Growth Pattern of Picea Rubens Prior to Canopy Recruitment

pr7 7

Growth pattern of Picea rubens prior to canopy recruitment

Xinyuan Wu l , J. Frank McCormick2 Richard T. Busing3 1 Department of Rangeland Ecology and Management Texas AM University, College Station, TX 77843, USA; 2Department of Ecology and Evolutionary Biology, The University of Tennessee Knoxville, TN 37996, USA; 3Forestry Sciences Laboratory 3200 SW Jefferson Way Corvallis, OR 97331, USA

Received 22 July 1997; accepted in revised form 6 October 1998

Key words: Canopy disturbance regime, Great Smoky Mountains, Radial growth, Red spruce, Suppression and release

Abstract

A majority (72%) of Picea rubens Sarg. (red spruce) trees in an old-growth spruce-fir forest in the Great Smoky Mountains underwent episodes of radial growth suppression and release before they reached the forest canopy. Prior to canopy recruitment, trees experienced an average of 1.43 and a maximum of 7 suppression periods with an average ring width of 0.257 mm. Duration of suppression periods ranged from 4 to 79 years with an average of 19.05 years, which was significantly shorter than the average duration of release periods (29.00 years). Mean ring width in a suppression period was negatively correlated with duration of the suppression period. The opposite was true for release periods. The severity of suppression had no significant effect on mean ring width in subsequent release periods. Greater suppression was observed in the recent growth pattern of current non-canopy trees than in the historical growth pattern reconstructed from current canopy trees. Growth releases prior to canopy recruitment, triggered by small gaps or neighboring gaps, displayed a continuous pattern over the last two centuries and had consistent frequency in recent decades. By contrast, canopy recruitment resulting from large-gap disturbances was discontinuous over the last two centuries, and less frequent after 1850.

Introduction (Adelges piceae) an exotic insect (Eagar 1984). This and other recent disturbances have stimulated research on spruce-fir forest dynamics in the southern Ap- Study of understory growth patterns in shade-tolerant palachians (Busing Clebsch 1987; Busing et al. trees has lead to a better understanding of their ecol- 1988; Busing Wu 1990; Busing Pauley 1994; ogy and of canopy disturbance regimes in deciduous Nicholas et al. 1992; Peart et al. 1992; Stephenson forests (Canham 1985, 1989; Frelich Lorimer 1991; Adams 1984; White et al. 1985). Several of these Lorimer Frelich 1989). Multiple episodes of sup- studies have considered the population dynamics of pression and release prior to canopy recruitment have Picea rubens (Sarg.). Patterns of radial growth of been inferred from radial increments of shade-tolerant Picea rubens in the canopy are also well documented species. The frequency of such episodes is often (e.g., LeBlanc et al. 1992; Reams et al. 1993; Van high suggesting frequent disturbances to the canopy Deusen 1988). However, studies describing the growth (Canham 1990). pattern of this species prior to canopy recruitment are Southern Appalachian spruce-fir forests are the few (e.g., Mount Grove 1952, reported in McLin- southernmost high-elevation coniferous forests in tock Bickford 1957). Information on growth and eastern North America (Oosting Billings 1951; canopy recruitment is basic to an understanding of the White 1984; White Cogbill 1992; Whittaker 1956). population dynamics of Picea rubens, the community In recent decades, the southern Appalachian en- structure and dynamics of spruce-fir forest stands, and demic, (Pursh.) Poir. (Fraser fir), has Abies fraseri the effects of disturbances upon them. been severely affected by the balsam woolly adelges 246

