Page  2




Robert Tatina Department of Biological Sciences, Dakota Wesleyan University, Mitchell, SD 57301


Robinson Woods Preserve is a 32.4 ha forest in southwestern Michigan that had been cleared and cropped and then abandoned sometime in the 1920s, at which time secondary succession was initi- ated. The objective of this study is to describe the current tree species composition and the succes- sional changes that led to it. In 2011 and 2012, the T-square method and the point-centered quarter method were used to determine the current structure and composition of the forest. Data from these methods were compared with similar data from two earlier studies of the same area conducted in 1972 and 1986. Between 1972 and 2011–2012, shade-intolerant trees of Sassafras albidum and Prunus serotina had decreased in importance by 22% and 82%, respectively, while shade-tolerant trees of Fagus grandifolia, Quercus rubra and Acer rubrum had increased by 26,790%, 352% and 21%, respectively. Between 1986 and 2011–2012, the density of most species had declined, except for trees of Fagus grandifolia, whose density had increased by 44%. In addition, trees of S. albidum had been reduced in importance to a subdominant role in the presence of A. rubrum and Q. rubra. Seedling and sapling densities by size class show that most of the canopy trees are reproducing them- selves. However, no Ulmus americana and Fraxinus americana trees larger than 45 cm dbh were en- countered, the former having been killed by Dutch elm disease and the latter by the emerald ash borer. Over the 90 years since abandonment, a hardwood forest has developed, one in which pioneer tree species have been replaced in part by shade-tolerant trees. Based on these results, it is expected that the future forest will likely become an American beech–sugar maple forest, especially in the ab- sence of periodic fire.

KEYWORDS: secondary succession, tree composition, Berrien County


The southeastern edge of Lake Michigan has been a fruitful place to study the development of plant communities. It was there in the late 1800s that Henry Cowles carried out his pioneering research into primary succession and de- scribed the process as it occurred from the shore to adjacent coastal dunes (Cowles 1899). Later his work was extended by Olson (1958), who used quanti- tative measures to describe the various plant communities, their ages, and the rates at which they develop and are replaced. Cain (1935) initiated a series of plant community development studies at Warren Woods State Park, an old growth, beech-maple forest located within a mile of Lake Michigan, and con- cluded that sugar maple (Acer saccharum Marsh) was inexplicably increasing in abundance. Subsequently, the same forest was studied by Brewer and Merritt (1978), who found that gaps caused by wind-throw of canopy trees does not in-


crease the diversity of canopy species, because sugar maple or beech replace the downed tree. Woods (1979) explained the maintenance of codominance of sugar maple and American beech as a process of reciprocal replacement in which a gap left by a maple was filled by a beech, and vice versa. Poulson and Platt (1996) attributed the continued codominance of beech and maple to their specific re- sponses to light intensity in understory shade and in gap openings such that maple is favored by large gaps and beech by small gaps. Brewer et al. (1984), using land survey records, showed that most of pre-settlement southwestern Michigan contained forests of American beech and sugar maple. Finally, Don- nelly and Murphy (1987) concluded that the current composition of the old- growth beech–maple forest at Warren Woods State Park had changed little since pre-settlement times. Most of the state has since been logged, but some areas have developed into second-growth hardwood forests, albeit smaller in area and more fragmented (Dickmann and Leefers 2003).

Second growth forests in Berrien County in southwestern Michigan have been the subject of several studies. Wells and Thompson (1982) described the plant communities of the forested sand dunes, concentrating mostly on those of Grand Mere State Park. Smith and Woodland (2007) described the plant com- munities of the forested dunes at Warren Dunes State Park as a mosaic of plant communities dominated by Quercus rubra and Acer saccharum. Tatina (2010) described the composition of a small forest inland from the sand dunes, but near Warren Dunes State Park, as one in which Quercus rubra dominated drier sites and Acer rubrum wetter sites.

Community composition and dynamics have been studied in forests of the Great Lakes region. For example, Abrell and Jackson (1977) reported small changes in density and basal area of trees in an old growth beech-maple forest in Indiana. Abrams and Scott (1989) concluded that disturbance could accelerate rates of succession in forests in northern lower Michigan. Tyrell (2003) reported that fire suppression was responsible for declines in red oak and increases in sugar maple in forests of southeastern Wisconsin. Holmes (2006) showed that sugar maple was increasing and American beech decreasing following large scale disturbance in an old growth forest in Indiana. Most other legacy studies in eastern hardwood forests have shown that the composition of these forests pre- dictably changes over time, resulting in dominance by more shade-tolerant species, especially by Acer saccharum or Fagus grandifolia or both (Ebinger 1986, Wilder et al. 1999, Galbraith and Martin 2005, Pierce et al. 2008, Pinheiro 2008). Finally, others have shown that once American beech and sugar maple achieve canopy dominance, gap phase replacement results in community devel- opment changing from a successional process to dynamic equilibrium (Vankat et al. 1975, Diggins 2013).

Even though the composition of and succession in old growth forests and the composition of second growth forests in southwestern Michigan have been de- scribed, none of the reports has examined secondary succession starting from an old field, in spite of the fact that much of the area has been logged, and some of it farmed. Beckwith (1954) and Foster and Gross (1999) did study succession on old fields, but their studies were terminated before the development of forests at these sites.


