You are here

Harvard Forest Data Archive


History and Dynamics of American Beech in Coastal New England 1620-2006

Related Publications



  • Lead: Posy Busby, Glenn Motzkin, David Foster, Charles Canham
  • Investigators:
  • Contact: Information Manager
  • Start date: 1620
  • End date: 2006
  • Status: completed
  • Location: Naushon Island (MA)
  • Latitude: +41.508
  • Longitude: -70.708
  • Elevation: 10 to 50 meter
  • Taxa: Fagus grandifolia (american beech), Quercus alba (white oak), Quercus velutina (black oak)
  • Release date: 2015
  • Revisions:
  • EML file: knb-lter-hfr.195.4
  • DOI: digital object identifier
  • EDI: data package
  • DataONE: data package
  • Related links:
  • Study type: short-term measurement, historical, modeling
  • Research topic: historical and retrospective studies; physiological ecology, population dynamics and species interactions
  • LTER core area: disturbance
  • Keywords: dendrochronology, disturbance patterns, history, hurricane damage, modeling, successional dynamics
  • Abstract:

    Variation in forest response to hurricane disturbance in coastal New England

    Research on disturbance in forest ecosystems has generally focused on either catastrophic disturbances generating stand-replacing successional sequences, or small-scale disturbances (e.g. individual tree-fall gaps) resulting in tree-by-tree replacement. The role of moderate disturbances in forest development, in contrast, is poorly understood. While the most severe hurricanes can cause catastrophic disturbance, the vast majority of hurricanes that affect forests from the Caribbean to the northeastern United States are moderate in intensity. In this study we examine both individual and population level responses of beech (Fagus grandifolia) and oak species (Quercus spp.) to hurricanes of varying intensity in coastal Massachusetts. We characterize growth response to disturbance using a novel approach that explicitly compares the range of growth responses observed following known disturbances to the range of growth responses in non-event years. The tree species exhibited a wide range of growth and regeneration responses to hurricanes; however, only a single storm caused dramatic increases in growth and new establishment for beech. The results of this study highlight the importance of wind disturbance in the establishment and persistence of beech, and suggest that while some moderate disturbances have little or no effect on species growth and regeneration dynamics, individual storms may have substantial impacts. Forest response to wind storms of varying, but moderate intensities, depended on local site conditions, including environmental, meteorological, topographical, historical and biological factors.

    Beech dominance in a coastal New England forest

    Monodominant forests occur in a wide range of tropical and temperate ecosystems, but the mechanisms enabling their development are not well understood. This study examines the history and dynamics of beech-dominated forests in coastal New England in order to identify factors that facilitate beech dominance. We also characterize beech structural variation with respect to edaphic and environmental conditions, and describe ‘dwarf beech forests’ which have not been documented in the literature. The development of beech dominance in the study area, Naushon Island, Massachusetts, was influenced by numerous factors - selective oak harvesting in the early historical period, infrequent fire, frequent and occasionally intense hurricanes, intense deer herbivory, minimal anthropogenic disturbance over the past 150 years, and geographic isolation – and facilitated by beech’s ability to persist in the understory and reproduce vegetatively. This study highlights the importance of both disturbance history and species-specific traits in the development of monodominant forests, and suggests that these stands can persist for long periods of time in the absence of pest, pathogen, or other disturbances.

  • Methods:

    Forest history

    An unusually strong record of historical documents and maps for the study area was used to determine the timing, and in some cases extent, of natural and anthropogenic disturbance events, and to track compositional change in forested areas. Early nautical maps (Des Barres 1780) and a detailed US Coast and Geodetic Survey map (1845, scale: 1:10,000) identified areas that were forested through the 19th C., and a series of aerial photographs were used to track changes over the past century. Aerial photo-delineations were geo-referenced to US Geological Survey topographic sheets using a zoom-transfer scope, and then digitized (1938, 1951, 1971, and 1999). Study sites were chosen in areas continuously forested throughout the historical period, which we defined as areas consistently mapped as forested for which we found no contradictory documentary or field data to suggest otherwise.

