important processes in the life of oak reproduction. Although recurrent shoot dieback is common to most hardwoods, it is especially prominent and ecologically important in the xerophytic oaks, which are morphologically and physiologically adapted to survival in environments subjected to repeated fire and drought (Abrams 1990, Grimm 1984, Wuenscher and Kozlowski 1971; Johnson 1993).

The natural environment of seedling sprouts of the xerophytic oaks imposes stresses that periodically decrease shoot mass and leaf area through shoot dieback. Surviving seedling sprouts thus develop increasingly greater root:shoot ratios as roots grow incrementally larger and shoots recurrently die back. In turn, high root-shoot ratio and large root mass enable oak reproduction to opportunistically respond to favorable environmental conditions by facilitating two or more long flushes of shoot growth (multiple flushing) during one growing season (Dickson 1991, Johnson 1979).Successional replacement of oaks by oaks thus heavily depends on conditions that favor the long-term accumulation of oak reproduction with high root:shoot ratios and large root mass. Lacking those characteristics, oaks are usually at a competitive disadvantage. This is especially true of the reproduction of the xeromorphic upland species, which grow slowly even under optimal conditions until they develop the requisite root mass and root:shoot ratio. Shoot dieback thus may be an important aspect of the evolutionary development and adaptive strategy of oaks (Johnson 1993).

The accumulation of oak reproduction under a parent stand is one of the most important aspects of the regeneration ecology of oaks. Oak silviculturists call this"advance reproduction" because, in the even-aged management of oaks, it is present in advance of final harvest. Its presence and development largely determine the importance of oaks after natural or human-caused events that destroy or remove the parent stand. Oaks opportunistically capitalize on this accumulation process because it facilitates the capture of growing space when the overstory is destroyed or removed. This capacity largely depends on the characteristics of the competing vegetation and the accumulated population of oak seedling sprouts with large roots. The size distribution and age distribution of this population, in turn, depend on the balance of birth, death, and growth rates of reproduction intrinsic to each type of oak forest (Johnson 1993).

Estimated Sprouting Probabilities

Stump sprouts originate from dormant buds at or near the base of stumps of harvested overstory trees. In the silviculture of central hardwoods, overstory trees are defined as those 2 in. d.b.h. and larger (Roach and Gingrich 1968). However, that definition has not been universally adopted. The biological distinction between a stump sprout and a seedling sprout is nevertheless arbitrary because all oaks, from small seedlings to large standing trees, have some potential to produce basal sprouts when the parent stem is cut. When wind, fire, or other factors destroy an oak stand, sprouts also may develop from the bases of trees that have broken off or from standing trees with dead tops (Johnson, 1993).

For several species of oaks, the percentage of stumps expected to produce sprouts after timber harvesting can be estimated from tree diameter and tree age (Johnson 1977). In general, the frequency of sprouting decreases with increasing tree diameter, age, and site quality (see figure). But other factors, such as season of cutting and shading, also can affect stump sprouting in hardwoods. For some species of oaks, there is evidence that stumps sprout more frequently when trees are cut or killed during the dormant season than during the growing season (Clark and Liming 1953, Kays and others 1988). However, some of the live oaks of the Western United States sprout prolifically regardless of season of cutting (Longhurst 1956). Although sugar maple stumps exposed to full light sprouted more frequently than shaded stumps (Church 1960), similar responses of oak stumps to shading have not been reported. McGee and Bivens (1984) observed that numbers of stems in white oak sprout clumps were about the same for stumps that had been released from directly overtopping trees and stumps that were not released. Regardless of treatment, nearly 100 percent of the stumps of trees between 2 and 8 in. d.b.h. sprouted. Larger trees or those older than 60 years produced few or no sprouts (Johnson 1993). Stump sprouts originating from pole-size and larger parent trees are, in effect, mature root systems connected to juvenile shoots. This root:shoot combination results in rapid height growth. During their first decade, open-grown stump sprouts can produce four or more flushes of shoot growth per year totaling 3 ft. or more even under droughty conditions (Cobb and others 1985, Johnson 1979, Reich and others 1980). The large root mass of stump sprouts and their large carbohydrate storage and absorptive capacity, together with other factors, facilitate multiple flushing in oaks. In contrast, multiple flushes are not produced in mature oaks, shaded seedlings and seedling sprouts, and small seedlings and seedling sprouts under water stress (Borchert 1976, Buech 1976, Cook 1941, Johnston 1941, Kienholz 1941, Longman and Coutts 1974; Johnson 1993).