The pattern of growth by which shade-tolerant stands with a Picea overstory. Stand I was on a south- trees reach the canopy can be described in terms of facing slope. The point on the trail nearest (about periods of suppression and release they undergo (Can- 75 m) to the center of stand I was approximately 260 m ham 1985). There are two basic modes of canopy east of the observation tower along the trail. Stand II recruitment. Young trees may start in large canopy was on a north-facing slope. The point on the trail openings or successively expanding smaller openings nearest (about 120 m) to the center of stand H was (Foster Reiners 1986; Runkle 1985) and grow di- approximately 425 m east of the tower along the trail. rectly into the canopy without any growth suppression. A 10 x 50 m transect perpendicular to the ridge was Alternatively, individuals may persist under a closed conducted in each stand to characterize stand com- canopy with suppressed growth, and respond with position and structure. Live stems of all tree species released growth to canopy openings created by the with a dbh (diameter at breast height, 1.37 m) of death of canopy trees or the falling of large branches. 3 cm or greater were sampled. Picea rubens, Abies They may reach the canopy in a single period of re- fraseri, Betula alleghaniensis Britton (yellow birch). lease, or they may undergo several periods of release Acer spicatum Lam. (mountain maple) were present and suppression before reaching the canopy. The rel- in both stands. Prunus sp. (cherry) was also present ative importance of these two modes for Picea in in stand I. Picea dominated both stand I (680 stems two old-growth stands of the southern Appalachians ha-1 , 85.89 m 2 ha— I basal area) and stand II (660 is addressed in this study. stems ha-1 , 57.23 m2 ha— I basal area). It represented Growth patterns may be used to infer distur- 89.3% and 93.4% of the basal area of all tree species bance regimes over several decades or centuries (e.g., in stand I and stand II, respectively. Abies had the Lorimer Frelich 1989; Frelich Lorimer 1991). lowest density (40 stems/ha) and basal area (0.05- Radial increment series can be used not only to in- 0.07 m2 ha-1 ) among tree species in both stands. The fer the frequency and severity of local canopy dis- overwhelming dominance of Picea and low impor- turbances, but also to detect changes in disturbance tance of Abies were typical of stands at this elevation regimes. The constancy of regimes can be explored (Busing et al. 1993). using series from different periods. For comparison to recent growth patterns of trees that have not reached Sampling and measurements the canopy, the historical growth pattern of Picea can be reconstructed from tree ring series of canopy trees. Field sampling was conducted in 1986. Increment Differences between recent and historical growth pat- cores were extracted at breast height from 47 canopy terns may reflect environmental changes in recent Picea trees (25 in stand I and 22 in stand II) and 50 centuries. non-canopy Picea trees (29 in stand I and 21 in stand The primary objectives of this study are: ( 1 ) to re- II). In this study, canopy trees included superdomi- construct the growth pattern of Picea from tree ring nants, dominants and co-dominants visually classified series of canopy trees; (2) to compare the recent based on their height and position relative to trees growth pattern of non-canopy trees with the histori- in the main canopy. All other trees (subcanopy and cal growth pattern of canopy trees; and (3) to make understory) were considered non-canopy trees. One inferences on the canopy disturbance regime. Our core was taken in the direction parallel to terrain con- findings contribute to the understanding of Picea ecol- tour. Trees <5 cm or >70 cm in dbh were not sampled ogy, spruce-fir forest dynamics, and forest disturbance due to the concern of the effect of coring on small history in the southern Appalachians. trees, and the limitation of the equipment available at the time (the bit length of the borers) for coring very large trees. Trees of abnormal growth form (bent, Methods twisted or hollow) were excluded from sampling. A tree ring measuring system (Robinson Evens Study site 1980) was used to determine ring width (annual increments of radial growth) by ring number at breast Two stands were selected along Noland Divide within height of all trees sampled. Picea cores had dis- the spruce-fir forest zone in the Great Smoky Moun- tinct boundaries between rings, even during strongest tains National Park (35°35 N, 83°28 W), between periods of suppression. Tree ring series were not cross- 1620 m and 1 700 m elevation. Both were old-growth dated and standardized. This study focused on the 247 section of a tree ring series when the