Robinson Woods Preserve (RWP) provides an ideal opportunity to examine secondary succession, because its past disturbances are known and because its post-disturbance vegetational composition had been described previously. The forest at RWP was first studied in 1972 by Carter (1972), who described 62 plant communities based on the importance values of trees and shrubs. He also in- cluded an oral history by a nearby resident who described the land use of the pre- serve prior to its abandonment. Using floristic data, Riess (1986) repeated the Carter study and described ten plant communities. However, Riess did not con- nect her findings to Carter’s earlier results. The objective of the present study was to trace the development of the forest at RWP by comparing its current com- position with those reported by Carter (1972) and by Riess (1986). Studies such as the present one can provide benchmarks to allow detection of effects due to human-induced climate change (Dale et al. 2001, McKenney et al. 2007) and forest fragmentation; it has been predicted that the latter activity will remove 1.2 million acres of forest in the Lake States region by 2030 (Alig and Plantinga 2004). Examples of the consequences of forest fragmentation abound in the lit- erature and include increases in lyme disease (Allan et al. 2003) and in the fre- quency of invasive non-native species (Yates et al. 2004), selective losses of na- tive species (Tallmon et al. 2003), reduced nutrient cycling, increased soil erosion, and decreases in water quality (Newbold et al. 2016).


Site Description

Robinson Woods Preserve (Figure 1) is located in the Michigan Lake Plain Ecoregion (USGS 2010) of southwestern Michigan (Berrien County, Chikaming Township, N41o51., W86o38.) about

2.2 km east of Lake Michigan and 1.6 km directly east of the village of Lakeside. The 32.4 ha sec- ond growth forest is owned by Chikaming Open Lands of Sawyer, Michigan, a land conservancy or- ganization. Carter (1972) described the history of the property beginning in 1912 based on communications with Charles Kull, who lived nearby. Prior to the end of World War I, the land was farmed, and the southern half of the property was planted to orchards and row crops. Shortly after the war, drainage ditches were dug, and an attempt was made to place tiles, but the tiling project was abandoned before completion (Carter 1972). The farm was likely abandoned at that time.

An aerial photo of the area taken in June 1938 shows that most of the southern portion of the pre- serve contained abandoned cropland (Figure 2). Forested portions existed at the northern end and along part of the eastern border. Cores taken in 2012 of larger oak trees (mean dbh = 76.6 cm ± 8.5 cm, N = 10) in the forested parts of the preserve had a mean age of 100.5 ± 16.6 years, which sup- ports the conjecture that the area had been logged in the late 1800s.

In 1966, Jean C. and William S. Robinson purchased the property and built a cabin at the north- ern end. Two years later they donated the southern 26.3 ha to The Nature Conservancy (TNC). In 1973, the Robinsons donated the remaining 6.1 ha to TNC, and in 2009, the entire 32.4 ha were trans- ferred by TNC to Chikaming Open Lands.

The climate at RWP, moderated by Lake Michigan, becomes neither very hot nor very cold. Based on data from 1981 to 2010 at South Bend, Indiana, (39 km east-southeast of RWP), the aver- age high temperature of 28.2 °C occurs during July, and the average low of –7.8 °C occurs in Janu- ary. Precipitation is evenly distributed over the year, averaging 96.5 cm (U. S. Climate Data 2016).

At RWP the elevation varies between 186 m and 201 m, with the lowest areas at the bottom of the ravines running across the northern end and at a creek at the eastern edge of the property. The re- mainder of the property is level or gently undulating.

Three soil types have been mapped in the preserve (Larson 1980). Oakville fine sand is a shal- low, dark brown, acidic soil derived from glacial outwash that is rapidly permeable, producing soils


FIGURE 1. Robinson Woods Preserve, Berrien County, Michigan, showing approximate locations for dominant tree species. FAGR-ACSA = Fagus grandifolia–Acer saccharum, QURU = Quercus rubra, ACRU = Acer rubrum, SAAL = Sasafrass albidum, QUAL = Quercus alba, ULAM = Ulmus americana.

that wet rapidly, but dry quickly. It is found on the higher, drier parts at the northern and south cen- tral parts of the preserve and favors a dry southern forest community as described by Kost et al. (2007). Morocco loamy sand is a shallow, dark grayish brown, acidic soil with rapid permeability whose parent material is loamy glacial till. At RWP this type is located where the surface is level and may contain standing water at some time during the year. Rimer loamy fine sand, an acidic to neutral soil of very slow permeability that has developed from sandy glacial and lake deposits, is found in low, level parts of the property and in wet years may remain wet for much of the growing season. Areas of RWP underlain by Morocco and Rimer soils are located on the southeastern and southwest- ern parts of the property and would favor trees of a mesic southern forest community.

Sampling Methods

Sampling protocols described by Carter (1972) and Riess (1986) were followed to set up 13 south to north transects running the length of the property. The transects were situated 30 m from the prop-


FIGURE 2. Aerial photo images from 1938, 1950 and 1967 (left to right). Robinson Woods Preserve is enclosed by black lines in each photo. The light grey area at the southern half of the preserve in the 1938 photo is probably the part that had been cleared and then planted to row crops and orchards. Over the 29 years these photos span, trees can be seen to fill in much of the area that had been cleared. Photos were provided by RS & GIS Research and Outreach Services, Michigan State University. Used with permis- sion.

erty boundaries and were 30 m apart. In June to September, 2011 and 2012, 650 points were located 15 m apart along these transects, and a table of random numbers was used to select 317 of these points for sampling. This differed from the methods of Carter (1972) and Riess (1986), who used evenly spaced sampling points. Because Carter (1972) included trees . 10 cm dbh (diameter at breast height = 1.5 m above the soil surface) within the southern 26.3 ha only, while Riess (1986) sampled trees > 5 cm dbh within the entire 32.4 ha, two methods were used in this study to sample the trees.