    Field Data

    To characterize variation in response to hurricanes, we sampled vegetation in fixed area plots (400 m2) in the three forest types in both the East and West end forests (tall N = 6, intermediate N = 12, and dwarf N = 6). Whenever possible, sets of tall, intermediate and dwarf stands were selected in close proximity to document forest response to similar disturbance events across a topographic and forest structural gradient. Within plots, species and dbh were recorded for all trees greater than 10 cm dbh, and an increment core was taken from 15-20 trees greater than 7 cm dbh for age determination and radial growth analysis. Because of the low density of oak species in study plots, and their greater age than the more abundant beech, additional oak trees located outside of plots were cored to facilitate reconstructing long term forest history and dynamics. This included oak trees immediately adjacent to study plots, as well as trees not associated with particular study plots.

    Sound cores were collected from 647 trees: 433 beech, 146 white oak, and 68 black oak (East end N = 347, West end N = 300; tall N = 154, intermediate N = 198, dwarf N = 109 and outside-of-plot oak species N = 132). A relatively small number (N = 54) of beech and oak cores were retrieved from stands along the coastline and are included in species analyses, but not in analyses of structural types because of the limited occurrence of this type. Cores were taken as close to the base of each tree as possible (30 - 40 cm from the base) in order to determine the most accurate establishment date. Cores were dried, mounted, and sanded with increasingly fine sandpaper to reveal the cellular structure. Tree rings were counted and measured to the nearest 0.01 mm using a Velmex measuring system (East Bloomfield, NY, USA). Cores were used to determine stand age structure for modern forests, and to estimate the diameters of trees at the time of hurricanes by summing ring widths from the innermost ring to the hurricane year (excluding rotten cores and cores that substantially missed the pith). All cores were used in the radial growth analysis. A sub-sample of beech (N = 92) white oak (N = 58) and black oak (N= 25) were visually cross-dated, and verified using the program COFECHA. Cross-dated cores were used to examine the yearly growth response to hurricanes.

    Hurricane regime reconstruction

    Boose et al. (1994, 2001) have reconstructed hurricane frequency and intensity in New England from European colonization (1620) to 1997. All hurricanes with damaging winds were identified for the region by a comprehensive review of a wide range of sources: personal diaries and town histories for the period 1620-1699, contemporary newspapers from 1700-1997, and meteorological data from 1871-1997. Wind damage reports and meteorological observations for each storm were used to characterize regional patterns of damage. Boose et al. (1994, 2001) then developed a model, HURRECON, parameterized with data from recent hurricanes, which can be used to reconstruct maximum hurricane wind speeds, a measure of hurricane intensity, at specific sites. We used HURRECON to reconstruct hurricane frequency and intensity for Naushon Island for the period 1620 - 1997. HURRECON estimates of hurricane wind speeds are easily translated into a Fujita scale rating of storm severity (Fujita 1971). This rating system classifies F0 damage as a loss of leaves and branches and the uprooting of shallow rooted trees (sustained wind speeds 18-25 m/s), F1 damage as scattered blowdowns and small gaps (26-35 m/s), F2 damage as extensive blowdowns (36-47 m/s), and F3 damage as almost complete leveling of trees (greater than 48 m/s). Additionally, local documentary sources from Naushon Island - personal descriptions, anthologies and annual reports - were used to develop an independent reconstruction of hurricane strikes.

    We used the reconstructed hurricane record to select six storm events for which we had an adequate sample of tree ring data to examine in detail. The six hurricanes: 1) were characterized by Fujita scale ratings (1.4 - 2.2) greater than the median rating for all hurricanes (1.3); 2) were not preceded or followed (within six years) by another hurricane (Fujita scale rating greater than 1.4); and 3) impacted the study site according to documentary sources (excluding one storm, 1888). Less intense hurricanes were not included because these events were typically not recorded in historical documents, and were not thought to have had observable effects on population-level growth dynamics. We selected hurricanes not preceded or followed by another hurricane to determine growth response to a specific event, recognizing, however, that the effects of a previous hurricane may persist longer than six years and exert influence on wind damage and growth response to a subsequent storm. A longer period of separation between hurricanes was not possible given the high frequency of hurricanes. The selected hurricanes occurred in 1869, 1888, 1924, 1944, 1960, 1991, and had reconstructed Fujita scale ratings of 2.2, 1.4, 1.5, 1.9, 1.8, and 1.9, respectively. An earlier hurricane (1841, 1.8) was also evaluated for white oak, which was the only species that had sufficient numbers of stems predating this event. These seven hurricanes are representative of the more intense storms that have impacted the study area over the past 150 years.