tree was in the pared to the growth pattern based on the threshold of understory, and on the growth pattern in this sec- 0.40 mm yr-1 . None of the growth pattern parameters tion related to micro-environmental stand dynamics based on each of the four additional thresholds was rather than macro-environmental factors. The lack of significantly different from that based on the threshold cross-dating, an important step for establishing precise of 0.40 mm yr-1 , with two exceptions. As one would chronologies, might have introduced some error but expect, averages of ring width in suppression peri- should not have significantly affected characterization ods based on the thresholds 0.35 and 0.45 mm yr-1 of suppression and release periods (Canham 1985). were significantly different from the corresponding average based on the threshold of 0.40 mm yr -1 . With Growth pattern an increased threshold value, all the growth pattern parameters increased slightly except for the duration The term growth pattern refers to the alternating of release periods which decreased. The ranges of periods of suppression an release prior to canopy re- most of the growth pattern parameters changed little cruitment that occurs at the initial year of the release with the different threshold values. These results indi- period during which an individual advances to the cate that growth pattern is relatively insensitive to the canopy (Canham 1985, 1990). It should be noted that threshold value within the range tested. canopy recruitment events are inferred from tree ring patterns rather than actual observation. The growth pattern was analyzed for tree ring series of canopy Results and discussion trees with the following procedures. For the purpose of defining suppression and re- Growth patterns prior to canopy recruitment lease periods in a tree ring series, a growth threshold value for release was required. The mean ring width Twenty-eight percent of the Picea trees advanced di- of the last three years for the non-canopy trees beneath rectly into the canopy from breast height without any closed canopy averaged 0.25 mm yr-1 and that for the growth suppression. Most trees (72%) underwent one non-canopy trees in canopy openings of various sizes or more suppression periods from breast height up be- averaged 0.47 mm yr — I . Based on this information, a fore reaching the canopy (Table 1). Trees experienced threshold of 0.40 mm/yr was used to indicate release. an average of 1.43 and a maximum of 7 suppres- A suppression period was defined as an interval sion periods with an average ring width of 0.257 mm. that (1) included four consecutive years of growth These results supported the speculation of White et al. <0.40 mm or a five-year series with four years of (1985) that multiple release events are often required growth <0.40 mm, and (2) did not include three con- for Picea to reach the canopy. The proportion of secutive years of growth > 0.40 mm or a five-year trees that underwent two or more suppression peri- series with four years of growth > 0.40 mm. This de- ods (40%) was considerably lower than that reported finition filtered out short-term fluctuations in growth for Acer saccharum Marsh. (sugar maple) and Fagus likely due to factors other than canopy opening and grandifolia Ehrh. (American beech) (>60%) in east- closure. ern deciduous forests (Canham 1990). A relatively Parameters describing the growth pattern included large proportion of Picea (30%) underwent just one the number of suppression/release periods per tree, suppression period before reaching the canopy (Acer duration of suppression/release periods, average ring <10%, Fagus <20%; Canham 1990). The propor- width in suppression/release periods, years (number of tion of Picea trees that underwent suppression may be rings) and cumulative ring width in suppression peri- underestimated as suppression before attaining breast ods, and age (number of rings) and size (cumulative height is not considered. ring width) at canopy recruitment. These suppression Three tree ring series are shown in Figure 1 as ex- and release periods occurred from breast height on, amples of different growth patterns. Tree #159 reached until canopy recruitment. canopy in a single period of release (Figure la). After a In order to evaluate the sensitivity of the growth release period, tree #182 had a long suppression period pattern to the threshold value, four additional thresh- with very low annual growth before canopy recruit- olds, 0.35, 0.38, 0.42 and 0.45 mm yr -1 represent- ment (Figure I b). The oldest tree (370 years) sampled ing 5% and 12.5% of increases and decreases, were in this study had seven release periods and seven sup- used to generate growth patterns which were corn- pression periods (tree #249; Figure lc). This tree was 248