Sampling and Calculations Relative to Carter (1972)

The T-square method (Greenwood 1996) was used to locate two trees . 10 cm in diameter at 1.5 m above ground level, and their circumferences at breast height were recorded by species at sampling points in the southern 26.3 ha of the preserve. The first tree was that nearest the sampling point, and the second was the tree nearest the first but outside a line perpendicular to a line drawn between the first tree and the sampling point. Circumferences were converted to diameters and diameters to basal areas. The number of sampling points for each species was divided by the total for all species and multiplied by 100 to yield the relative frequency value for each species. The number of trees of each species was divided by the total for all species and multiplied by 100 to produce the relative density value for each species. The basal area for each species was divided by the total for all species and multiplied by 100 to calculate the relative dominance value for each species. Finally, the three rela- tive values for each species were summed to yield an importance value for each species. This value allows each species to be compared to the others based on frequency of occurrence, density and basal area. Density of trees was calculated from T-square distance measurements using the following for- mula: density = n2/2.2828.x.y, where n is the number of sampling points, x is the distance to the nearest tree from a sampling point, and y is the distance from the first tree to its nearest neighbor (Greenwood 1996).

Saplings (that is, trees > 30 cm tall and <10 cm dbh) were counted in each of 5 diameter size classes (< 2; . 2 < 4; . 4 < 6; . 6 < 8; . 8 < 10 cm) within a 0.01 ha circular plot centered on each sampling point. These results were used to calculate the relative values, which were then summed to yield importance values. Seedlings (tree species < 30 cm tall) were counted in 0.002 ha circular plots centered on each sampling point. Their numbers were used to calculate seedling density on a per hectare basis.

Tree and sapling densities were compared with the density data reported by Carter (1972). Be-


cause it was impossible to locate the boundaries of the 62 communities of Carter (1972), who de- scribed each community by importance values and densities for trees and saplings, his values for these were converted to reflect the entire 26.3 ha. This was done by cutting out each community from a paper copy of his map of the 62 communities (Figure 1 in Carter 1972) and weighing each sepa- rately to the nearest mg. Because his delineation of each community is an approximation, it was felt that using optical methods to determine the area of each community would not improve the accuracy of the measurements. All the weights were then totaled, and the proportion of the entire area repre- sented by each community was determined. The sapling and tree densities for each community were each multiplied by the proportion for that community to determine the actual densities, and the sum of all of these densities was divided by 26.3 ha to determine the total density of saplings and trees per hectare. The importance value for each community was multiplied by the appropriate proportion and then summed by species for the entire sampling area. The sum for each species was then divided by the sum for all species and the resulting quotient multiplied by 100 to obtain the importance value.

Sampling and Calculations Relative to Riess (1986)

At 317 sampling points randomly distributed along north-south transects within the entire 32.4 ha, the circumference at breast height of four trees whose dbh > 5 cm and the distance in meters be- tween those trees and the sampling point were measured following the point-centered quarter method of Cottam and Curtis (1956). The number of sampling points at which each tree species occurred was used to determine relative frequency, and the number of trees was used to calculate relative density. The tree density (number of trees per ha) of each species was calculated by the following formula: 10,000 x (number of individuals of the given species / number of individuals of all species) / (mean distance from points to trees)2. Finally, tree circumferences were summed for each species, and the total of all species was calculated. The sum for each species was then divided by the total for all species and the quotient multiplied by 100 to yield the relative dominance of that species.


A comparison of aerial photos from 1938 to 1967 and of the tree species com- position of Robinson Woods Preserve from 1972 to the present shows the changes that have occurred in the vegetation since the property was abandoned.

Changes before 1972

Aerial photos (Figure 2) show that in 1938 the southern half of RWP had not yet developed a cover of trees in the portions of the preserve that had been cleared and cropped. By 1950 a cover of trees is apparent, especially in the northern portion of the cropped area, with isolated clumps of trees scattered throughout the southern portion. In 1967, most of the preserve, except for the southeastern corner, was forested. Current aerial images of RWP, such as those available on Google Earth, show the entire property to be forested.

Changes between 1972 and 2011–12

With importance values (IV) of 74.81, 65.21 and 34.38, respectively, Acer rubrum L., Sassafras albidum (Nutt.) Nees, and Prunus serotina Ehrh. were the most abundant saplings on the southern 26.3 ha in 1972 (Table 1). Of the 62 communities Carter (1972) described, 37 were dominated by Sassafras albidum and Prunus serotina, and 52 contained these two species as trees and/or saplings. As shown in Table 1, S. albidum and P. serotina had high importance values as


TABLE 1. Importance values (IV) for saplings and for trees, respectively, and the percentage change in each, for the species recorded at Robinson Woods Preserve in 1972 (calculated from statistics pro- vided in Carter 1972) and in 2011–2012 (this study). An asterisk (*) in the percentage change col- umn indicates that the species was not encountered at sampling points in 1972. The importance val- ues in 1972 do not add up to 300, because Carter excluded species with low values.