    Population level growth changes in hurricane vs. non-hurricane years

    We have developed a growth change (GC) analysis that explicitly compares the range of growth responses among individuals of a population of a particular species following a hurricane to the range of growth responses in non-event years. We distinguish between two types of non-event years: non-hurricane years, i.e. those that do not fall on a hurricane year, and quiet years, i.e. those which are not preceded or followed (within eight years) by a hurricane with severity > F1. A greater separation between quiet years and hurricanes was not possible given the high frequency of hurricanes in the study area. We examined the same number of quiet and non-hurricane years as the number of hurricane years examined (N = 6 for each of the three types of years). We also examined the effect of a major cutting event on white oak growth (y = 1826). The number of trees sampled in each structural type was not sufficient to compare growth responses among the different structural types, so data from all structural types were pooled, by species, for this analysis.

    For each core, percent growth change (GC) was calculated for each hurricane and non-event year (y) using prior (M1) and subsequent (M2) ten-year growth means: GC = [(M2 - M1) / M1] x 100. We excluded cores from this analysis that were rotten or did not reach the pith. For example, for GC for the year 1944, M1 covers the period from 1935-1944 and M2 covers 1945 - 1954 (Nowacki and Abrams 1997). We examined growth changes based on ten-year averages to filter out short term tree responses to climate while detecting sustained growth responses caused by disturbance (Lorimer and Frelich 1989, Nowacki and Abrams 1997). However, by using ten-year growth averages, GC in non-event years may overlap the M1 or M2 of hurricane years. GC values were correlated with tree size in beech, with smaller trees showing greater GC. Thus, we relativized GC by multiplying GC by the diameter at year y, calculated by summing the ring widths from the innermost year to year y. GC was not correlated with tree size in white and black oak, and thus was not relativized for those two species.

    We generated GC frequency distributions for beech and oak species for the six selected hurricane years and non-event years. Differences among the shapes of GC frequency distributions following the selected hurricane years, and differences between the hurricane years and the non-event years, are interpreted as resulting from the differential response of trees to the different storms. Similar analysis of growth change frequency distributions has also been used to assess the likelihood of gap versus understory origin for seedlings (Lorimer et al. 1988). In non-event years, we expected GC would be tightly distributed, with a relatively small number of individuals exhibiting above or below average growth. In response to stand-level disturbance events, in which a minimum of ~ 25% of trees typically experience growth release (Nowacki and Abrams 1997), we expected distributions would have long right tails and would be characterized by significantly higher third quartile values relative to non-event years. We use the third quartile value as a proxy for the level of GC exhibited by the fastest growing trees. For example, a higher third quartile value in a hurricane year relative to a non-event year would indicate that 25% of trees responded to the hurricane with greater GC than the third quartile value of the non-event year.

    We report statistics characterizing differences among hurricanes, and between hurricane and non-event years: 10%, 25%, 75% and 90% quantiles, maximum, mean, median, skewness, and variance. One-way analysis of variance was used to compare these values for the three categories of years (hurricane, non-hurricane and quiet). We also present an illustrative example of GC frequency distributions following a major hurricane, 1944, and a non-event year 1934, for all species.

    Transient patterns of growth following hurricanes

    To characterize transient patterns in growth following hurricanes, we examined annual variation in growth following hurricanes using cross-dated beech, white and black oak cores. First, to determine how the mean growth change (M2) varied over the ten-year period following hurricanes, we examined variation in yearly response around the post-hurricane ten-year mean for the two hurricanes that elicited the greatest growth responses (1924 and 1944). For example, in the years immediately following the hurricanes, positive growth relative to the post-hurricane mean would indicate rapid attainment of maximum release, whereas negative growth followed by positive growth would indicate gradual release, or damage followed by recovery. Year-to-year residuals (Ry) - the difference between a measured annual ring width (Wy) and the ten-year post-hurricane mean (M2) - were calculated for individuals of all three species for the 10 years following the 1924 and 1944 hurricanes: Ry = Wy - M2. In order to compare the range of response among trees of differing species, life stages and sizes, we standardized residuals by dividing Ry by M2.