Table 1. Growth pattern of Picea rubens prior to recruitment, reconstructed from tree ring series of 47 canopy trees.

Parameter mean range std

Number of trees smapled with suppression period(s) 34 Number of suppression periods per tree 47 1.43 0-7 1.38 Duration of suppression periods (yr) 67 19.05 4-79 15.01 Average ring width in suppression periods (mm) 67 0.257 0.139-0.375 0.070 Years (number of rings) in suppression periods 47 27.15 0-150 30.61 Cumulative ring width in suppression periods (mm) 47 6.15 0-37.83 7.04 Number of release periods per tree 47 1.34 0-7 1.36 Duration of release periods (yr) 63 29.00 3-106 26.04 Average ring width in release period (mm) 63 0.735 0.431-1.629 0.256 Age (number of rings) at recruitment (yr) 47 67.02 1-228 8.73 Size (cumulative ring width) at recruitment (mm) 47 39.94 0-14458 5.84

5 - a) 1.8 - a ca 4 - Tree #159 (dbh=59.1cm) a) 1.6 - 0 3 T2 o 2 - .- 1.4 is O O 1 E 1.2 0 E 0 00 000 0 1600 1650 1700 1750 1800 1850 1900 1950 2000 1 0 0 0 0 0 0 0 0 0.8 O 0 0 00 b 0 o o 00 0 Tree #182 (dbh=54.5cm) 000 0 0 OC 0.6 0 rn 6600$- ° 0.4 a 0.2 $ • •

T-7 0 1600 1650 1700 1750 1800 1850 1900 1950 2000 0 20 40 60 80 100 120 Duration of suppression/release period, yr 5 4 Tree #249 (dbh=49.3cm) Suppression period o Release period

3 Figure 2. Mean ring width vs the duration of suppression or release 2 periods of Picea rubens based on tree ring series of 47 canopy trees. 1

0 1600 1650 1700 1750 1800 1850 1900 1950 2000 facing stand II, there were no significant differences Year, AD in the patterns of suppression and release. All results Figure 1. Picea rubens tree ring series. Horizontal lines at 0.40 mm that follow were based on an aggregation of data from indicate release periods. Arrows indicate the initiation of canopy stands I and II. recruitment. Mean ring width in suppression periods was negatively correlated with duration of those periods (r = -0.566), but during release periods the correlation more than 230 years old before canopy recruitment. was positive (r = 0.578) (Figure 2). Suppression There was considerable oscillation in radial growth was consistently severe in long suppression periods after canopy recruitment in tree ring series from the (>30 yr). In shorter suppression periods, however, two stands, corroborating the results of Reams et al. the growth condition was typically less severe and (1993) from nearby higher elevation sites. inure variable than in longer ones (Figure 2). Appar- Although canopy trees on the south-facing stand I ently, short release periods, likely resulting from small were slightly larger and older than those on the north- canopy openings, provided moderate release condi-

249

0.10 tions. In longer release periods, presumably associated a with single large canopy openings or openings that ex- 0.05 pand over time (Foster Reiners 1986; Runkle 1985), 0.00 trees often experienced faster growth. 0.05 It has been documented for other species that long- 0.10 E suppressed trees have greater growth upon release E 0.15 (Marshall 1927; Canham 1988). Mean ring width of =- -0.20 28 trees in the first three years of a release pe- Picea 0.25 I I riod weakly correlates positively with the duration Ern -4 -3 -2 -1 0 1 2 3 4 (r = 0.099) and negatively with mean ring width Time since suppression, yr

(r = -0.214) of the preceding suppression period. a) rn 0.25 The large variance in these relationships suggests that b for Picea, the duration of the preceding suppression .c 0.20 period may not influence the intensity of the release as 0 0.15 a) much as the conditions at release such as gap size and 0.10 orientation. lY 0.05 The abruptness of transitions between periods of 0.00 contrasting growth rates may be used to infer canopy -0.05 dynamics and understory response. For transitions 0.10 where suppression and release periods were both at -4 -3 -2 -1 0 1 2 3 4 least 6 years in length, the change in growth rate (ring Time since release, yr width) from one year to the next was calculated for Figure 3. Rate of change in ring width in the 4 years immediately the 4 years immediately before and after the tran- preceding and following the transition between suppression and re- sition to determine the abruptness of the transitions lease periods of Picea rubens. Filled boxes indicate means that are (Canham 1985). The beginning of release period was significantly (p < 0.05) different from 0. characterized by an abrupt increase in growth rate by 0.21 mm yr I (from 0.32 in the year prior to re- gradual closure of canopy openings. Presumably, lat- lease to 0.53 mm yr- I in the year of release) and the eral canopy growth from adjacent trees is responsible, beginning of suppression period was marked by an in part, for the gradual reduction in growth. abrupt reduction in growth rate of similar magnitude (0.19 mm yr- 1 , from 0.52 in the year prior to sup- Historical vs. recent growth patterns pression to 0.33 mm yr-1 in the year of suppression) (Figure 3). These results indicated distinct transitions The growth pattern prior to canopy recruitment ob- between suppression and release recognized by sup- tained from canopy trees is historical. Contrasts be- pression and release periods as defined in this study. tween the historical growth patterns and a recent The growth transition in Picea was not as abrupt as growth pattern, constructed from non-canopy trees, in Acer and Fagus (Canham 1985, 1990). For Fa- are evident. Growth suppression is greater in the recent gus, the release increment was 0.37 mm yr I (0.34 to growth pattern than in the historical pattern (Table 2). 0.71 mm) with no significant increase in growth rate The recent pattern has significantly more suppression in the 3 years before and after (Canham 1990). The periods per tree, a significantly higher percentage of sharp increase in growth at the beginning of release rings and cumulative growth in suppression periods, for Picea was preceded by a small but statistically and significantly longer suppression periods (Table 2). significant increase in growth, indicating the possibil- Whereas the growth pattern prior to canopy re- ity of a delay between gap creation and release. The cruitment obtained from canopy trees is complete, the sharp increase was followed by more than one year of pattern from non-canopy trees is incomplete. There gradual increase leading to a maximum growth rate of may be additional suppression and release periods be- 0.77 mm yr- I (Figure 3). Although not apparent for fore these understory trees reach the canopy. A certain Acer and Fagus in northern hardwood forests (Can- percentage of understory trees will die before they ham 1985, 1990), a significant reduction of growth in reach the canopy, therefore the understory growth pat- the year preceding the transition from release to sup- tern is not expected to be the same as the historical pression was evident for Picea (Figure 3), suggesting one. In order to compare the recent with the historical 250