Saplings Trees

IV in IV in Percentage IV in IV in Percentage Species 1972 2011–12 Change 1972 2011-12 Change Acer rubrum 74.81 56.56 –24 71.48 86.63 21 Quercus rubra 15.00 26.79 79 12.24 55.32 352 Sassafras albidum 65.21 20.39 –69 58.44 45.63 –22 Quercus alba 5.41 19.84 267 21.51 28.34 32 Fagus grandifolia 1.60 74.10 4531 0.10 26.89 26790 Liriodendron tulipifera 3.42 5.27 54 12.61 15.19 20 Nyssa sylvatica 12.05 19.71 64 13.08 8.86 –32 Prunus serotina 34.38 9.42 –73 45.07 8.28 –82 Fraxinus americana 7.58 6.94 –8 4.63 5.45 18 Acer saccharum 1.35 5.81 330 0.50 4.87 874 Ulmus americana 3.51 12.07 244 4.33 3.52 –19 Pinus strobus 0 0 * 0 2.82 100 Populus deltoides 0 0 * 5.52 2.20 –60 Carya ovata 0 10.41 * 0.95 1.58 66 Tilia americana 0.73 0.83 14 0.81 1.49 84 Quercus palustris 3.30 4.24 28 1.49 0.96 –36 Populus grandidentata 10.94 0 –100 9.97 0.51 –95 Amelanchier arborea 0.28 5.34 1807 0 0.49 100 Cornus florida 2.00 3.69 85 0 0.49 100 Quercus muehlenbergii 0 0 * 0 0.49 100 Populus tremuloides 2.37 0 –100 1.21 0 –100 Carpinus caroliniana 1.35 7.42 450 0 0 0 Pyrus malus 0 0 * 3.17 0 –100 Rhus typhina 0 0 * 0.11 0 –100 Salix sp. 0 0 * 0.44 0 –100 Ostrya virginiana 1.18 1.77 50 0 0 0 Crataegus sp. 3.23 2.19 –32 0.32 0 –100 Acer saccharinum 0 0.18 * 1.39 0 –100 Hamamelis virginiana 0 2.97 * 0 0 0 Asimina triloba 0 0.25 * 0 0 0 Cornus sp. 0 0.84 * 0 0 0 Viburnum lentago 0 0.24 * 0 0 0 Rhus copallina 0.29 0 –100 0 0 0 Betula papyrifera 0.21 0 –100 0 0 0 Ulmus rubra 1.78 0 –100 0 0 0 Prunus pensylvanica 0 2.73 * 0 0 0 Totals 251.96 300 269.37 300.01 Indviduals/ha 1042 948 –9 527 466 –12

saplings and as trees, second only to Acer rubrum, which was the dominant tree in 20 communities and a dominant sapling in 19 (Carter 1972). Between 1972 and 2011–2012, 10 species of saplings had declined in importance (Table 1). These included Acer rubrum (–24%), Sassafras albidum (–69%), Prunus serotina (–73%), and Fraxinus americana (-8%). Thirteen species of saplings


had increased in importance. These included Quercus rubra L. (79%), Quercus alba L. (267%), and Fagus grandifolia Ehrh. (4531%). Five species present as saplings in 1972, including Populus tremuloides Michx., Rhus copallina L. and Betula papyrifera Marsh., were not recorded in 2011 and 2012. Seven sapling species not previously noted were recorded in samples in 2011 and 2012; among these were Carya ovata (Mill.) K. Koch, Hamamelis virginiana L., Asimina triloba (L.) Dunal, and Prunus pensylvanica L.f.

By 2011–2012, 13 species of trees had increased in importance (Table 1), most notably Acer rubrum (21%), Quercus rubra (352%), Quercus alba (32%), and Fagus grandifolia (26,790%). Trees of Sassafras albidum and Prunus serotina had decreased in importance by 22% and 82%, respectively.

During the 40 years between sampling dates, the density of saplings had de- creased by 9% and that of trees by 12% (Table 1). In 1972 the dominant tree species were the same as the dominant sapling species; of these, only trees of Acer rubrum retained their dominant position in 2011 and 2012.

Changes between 1986 and 2011–12

Twenty-seven species were recorded in 1986 and 25 in 2011 and 2012 (Table 2). Over the 25 or 26 years between sampling events, the density of most species declined, except for trees of Fagus grandifolia, whose density had increased by 77%.

In 1986 the species with the greatest importance values were Sassafras al- bidum (IV = 71.53), Acer rubrum (IV = 59.96), and Quercus rubra (IV = 50.61) (Riess 1986 and Table 2). By 2011 and 2012, trees of S. albidum (IV = 33.29) had been reduced to a subdominant role in the presence of A. rubrum (IV = 71.33) and Q. rubra (IV = 63.97), whereas trees of both of the latter species showed increases in importance values, even though the density of Q rubra had declined from 198 to 132 trees per hectare. An additional 15 species increased in importance values over the 25-to 26-year period (Table 2). Trees of Fagus gran- difolia had a large increase in importance value (from 12.55 to 27.22) due to a 76.75% increase in density. Liriodendron tulipifera L. (27%), Quercus alba (24%), Nyssa sylvatica Marsh. (42%), and Acer saccharum (46%) likewise ex- hibited increases in their importance values; of these, L. tulipifera (–49%) and Q. alba (-41%) decreased in density, while trees of N. sylvatica (6%) and A. sac- charum (2%) increased slightly. Finally, the trees of 13 species decreased in im- portance values, most notably, Sassafras albidum (–75%), Prunus serotina (- 71%), and Fraxinus americana L. (–75%).

Current Composition

Compositional statistics for trees and for saplings and seedlings on the 32.4 ha are presented in Appendices A and B for use in future legacy studies. In 2011–2012, 17 species of trees were measured producing a total density of

431.51 trees per hectare. Quercus rubra (IV = 72.61) and Acer rubrum (IV = 71.46) were the two most important species (Appendix A), Q. rubrum on well drained soils and Acer rubrum on poorly drained soils. Seedling and sapling den- Page  10 10 THE MICHIGAN BOTANIST Vol. 55

TABLE 2. Density and importance values of trees (>5 cm dbh) at Robinson Woods Preserve in 1986 and in 2011–12. Density values for 1986 are calculated from statistics in Riess (1986); importance values for 1986 are takes from Riess (1986).