    Second, we generated mean ring width indexes for beech and oak species to examine variation in annual growth following hurricanes relative to pre-hurricane growth, and more broadly, relative to growth throughout the 150 year study period. Species indexes are composed of standardized mean indices for all cross-dated cores. Standardization involves fitting a curve or straight line to the average tree growth as it changes over time to ‘correct’ for age-related growth trends (Fritts and Swetnam 1989). Ring widths are divided by the value of the curve, and expressed as an index of the potential average growth for that year. Since we observed no overall age-related growth trends for beech, indices were calculated by dividing annual growth values by the overall mean growth increment (straight line standardization). In contrast, since oaks showed declining growth with age, indices were calculated by dividing annual growth values by expected values obtained by fitting a negative exponential or negative linear curve to measured values. To gauge the impact of climatic conditions on observed growth patterns, we examined the Palmer Drought Severity Index (PDSI) for the Massachusetts NCDC Climate Division 3. Monthly PDSI values were averaged to generate a yearly signal for the period 1895 - 2004 (NOAA CLIMVIS).

    Modern forest dynamics

    To characterize tree response to disturbance in the study area, and to determine whether structural variation is related to disturbance history, we generated stand-level disturbance chronologies for beech and oak species (East and West end data pooled), and by structural type (beech only, East and West end data pooled). By identifying the percentage of trees that experienced growth releases each decade, a disturbance chronology is used to estimate the average level of decadal small-scale disturbance, and to approximate the timing of stand-level disturbance events based on pulses in decadal release. The severity of a disturbance event is estimated by the percentage of trees released, with a stand-level disturbance defined as growth release in a minimum of ~ 25% of stems (Lorimer 1980, Nowacki and Abrams 1997).

    Criteria used to identify release vary based on tree species and size, and canopy position (Rubino and McCarthy 2004). For example, higher thresholds are typically used for shade-tolerant trees, like beech, that have greater potential for release than less shade-tolerant trees, like oak species. We identified release using criteria developed in previous studies of beech and oak growth response to disturbance (Lorimer and Frelich 1989, Nowacki and Abrams 1997, Rentch 2002). Moderate and major releases for beech were defined as a growth change of 50-100% and greater than 100% (Lorimer and Frelich 1989). Moderate and major releases for oak species were defined as a growth change of 25-50% and greater than 50% (Nowacki and Abrams 1997). Using all cores, percent growth change (GC) was calculated for all years using prior (M1) and subsequent (M2) ten-year growth means: GC = [(M2 - M1) / M1] x 100. Running comparisons of sequential ten-year means were made and release dates were assigned to years in which the maximum GC reached the predetermined threshold (Nowacki and Abrams 1997). We examined growth changes based on ten-year averages to filter out short term tree responses to climate while detecting sustained growth responses caused by disturbance (Lorimer and Frelich 1989, Nowacki and Abrams 1997).

    In addition to causing blowdown, tree death, and release among surviving trees, disturbance events can also cause significant structural damage to surviving trees. Growth suppression was identified in a sub-sample of trees using a GC threshold of -50% or less for all species (Foster 1988, Motzkin et al. 2002b). For this analysis we used the sub-samples of crossdated beech (N = 92), white (N = 58) and black oak (N = 25).


    Mineral soil samples (0-15 cm) from study plots were analyzed to evaluate the relationship between physical and chemical soil properties and forest structure. We expected soil properties and forest structure to vary along a topographic gradient, with tall forest structural types on finer textured soils, and stunted trees in dwarf types restricted to excessively drained, coarse-textured, nutrient poor soils.

    Soil samples were oven-dried (105 deg C for 48 hours) and sieved (2 mm). Samples were analyzed by Brookside Laboratories (New Knoxville, OH) to determine soil texture, pH (McLean 1982), total exchange capacity (TEC), percent organic matter (SOM%; Store 1984), and exchangeable cation and macronutrient concentrations (ppm) (P, Ca, Mg, K, Na) (Mehlich 1984). Sub-samples were ground (less than 250 micro meters) prior to total carbon and nitrogen analysis, which was determined by the Analytical Chemistry Laboratory (University of Georgia) by micro-dumas combustion.