Table 2. Comparison of historical growth pattern and recent growth pattern of Picea rubens (mean ± sd (n))

Parameter Historical Recent growth pattern growth pattern

Number of suppression periods per tree 1.43 ± 1.38 (47) 2.69 ± 1.54 Percent of rings in suppression periods 30.48 ± 25.88 (47) 48.84 ± 22.61 Percent of cumulative ring width in suppression periods 15.46 ± 18.91 (47) 25.16 ± 19.94 (51) Duration of suppression period (yr) 19.05 ± 15.01 (67) 25.46 ± 27.84 (137) Average ring width in suppression periods (mm) 0.257 ± 0.070 (67) 0.249 I 0.067 (137) Duration of release period (yr) 29.00 ± 26.04 (63) 23.91 ± 22.36 (147) Average ring width in release period (mm) 0.735 ± 0.256 (63) 0.715 ± 0.249 (147)

Indicate significant (p < 0.05) difference between the two means using two-tailed t-test. growth pattern, the non-canopy tree data set was ad- respectively. The results of Monte Carlo test compar- justed with sub-sampling. Based on a size-structured isons for cumulative growth in suppression periods life table of a Picea population in a nearby site, the and duration of suppression periods were not as signif- estimated mortality during the period that trees grow icant as, but showed trends consistent with the results from the average dbh of the non-canopy trees to the of tests with all non-canopy trees (Table 3). To assess average dbh of the canopy trees is approximately 45% the sensitivity of the comparison to the 55% survival (Busing Wu 1990). A recent growth pattern derived rate, Monte Carlo tests with 50% and 60% survival from a subset (about 55%) of the non-canopy trees rates were also conducted and the results were very might be more appropriate for comparison. In order to similar to those with 55% survival (Table 3). To test select the subset of non-canopy trees for analysis, one the extreme that suppression history is all important, could use the extreme assumption that the 45% lost be- recent growth patterns based on subsets (50, 55 and fore canopy recruitment were those experiencing the 60%) of non-canopy trees with the least suppression most suppression. This is unlikely, however, as Picea (measured by percent cumulative growth that was in is very shade tolerant, subject to understory mortality suppression periods) were compared to the historical when toppled by falling limbs and boles (White et al. growth pattern. The recent (understory-based) growth 1985), and is often recruited in large gaps that may pattern still had significantly more suppression peri- be independent of suppression history. A null alter- ods per tree. Differences in other parameters became native is that understory mortality is independent of statistically insignificant, however. suppression in Picea. Temporal distribution of canopy and non-canopy In order to compare the degree of suppression trees with different numbers of suppression periods in the historical growth pattern to that of the recent appears in agreement with that growth suppression is growth pattern based on non-canopy data adjusted greater in the recent growth pattern than in the histor- with random understory mortality, a Monte Carlo test ical pattern. Eighty-eight percent of the canopy trees, (Manly 1997) was conducted in which 1,000 sub- the same as the percentage for non-canopy trees, had sets of non-canopy trees were formed, each consisting release periods at breast height. By contrast, 28% of of a random selection of non-canopy trees. A re- canopy Picea trees reached the canopy in a single pe- cent growth pattern was derived from each subset riod of release from breast height (Table 1), whereas and compared to the historical growth pattern. The only 3.7% of non-canopy trees had had no suppression percentage of the tests that were significant for each period and had a chance to reach canopy in a single growth pattern parameter, as well as the trend for all period of release (Figure 4). The range of the begin- the significant tests, were noted. Results of the Monte ning dates of canopy trees (1617-1868) is 56% longer Carlo test based on 55% survival subsets (Table 3) are than that of non-canopy trees (1761-1929) (Figure 4). similar to those of the test using all non-canopy trees Only if we assume that most of the non-canopy trees (Table 2). The suppression periods per tree and the would have no additional suppression periods, and that percentage of rings in suppression periods in the recent new non-canopy trees in the next 90 years would have growth pattern were significantly greater than those few or no suppression periods before they reach the in the historical pattern 100% and 98.8% of the time, canopy, would suppression in the recent growth pat- 251