Density (trees/ha) Importance Value Percentage Percentage Species 1986 2011–12 Change 1986 2011-12 Change Sassafras albidum 328 82.707 –75 71.53 33.291 –53 Acer rubrum 271 175.618 –35 59.96 71.328 19 Quercus rubra 198 132.116 –33 50.61 63.968 26 Liriodendron tulipifera 57 29.001 –49 18.87 23.982 27 Quercus alba 65 38.131 –41 18.69 23.252 24 Prunus serotina 50 14.501 –71 14.13 6.391 –55 Fagus grandifolia 43 76.262 77 12.55 27.218 117 Fraxinus americana 41 10.204 –75 9.51 3.894 –59 Nyssa sylvatica 33 34.909 6 9.15 12.948 42 Acer saccharum 29 29.538 2 6.74 9.863 46 Cornus florida 24 3.222 –87 5.51 1.225 –78 Ulmus americana 18 16.649 –8 4.23 5.340 26 Quercus palustris 13 8.056 –38 3.51 3.403 –3 Carpinus caroliniana 9 6.445 –28 2.27 2.014 –11 Tilia americana 7 1.611 –77 2.27 0.669 –71 Ostrya virginiana 7 4.834 –31 1.54 1.802 17 Hamamelis virginiana 7 0.537 –92 1.44 0.203 –86 Populus grandidentata 5 1.611 –68 1.35 0.916 –32 Populus tremuloides 5 0.000 –100 1.21 0.000 –100 Crataegus sp. 6 2.685 –55 1.07 0.789 –26 Carya ovata 3 4.834 61 0.95 1.955 106 Amelanchier arborea 4 3.759 –6 0.83 2.627 216 Populus deltoides 2 2.148 7 0.7 1.211 73 Betula papyrifera 2 0.000 –100 0.45 0.000 –100 Asimina triloba 2 1.074 –42 0.36 0.287 –20 Pinus sylvestris 2 0.000 –100 0.36 0.000 –100 Acer saccharinum 0 1.074 * 0 0.511 * Quercus muehlenbergii 0 0.537 * 0 0.214 * Pinus strobus 0 1.074 100 0 0.701 * Totals 1232 683.136 –45 300 300.000

sities by size class (Appendix B) show that most of the canopy trees are repro- ducing themselves. However, no Ulmus americana and Fraxinus americana trees larger than 45 cm dbh were encountered.


When RWP was abandoned after having been logged and then cleared in part for row crops and an orchard, the forest began to recover in the process of sec- ondary succession. While there is no record of the nature of the plant communi- ties that followed immediately after abandonment in the early 1920s and until 1972, studies of old field succession from nearby areas offer clues as to the likely series of stages and their dominant plant species. Beckwith (1954) described


four stages within the first 25 years following abandonment of cropped land in Washtenaw County, Michigan, about 240 km east-northeast of RWP: an an- nual–biennial stage, followed by a grass and perennial herb stage, which is re- placed by a mixed herbaceous perennial stage. The fourth stage, which is domi- nated by shrubs, may be initiated by propagules of wind-dispersed species followed by propagules of bird-dispersed species (Foster and Gross 1999). While the duration of each stage depends on the specific type of agricultural practice that had occurred on the land (Beckwith 1954), Evans and Dahl (1955) and Weigert and Evans (1964) have found that the herbaceous perennial stage of old fields in Michigan may persist for longer than 50 years. Beckwith (1954) and Huberty et al. (1998) each provide a list of possible plant species that dominate each stage.

Although the first 50 years of secondary succession at RWP can only be pre- sumed as a matter of extrapolation from nearby old fields, the record shows that after those 50 years, it had the characteristics of a young forest, one where dom- inance was shared by shade-intolerant pioneer species as documented by Carter (1972).

Oliver and Larson (1996) describe forest development as a sequence of four stages: 1) a stand initiation stage, 2) a stem exclusion stage, 3) an understory re- initiation stage, and 4) an old growth stage. If the stand development described by Oliver and Larson (1996) is fitted to the aforementioned description, then the forest in 1972 was late in the first stage (stand initiation) of forest development, in which pioneer species, developing from propagules carried into the area or from saplings and or a seed bank that had survived the clearing of the land, were the dominants, and new species were continually entering. At that time, the short stature of the trees lent a scrubby appearance to the forest (J. Carter pers. comm. 2012).

In the 40 years since the Carter (1972) study, the forest has changed markedly. Sapling and tree densities declined as competition for sunlight, water, and soil nutrients (Coomes and Grubb 2000) winnowed out the less adapted, most of which were trees of shade-intolerant species. Gains were made by saplings of shade-tolerant species, most notably Hamamelis virginiana and Asimina triloba, two species commonly present in the understory of nearby old growth forests (Cain 1935). Acer rubrum trees, already important in 1972, were among the dominants in 2011 and 2012. Shade-intolerant species disappeared from the canopy as the area started to resemble a mature forest with abundant oaks. Al- though they were still not dominant in 2011 and 2012, trees of Fagus grandifolia and Acer saccharum showed large gains in importance values, perhaps foreshad- owing their position in the future. Where disturbance was greater due to plowing, the portion of the forest where row crops had been planted is currently in the stem exclusion stage of Oliver and Larson (1996), in which canopy closure be- gins to eliminate shade-intolerant species, and the portion where disturbance had been limited to logging is in the understory re-initiation stage, which is charac- terized by advance regeneration of shade tolerant species.

Riess’s (1986) study of the same area allows for a corroborating look at and fine tuning of those stages of forest development that have occurred since 1986. By 1986, Sassafras albidum and Acer rubrum were still dominants, as they were


in the Carter (1972) study, but trees of Quercus rubra had increased in impor- tance value, either through recruitment from advance reproduction or because more oaks were included with the additional 15 acres that Riess (1986) had sam- pled. However, of the nine forested communities she described, only two con- tained Sassafras albidum trees as a dominant; the others had shade-tolerant species (e.g., Acer rubrum and Quercus rubra, Fagus grandifolia, and Acer sac- charum) as dominants. Thus, by about 65 years after abandonment, the forest had already entered the understory re-initiation stage of Oliver and Larson (1996). Since then the changes that have occurred have paralleled those de- scribed earlier in the comparison between 1972 and 2011-2012. From 1986 to 2011-2012, species diversity remained constant, total tree density declined, and dominance shifted from Sassafras albidum to Acer rubrum and Quercus rubra. In addition, there was a noteworthy increase in the density and importance value of Fagus grandifolia.