    Data Files

    HF195-04 contains tree ring data in a standard tab-delimited format. Each line includes core ID, initial year, and ring widths in 0.001 millimeters. HF195-05 contains spatial data for the project, including ArcGIS shape files and historical maps in jpeg and tif formats. See the file Metadata for Spatial Data.pdf for details.

  • Use:

    This dataset is released to the public under Creative Commons CC0 1.0 (No Rights Reserved). Please keep the dataset creators informed of any plans to use the dataset. Consultation with the original investigators is strongly encouraged. Publications and data products that make use of the dataset should include proper acknowledgement.

  • Citation:

    Busby P, Motzkin G, Foster D, Canham C. 2015. History and Dynamics of American Beech in Coastal New England 1620-2006. Harvard Forest Data Archive: HF195 (v.4). Environmental Data Initiative:

Detailed Metadata

hf195-01: soil

  1. site: name of study plot
  2. total.n: percent total nitrogen (unit: dimensionless / missing value: NA)
  3. total.c: percent total carbon (unit: dimensionless / missing value: NA)
  4. cn.ratio: carbon to nitrogen ratio (unit: dimensionless / missing value: NA)
  5. sand: percent sand (unit: dimensionless / missing value: NA)
  6. silt: percent silt (unit: dimensionless / missing value: NA)
  7. clay: percent clay (unit: dimensionless / missing value: NA)

hf195-02: trees

  1. site: name of study plot
  2. stand.type: structural stand type
  3. size: size of study plot (unit: meterSquaredPerHectare / missing value: NA)
  4. utm.n: location of study plot (UTM coordinates) (unit: degree / missing value: NA)
  5. utm.e: location of study plot (UTM coordinates) (unit: degree / missing value: NA)
  6. tree core identification code
  7. dbh: diameter at breast height (1.4 m) (unit: centimeter / missing value: NA)
  8. dbh2: second diameter at breast height (1.4 m) (unit: centimeter / missing value: NA)
  9. dbh3: third diameter at breast height (1.4 m) (unit: centimeter / missing value: NA)
  10. species: tree species sampled
    • FAGR: Fagus grandifolia
    • QUAL: Quercus alba
    • QUVE: Quercus velutina
    • QURU: Quercus Rubra
    • QU: Quercus species
    • ACRU: Acer rubrum
    • OSVI: Ostrya virginiana
    • CAGL: Carya glabra
  11. location: whether tree is within plot or out of plot
    • P: in plot
    • OP: out of plot
  12. vigor: tree vigor
    • H: appears healthy
    • R: clearly rotten
    • D: dead
  13. canopy.position: position of tree within stand
    • D: dominant
    • CD: co-dominant
    • I: intermediate
    • S: suppressed
  14. crown.damage: level of damage to tree cown
    • ND: no damage
    • BT: broken top
  15. field.notes: notes taken in the field
  16. dendro.notes: notes taken once tree cores had been mounted, sanded, counted and measured

hf195-03: tree cores

  1. id: tree core identification code, typically abbreviated plot name followed by numerical code
  2. age: whether tree core reached (or neared) tree center, and therefore whether it was used for age determination
    • 1: tree core reached or neared pith and was used for age determination
    • -1: tree core was incomplete and was not used for age determination
  3. dbh: diameter of tree at breast height (1.4 m) (unit: centimeter / missing value: NA)
  4. species: tree species sampled
    • FAGR: Fagus grandifolia
    • QUAL: Quercus alba
    • QUVE: Quercus velutina
  5. vigor: vigor of sampled tree
    • H: appears healthy
    • R: clearly rotten
    • D: dead
  6. canopy.position: position of tree within stand
    • D: dominant
    • CD: co-dominant
    • I: intermediate
    • S: suppressed
  7. crown.damage: level of damage to tree crown
    • N: no damage
    • BT: broken top
  8. stand.type: structural stand type
    • OAK: trees sampled by reconnaissance
    • D: dwarf stand type
    • C: coastal stand type
    • H: tall stand type
    • G: intermediate stand type
  9. location: location of sampled tree
    • E: east end forest
    • W: west end forest

hf195-04: tree rings

  • Compression: zip
  • Format: text
  • Type: document

hf195-05: historical maps and GIS file

  • Compression: zip
  • Format: Esri shapefile, jpeg, pdf, tiff
  • Type: document, image, vector GIS