Table 3. Comparison of historical growth pattern (H) vs recent growth pattern (R) of Picea rubens based on randomly drawn sub-populations of non-canopy trees with 60%, 55% and 50% survival rates. For each survival rate, 1000 sub-populations were randomly drawn and each of their growth patterns compared with the historical one.

Parameter used in comparisons Percent of t-tests significant (trend of all significant tests) 60% survival 55% survival 50% survival

Number of suppression periods per tree 100.0% (R>H) 100.0% (R>H) 100.0% (R>H) Percent of rings in suppression periods 99.7% (R> H) 98.8% (R>H) 98.3% (R>H) Percent of cumulative ring width in suppression periods 64.9% (R>H) 58.6% (R>H) 54.4% (R>H) Duration of suppression period (yr) 24.6% (R>H) 25.2% (R>H) 24.2% (R>H) Ring width in suppression periods (mm) 0.1% (H>R) 0.5% (H>R) 0.6% (H>R) Duration of release period (yr) 5.9% (H>R) 7.2% (H>R) 7.7% (H>R) Ring width in release period (mm) 0.1% (H> R) 0% (—) 0.1% (H>R)

8 disturbance. They estimated that the return interval of

7- 0 small gaps was about 100 years, the recovery time for small gaps was about 50 years, and the average canopy residence time was about 112 in a spruce-fir forest of the Great Smoky Mountains. In our study, release periods prior to canopy recruitment tended to be less than 60 years in duration with an average of 29 years 0 (Table 1). This average is shorter than the estimated 0 CDO 0 G- 01- 00 ++ + + .++ 50 years for gap closure. Also, the average duration of suppression periods (19 years) is considerably shorter 0 010 CD40 CCLDOCD ++ -1-141-4++ than average canopy residence time. These differences 0- 00 0 0 CCD aD 0 suggest that canopy gaps need not be directly overhead

1 for release of understory Picea. Increased light from 1600 1650 1700 1750 1800 1850 1900 1950 smaller openings created by limb fall, or from nearby Approximate starting year (at breast height), AD gaps, must enhance growth. Similar conclusions were reached in studies of Acer and Fagus in eastern de- 0 Canopy + Non-canopy ciduous forest (Canham 1985, 1990; Poage Peart Figure 4. Approximate starting date of Picea rubens tree ring series 1993). vs number of suppression periods per tree. The proportions of canopy trees that started their canopy recruitment in different decades can be used tern, when it is completed, be similar to that of the to construct a chronology of canopy recruitment (Can- historical growth pattern. Even so, the temporal distri- ham 1990). This chronology is an estimate of distur- bution of tree ring series would still show considerably bance events creating large or expanding gaps. Be- greater suppression over the period 1825-1925 than cause it does not include non-canopy trees that have during the previous 100-150 years (Figure 4). recently started growth into the canopy it may under- estimate the canopy recruitment in the recent decades. Disturbance regime Lorimer Frelich (1989) included non-canopy trees advancing into the canopy in the construction of a In forests with dense canopies, growth release periods disturbance chronology; such trees were defined as in shade tolerant trees result from canopy disturbances `canopy trees in that study. Using a similar approach, (Canham 1989; Lorimer Frelich 1989). Disturbance a 20-decade chronology (1770-1970) of canopy re- events creating large or expanding gaps are particu- cruitment was constructed for Picea (Figure 5a). It larly conducive to canopy recruitment. In a previous included all canopy trees and those non-canopy trees study of natural disturbance in southern Appalachian that have started advancing into the canopy (currently spruce-fir forests, White et al. (1985) concluded that small gaps (<200 m 2 ) were the predominant form of