Abrams (1998) has shown that the number of Acer rubrum trees has been in- creasing in eastern deciduous forest since pre-settlement times and can effec- tively compete for both wet and dry sites, while Acer saccharum will do the same on mesic sites. Acer saccharum has also been shown to replace oaks in the absence of fire (Abrams 2003). If the trends observed in the composition of the forest at RWP continue, and the predictions of Abrams (1998, 2003) are taken into account, then the forest at RWP may come to look more like a Fagus gran- difolia–Acer saccharum–Acer rubrum forest. However, in light of recent predic- tions about global climate change, determining the future of any forest must be considered to be extremely tentative.

The present forest, which is in a mid-successional stage, is undergoing further development as shown in the size class distributions of its trees (Appendix B). Interpretation of these leads to the hypothesis that the shade-tolerant species will continue to replace themselves, with the wetter parts dominated by Acer rubrum trees, and the drier parts becoming an oak forest of Quercus rubra and Q. alba. Exclusion of fire will probably favor sugar maple to the detriment of the oaks, as has been found in many other eastern deciduous forests (Abrams 2003) and thus the drier forest may become a beech-maple forest. Thus, as the forest at RWP de- velops to the old growth stage, its composition and dynamics may resemble that of two nearby forests. One is Warren Woods State Park, located 2.6 km east- southeast from RWP. It has been described as an old growth beech-maple forest that is sustaining itself through gap-phase replacement of Fagus grandifolia and Acer saccharum (Brewer and Merritt 1978; Woods 1979; Donnelley and Murphy 1987; Poulson and Platt 1996). Size class distributions (Appendix B) of trees at RWP show that both Fagus grandifolia and Acer saccharum are increasing and, in the absence of major disturbance, may become the dominants in an old growth beech-maple forest. A similar, albeit smaller, beech-maple forest, Toumey Wood- lot, 200 km northeast, which had been protected from timbering and grazing for 92 years, has been shown to be changing as sugar maple was increasing and American beech was decreasing slightly (Schneider 1966). In it, tree density in- creased over the most recent 20-year period. At RWP, tree density decreased by almost 50% as a consequence of the loss of pioneer species. It is interesting to speculate that if the pattern at Toumey Woodlot is the norm for old growth


forests in this area, then at some point in time the downward trend in density at RWP may be reversed.

Future trends extrapolated from past events assume that environmental condi- tions remain relatively constant, a condition which may not hold as a conse- quence of global climate change. The predicted increases in average tempera- tures will cause a northern shift in the distribution of some tree species (McKenney et al. 2007). While none of the species at RWP appear to be at risk of extirpation due to a northern shift (see Woodall et al. 2010), changes in tem- perature may alter competitive interactions favoring an unpredictable change in species composition. Climate change is also predicted to be accompanied by changes in the duration and frequency of disturbances, such as drought, fire, storms, insect outbreaks and invasions of alien species (Dale et al. 2001). If such disturbances create large gaps in the current forest, Acer saccharum may be fa- vored over Fagus grandifolia (Poulson and Platt 1996). If droughts are prolonged with or without increased frequency of forest fire, oaks may be favored over beech and maples (Abrams 2003) and the entire forest might become a Quercus rubra—Q. alba stand.


Thanks are extended to Chikaming Open Lands for permission to study the forest at RWP and to Peg Kohring, Midwest Regional Director, The Conservation Fund, Pamela Smith, CSU Colorado Natural Heritage Program, Nancy Baird, and an anonymous reviewer for their careful reading of ear- lier versions of this manuscript and for their suggestions which improved it.


Abrams, M. (2003). Where have all the white oaks gone? BioScience 53: 927–939. Abrams, M. D. (1998). The red maple paradox. Bioscience 48: 355–364. Abrams, M. D., and M. L. Scott. (1989). Disturbance-mediated succession in two Michigan forest

types. Forest Science. 35: 42–49. Abrell, D. B., and M. T. Jackson. (1977). A decade of change in an old growth beech-maple forest in Indiana. American Midland Naturalist 98: 23–32. Allan. B. F., F. Keesing, and R. S. Ostfeld. (2003). Effect of forest fragmentation on Lyme disease risk. Conservation Biology 17: 267–272. Alig, R., and A. J. Plantinga. (2004). Future forestland area: Impacts from population growth and other factors that affect land values. Journal of Forestry 102: 19–24. Beckwith, S. L. (1954). Ecological succession on abandoned farm lands and its relationship to wildlife management. Ecological Monographs 24: 349–376. Brewer, L. G., T. W. Hodler, and H. A. Raup. (1984). Presettlement vegetation of southwestern Michigan. The Michigan Botanist 23: 153–156. Brewer, R., and P. G. Merritt. (1978). Wind throw and tree replacement in a climax beech-maple for- est. Oikos 30: 149–152. Cain, S. A. (1935). Studies on virgin hardwood forest: III. Warren’s Woods, a beech-maple climax forest in Berrien County, Michigan. Ecology 16: 500–513. Carter, J. (1972). A floristic analysis and mapping study in the Robinson Preserve. M.A. thesis, An- drews University, Berrien Springs, Michigan. Coomes, D. A., and P. J. Grubb. (2000). Impacts of root competition in forests and woodland: a the- oretical framework and review of experiments. Ecological Monographs 70: 171–207. Cottam, G., and J. T. Curtis. (1956). The use of distance measures in phytosociological sampling. Ecology 37: 451–460.