Frequency of flushing and total shoot elongation in oaks usually decline as stems increase in size and age and as root systems approach their maximum size. The number of flushes in scarlet oak stump sprouts decreased from an average of about two per growing season the first year to one by the fourth growing season (Cobb and others 1985). Thus, by the fifth year, the pattern of shoot growth approached that of the single flush of a mature tree. The progression from multiple to single flushes may be attributable to a declining root:shoot ratio that results in increasingly longer periods for roots and shoot to restore "functional balance" after shoot elongation and leaf expansion (Borchert 1975; Johnson 1993).

The number and spatial distribution of sprouts around the stump also influence sprout growth. The importance of sprout distribution around the stump may be related to the pattern of vascular connections that develop between sprouts and the parent tree root system, with each stem helping to sustain a portion of the root system (Kharitonovich 1937, Kramer and Kozlowski 1979, Roth and Sleeth 1939, Wilson 1968). In the Missouri Ozarks, the number of sprouts per stump was positively correlated with the early height growth of five oak species (Johnson 1977). The same relation also was observed for various oak species in other regions (P. S. Johnson 1975, Ross and others 1986, Schwarz 1907). Collectively, the observed spatial distribution and clump density effects suggest that numerous well-distributed sprouts maintain the parent tree root system and thus an efficient root-shoot feedback system that promotes rapid early height growth (Johnson 1993).

However, the apparent benefits of a balanced distribution of stems and high clump density are short lived. In Wisconsin, rapid growth of the dominant stem in unthinned clumps of 4- to 23-year-old northern red oak stump sprouts was associated with high clump density (P. S. Johnson 1975). Similarly, northern red oak clumps thinned to one stem as early as age 4 subsequently grew faster than stems in unthinned clumps (Johnson and Rogers 1984). This would seem to indicate that competition between stems in the same clump begins very early. However, 12-year-old northern red oak stump sprouts in Appalachian forests did not respond to clump thinning (Lamson 1988). Although northern red oak commonly initiates several dozen sprouts, stem crowding soon induces rapid stem mortality so that by the end of the first decade only four or fewer stems per clump typically remain. But this natural clump thinning process varies greatly among and within species (cf. P. S. Johnson 1975, Roth and Hepting 1969, Schwarz 1907, Johnson 1993).

The diameter of the parent tree and the correlated size of the root system also affect the growth of oak stump sprouts. For five species of oaks in the Missouri Ozarks, the correlation between stump diameter and height growth of the dominant stem within a sprout clump was consistently negative for 5-year-old sprouts of all species (Johnson 1977). However, the opposite was true of oak sprouts in Virginia (Ross and others 1986). Such discrepancies might be explained by the range of stump diameters observed in any given study. For example, data from a study of black oak and white oak sprouting that included trees ranging from less than 1 in. to more than 12 in. in basal diameter showed that the most rapid height growth occurred in clumps originating from 6-in. stumps (Johnson 1979). Sprouts from stumps larger or smaller than that grew less. The height growth of oak reproduction thus changes continuously, but not unidirectionally, from small advance reproduction to large-diameter overstory trees (see figure). (Johnson, 1993)

Height Growth and Parent Stem Diameter

Because of the large variation in number of stems per clump within a stump diameter class, we might expect roots and shoots of many young sprout clumps to be physiologically imbalanced. Accordingly, among the imbalanced clumps, rapid changes in clump structure would be expected as the clumps move toward functional balance. This view is supported by the initially large but rapidly decreasing variation in shoot growth among dominant shoots within black oak and white oak sprout clumps during the first 4 years (Johnson 1979). The large amount of unexplained variation in the relation between shoot growth and stump diameter also reflects rapidly changing root-shoot relations. Other factors that may be significant sources of variation in the shoot growth of oak sprouts include site quality, genetic variation, competition, parent tree age, and season of cutting. The significance of these factors may vary among oak species and regions (Johnson, 1993).