▪

252

20% be quite variable over time, whereas small distur- a. bances triggering episodes of growth release tend to c 15% 0 be consistent and sustain the non-canopy population. 10% 0 5% Acknowledgements

0% I,Ii, , 1,1,1, , . ,I, , 1, 0 0 0 0 0 0 0 0 We thank John Rennie, Clifford Amundsen, Peter 1"-- CD ,- CO ,C) I"- Cil r-- I"- CO 03 CO CO CO CD White and Christopher Eagar for their helpful sugges- tions during the course of this study, and Chengxia 15% You and Yao Yin for their assistance in the field. Charles Canham, Randy Bruins, and three anonymous 12% reviewers provided comments that lead to consider- 0 9% able improvement of the manuscript. This research 00 0 6% was supported in part by the Graduate Program in a Ecology of the University of Tennessee, the Great 3% Smoky Mountains Conservation Association through 0% a Carlos C. Campbell Memorial Research Fellow- ship, and the Graduate School of the University of Tennessee through a Hilton A. Smith Graduate Fel- lowship. ■ Canopy trees 0 Non-canopy trees

Figure 5. Twenty-decade chronology (1770-1970) of canopy recruitment and pre-canopy recruitment release period showing (a) the proportion of canopy recruitment occurred in each decade, and References (b) the proportion of release periods (prior to canopy recruitment) started in each decade. Busing, R. T. Clebsch, E. E. C. 1987. Application of a spruce-fir forest canopy gap model. For. Ecol. Manage. 20: 151-169. Busing, R. T. Wu, X. 1990. Size-specific mortality, growth, and in a release period initiated before 1970, and in canopy structure of a Great Smoky Mountains red spruce population. Can. J. For. Res. 20: 206-210. openings). Busing, R. T., Clebsch, E. E. C., Eagar, C. Pauley, E. F. 1988. More than half of the canopy recruitment occurred Two decades of change in a Great Smoky Mountains spruce-fir in the first eight decades (1770-1850). A pulse of high forest. Bull. Torrey Bot. Club 115: 25-31. recruitment was evident circa 1790. From 1850 to Busing, R. T., White, P. S. MacKenzie, M. D. 1993. Gradient analysis of old spruce-fir forests of the Great Smoky Mountains 1970 there were lapses in canopy recruitment lasting circa 1935. Can. J. Bot. 71: 951-958. one or two decades. These discontinuities suggest that Busing, R. T. Pauley, E. F. 1994. Mortality trends in a south- the frequency of disturbance events creating large or ern Appalachian red spruce population. Forest Ecology and expanding gaps decreased after 1850. Management 64: 41-45. Canham, C. D. 1985. Suppression and release during canopy Frequent small gap disturbances were not rep- recruitment in Acer saccharutn. Bull. Torrey Bot. Club 112: resented in the chronology of canopy recruitment 134-145. (Lorimer Frelich 1989). Nonetheless, they were Canham, C. D. 1988. Growth and canopy architecture of shade- important to the survival and growth of non-canopy tolerant trees: response to canopy gaps. Ecology 69: 786-795. Canham, C. D. 1989. Different responses to gaps among shade- Picea. A chronology of small-gap disturbances can tolerant tree species. Ecology 70: 548-550. be constructed using the approximate starting dates of Canham, C. D. 1990. Suppression and release during canopy re- release periods of all canopy and non-canopy trees be- cruitment in Fugus grandifolia. Bull. Torrey Bot. Club 117: fore canopy recruitment (Figure 5b). This chronology 1-7. Eagar, C. 1984. Review of the biology and ecology of the bal- of growth releases resulting from small disturbances sam woolly aphid in the southern Appalachian spruce-fir forests. had a pattern similar to that of canopy recruitment pp. 36-50. In: White, P. S. (ed.), The southern Appalachian from 1770 to 1850. Unlike the pattern of canopy re- spruce-fir ecosystem: its biology and threats. USDI, National cruitment, the small-disturbance chronology does not Park Service, Research/Resources Management Report SER-71. Foster, J. R. Reiners, W. A. 1986. Size distribution and expan- show a decrease in frequency after 1850. Apparently, sion of canopy gaps in a northern Appalachian spruce-fir forest. large disturbances leading to canopy recruitment can Vegetatio 68: 109-114. 253