Cowles, H. C. (1899). The ecological relations of the vegetation on the sand dunes of Lake Michigan. Botanical Gazette 28: 95–117, 167–202, 281–308, 361–391.

Dale, V. H., L. A. Joyce, S. McNulty, R. P. Nielson, M. P. Ayres, M. D. Flannigan, P. J. Hanson, et al. (2001). Climate change and forest disturbances. BioScience 51: 723–734.

Dickmann, D. I., and L. A. Leefers. (2003). The forests of Michigan. University of Michigan Press, Ann Arbor.

Diggins, T. P. (2013). A 300-year successional sequence in an Eastern United States riparian hard- wood forest. Journal of the Torrey Botanical Society. 140: 65–88.

Donnelley, G. T., and P. G. Murphy. (1987). Warren Woods as forest primeval: A comparison of for- est composition with pre-settlement beech-sugar maple forests of Berrien County, Michigan. The Michigan Botanist 26: 17–24.

Ebinger, J. 1986. Sugar maple, a management problem in Illinois forests? Transactions of the Illinois Academy of Science 79: 25–30.

Evans, F. C., and E. Dahl. (1955). The vegetational structure of an old field community in southeast- ern Michigan and its relation to environmental factors. Ecology 36: 685–706.

Foster, B. L., and K. L. Gross. (1999). Temporal and spatial patterns of woody plant establishment in Michigan old fields. American Midland Naturalist 132: 229–243.

Galbraith, S. L., and W. H. Martin. (2005). Three decades of overstory and species change in a mixed mesophytic forest in eastern Kentucky. Castanea 70: 115–128.

Greenwood, J. J. D. (1996). Basic techniques. Pp. 60–62 in Ecological census techniques, W. J. Sutherland, editor. Cambridge University Press, New York, N.Y.

Holmes, T. (2006). Four decades of change on old growth beech-maple forest in Indiana. M.A. the- sis, University of Missouri-Columbia., Columbia, Missouri. Available at https://mospace. (Ac- cessed October 18, 2012).

Huberty, L. F., K. I. Gross and C. J. Miller. (1998). Effects of nitrogen addition on successional dy- namics and diversity in Michigan old fields. Journal of Ecology 86:794–803.

Kost, M.A., D.A. Albert, J.G. Cohen, B.S. Slaughter, R.K. Schillo, C.R. Weber and K.A. Chapman. (2007). Natural communities of Michigan. Michigan Natural Features Inventory, Report Number 2007-21, Lansing, Michigan.

Larson, J. D. (1980). Soil survey of Berrien County, Michigan. USDA Soil Conservation Service, Washington, D.C.

McKenney, D. W., J. H. Pedlar, K. Lawrence, K. Campbell, and M. F. Hutchinson. (2007). Potential impacts of climate change on the distribution of North American trees. BioScience 57: 939–948.

Newbold, T., L. N. Hudson, A. P. Arnell, S. Contu, A. De Palma, S. Ferrier, S. L. L. Hill, et. al. (2016). Has land use pushed terrestrial biodiversity beyond the planetary boundary? A global assessment. Science 353: 288–291.

Oliver, C. D., and B. C. Larson. (1996). Forest stand dynamics (Update edition). Wiley and Sons, New York, New York.

Olson, J. S. (1958). Rates of succession and soil changes on southern Lake Michigan sand dunes. Botanical Gazette 119: 125–169.

Pierce, A. R., W. R. Bromer, and K. N. Rabenold. (2008). Decline of Cornus florida and forest suc- cession in a Quercus-Carya forest. Plant Ecology 195: 45–53.

Pinheiro, N. R. (2008). Composition, structure and stand development of old-growth and second- growth beech-maple stands of Crall woods, north-central Ohio. M.S. thesis, Ohio State University, Columbus, Ohio. Available at 20Pinheiro%20Thesis%20Final.pdf. (Accessed October 12, 2012).

Poulson, T. L., and W. J. Platt. (1996). Replacement patterns of beech and sugar maple in Warren Woods, Michigan. Ecology 77: 1234–1253.

Riess, A. H. (1986). A floristic study of the Robinson Preserve, Chikaming Township, Berrien County, Michigan. M.A. thesis, Andrews University, Berrien Springs, Michigan.

Schneider, G. (1966). A 20-year investigation in a sugar maple-beech stand in southern Michigan. Research Bulletin 15. Agriculture Experiment Station, Michigan State University. Lansing.

Smith, P. F., and D. W. Woodland. (2007). Forest composition study of the Great Lakes coastal forest at Warren Dunes State Park, Berrien County, Michigan. The Michigan Botanist 46: 33–62.

Tallmon, D. A., E. S. Jules, and N. J. Radke. (2003). Of mice and men: Cascading effects of forest fragmentation. Ecological Applications 13: 1193–2103.


Tatina, R. (2010). Tree composition and demography of a second growth hardwood forest in Berrien County, Michigan. The Michigan Botanist 49: 118–131.

Tyrell, C. D. (2003). Twenty-five years of change in a dry-mesic forest of south eastern Wisconsin. The Michigan Botanist 42: 95–102.

U. S. Climate Data. 2016. Available at unitedstates/usin0624. (Accessed July 12, 2016). USGS. (2010). Level II and IV Ecoregions of Michigan. Available at EPADAtaCommons/ORD/Ecoregion/mi/mi_eco.pdf. (Accessed July 14, 2016). Vankat, J. W., W. H. Blackwell, Jr., and W. E. Hopkins. (1975). The dynamics of Hueston Woods and a review of the successional status of the southern beech-maple forest. Castanea 40: 290–308. Weigert, R. G., and F. C. Evans. (1964). Primary production and the disappearance of dead vegeta- tion on an old field in southeastern Michigan. Ecology 45: 49–51. Wells, J. R., and P. W. Thompson. (1982). Plant communities of the sand dunes region of Berrien County. The Michigan Botanist 21: 3–38 Wilder, C. M., F. W. Holtzclaw, Jr., and E. E. C. Clebesch. (1999). Succession, sapling density and growth in canopy gaps along a topographic gradient in a second growth east Tennessee forest. American Midland Naturalist 142: 201–212.