Frelich, L. E. Lorimer, C. G. 1991. Natural disturbance regimes Reams, G. A., Nicholas, N. S. Zedaker, S. M. 1993. Two hundred in hemlock-hardwood forests of the upper Great Lakes region. year variation of southern red spruce radial growth as estimated Ecol. Monogr. 61: 145-164. by spectral analysis. Can. J. For. Res. 23: 291-301. LeBlanc, D. C., Nicholas, N. S. Zedaker, S. M. 1992. Prevalence Robinson, W. J. Evens, R. 1980. A microcomputer-based tree- of individual-tree growth decline in red spruce populations of the ring measuring system. Tree-ring Bull. 40: 59-64. southern Appalachian Mountains. Can. J. For. Res. 22: 905-914. Runkle, J. R. 1985. Disturbance regimes in temperate forests. Pp. Lorimer, C. G. Frelich, L. E. 1989. A methodology for estimating 17-33. In: Pickett, S. T. A. White, P. S. (eds), The ecology of canopy disturbance frequency and intensity in dense temperate natural disturbance and patch dynamics. Academic Press, New forests. Can. J. For. Res. 19: 651-663. York. Manly, B. F. 1997. Randomization, Bootstrap and Monte Carlo Stephenson, S. L. Adams, H. S. 1984. The spruce-fir forest on the Methods in Biology. Chapman Hall, London. summit of Mt. Rogers in southwest Virginia. Bull. Torrey Bot. Marshall, R. 1927. The growth of hemlock before and after release Club 111: 69-75. from suppression. Harvard For. Bull. 1 1 : 1-43. Van Deusen, P. C. 1988. Analyses of Great Smoky Mountain red McLintock, T. F. and C. A. Bickford. 1957. A proposed site index spruce tree ring data. USDA Forest Service, General Technical for red spruce in the Northeast. USDA For. Serv. Northeast. For. Report SO-69, New Orleans, LA. Exp. Stn. Pap. 93. White, P. S. 1984. Spruce-fir ecosystem, an introduction. Pp. 1- Mount, P. and W. Grove. 1952. Site index for spruce-fir stands based 21. In: White, P. S. (ed.), The southern Appalachian spruce-fir on age at breast height and total height. Univ. MA For. Dept. ecosystem: its biology and threats. USDI, National Park Service, Tech. Note 13. Research/Resources Management Report SER-7l. Nicholas, N. S., Zedaker, S. M. Eagar, C. 1992. A comparison White, P. S. Cogbill, C. V. 1992. Spruce-fir forests of Eastern of overstory community structure in three southern Appalachian North America. Pp. 3-39. In: Eagar, C. Adams, M. B. (eds), spruce-fir forest. Bull. Torrey Bot. Club 119: 316-332. Ecology and decline of red spruce in the eastern United States. Oosting, H. J. Billings, W. D. 1951. A comparison of vir- Springer-Verlag, New York. gin spruce-fir forest in the northern and southern Appalachian White, P. S., Mackenzie, M. Busing, R. T. 1985. Natural system. Ecology 32: 84-103. disturbance and gap phase dynamics in southern Appalachian Peart, D. R., Nicholas, N. S., Zedaker, S. M., Miller-Weeks, M. spruce-fir forests. Can. J. For. Res. 15: 233-240. M. Siccama, T. G. 1992. Condition and recent trends in high- Whittaker, R. H. 1956. Vegetation of the Great Smoky Mountains. elevation red spruce populations. Pp. 125-191. In: Eagar, C. Ecol. Monogr. 26: 1-80. Adams, M. B. (eds), Ecology and decline of red spruce in the eastern United States. Springer-Verlag, New York. Poage, N. J. and D. R. Peart. 1993. The radial growth response of American beech (Fagus grandifolia) to small canopy gaps in a northern hardwood forest. Bull. Torrey Bot. Club 120: 45-48.