Woodall, C. W., C. M. Oswalt, J. A. Westfall, C. H. Perry, M. D. Nelson, and A. O. Finley. (2010). Se- lecting tree species for climate change migration hypotheses using forest tree inventory data. For- est Ecology and Management 259: 778–785.

Woods, K. D. (1979). Reciprocal replacement and the maintenance of co-dominance in a beech- maple forest. Oikos 33: 31–39. Yates, E. D., D. F. Lewis, Jr., and C. L. Williams. (2004). Recruitment of three non-native invasive plants into a fragmented forest in southern Illinois. Forest Ecology Management 190: 119–130.

APPENDIX A. Density, basal area per hectare, relative frequency, relative density, relative domi- nance, and importance value for trees (> 10 cm dbh) of each species at Robinson Woods Preserve, derived from measurements taken at 317 random points sampled in 2011–2012.

Density Basal Area Relative Relative Relative Importance Species (trees/ha) per Hectare Frequency Density Dominance Value Quercus rubra 101.109 11.870 22.917 23.428 26.262 72.606 Acer rubrum 114.681 8.991 25.000 26.572 19.892 71.464 Sassafras albidum 76.680 3.190 16.458 17.767 7.057 41.283 Quercus alba 29.179 4.275 7.083 6.761 9.457 23.302 Fagus grandifolia 23.072 8.807 5.417 5.346 19.483 30.246 Liriodendron tulipifera 23.750 4.935 6.458 5.503 10.919 22.880 Acer saccharum 14.929 1.217 2.917 3.459 2.692 9.068 Nyssa sylvatica 12.215 0.260 3.542 2.830 0.576 6.947 Prunus serotina 10.179 0.501 2.917 2.358 1.108 6.383 Fraxinus americana 7.464 0.199 2.083 1.730 0.441 4.254 Ulmus americana 4.750 0.094 1.458 1.101 0.207 2.766 Pinus serotina 2.714 0.377 0.417 0.629 0.835 1.880 Populus deltoides 2.036 0.249 0.625 0.472 0.550 1.647 Carya ovata 2.036 0.060 0.625 0.472 0.133 1.230 Tilia americana 2.036 0.112 0.417 0.472 0.247 1.135 Ostrya virginiana 0.679 0.014 0.417 0.157 0.031 0.605 Amelanchier arborea 0.679 0.009 0.208 0.157 0.020 0.386 Quercus palustris 1.357 0.014 0.417 0.314 0.031 0.762 Populus grandidentata 0.679 0.016 0.208 0.157 0.035 0.400 Cornus florida 0.679 0.009 0.208 0.157 0.019 0.385 Quercus muehlenbergii 0.679 0.007 0.208 0.157 0.015 0.380 Totals 431.58 45.2 100 100 100 300.000


APPENDIX B. Sapling (9, 7, 5, 3, and 1 cm size class midpoints) and seedling densities (stems per hectare) for 32.4 ha of Robinson Woods Preserve (2011–2012) from 317 sampling points.

Size class Quercus rubra Acer rubrum Sassafras albidum Quercus alba 9 cm 13.21 21.07 5.97 5.66 7 cm 12.26 30.50 11.64 6.60 5 cm 21.07 35.22 11.95 12.89 3 cm 19.18 50.31 16.98 23.27 1 cm 82.39 95.28 1023.27 137.74 Seedling 1778.30 3477.00 2654.00 277.30 Fagus Liriodendron Size class grandifolia tulipifera Acer saccharum Nyssa sylvatica 9 cm 11.32 0.94 2.83 5.66 7 cm 25.79 1.26 3.46 5.66 5 cm 73.27 3.77 7.55 14.15 3 cm 263.21 5.03 10.06 21.70 1 cm 1026.10 52.83 23.27 56.29 Seedling 104.30 197.80 69.30 592.80 Size class Prunus serotina Fraxinus americana Ulmus americana Pinus strobus 9 cm 4.72 2.52 4.40 0.00 7 cm 3.46 3.14 3.46 0.00 5 cm 4.72 3.14 10.69 0.00 3 cm 8.49 6.29 10.69 0.00 1 cm 239.62 380.19 13.21 0.31 Seedling 1680.80 547.60 4.50 0.00 Size class Populus deltoides Carya ovata Tilia americana Acer saccharinum 9 cm 0.00 5.66 0.31 0.00 7 cm 0.00 1.89 0.31 0.00 5 cm 0.00 1.57 0.63 0.31 3 cm 0.00 12.58 1.90 0.00 1 cm 0.94 68.87 0.00 0.00 Seedling 0.00 50.20 0.00 0.00 Ostrya Amelanchier Quercus Populus Size class virginiana arborea palustris grandidentata 9 cm 0.63 0.00 1.26 0.00 7 cm 1.57 0.63 2.20 0.00 5 cm 3.14 2.52 4.40 0.00 3 cm 6.29 12.89 0.94 0.00 1 cm 26.42 427.99 5.03 2.52 Seedling 7.60 609.30 7.00 0.00 Prunus Carpinus Hamamelis Size class Cornus florida pensylvanica caroliniana virginiana 9 cm 0.00 3.77 0.94 0.00 7 cm 3.77 0.00 1.57 0.00 5 cm 5.03 0.00 4.09 6.29 3 cm 6.29 1.26 17.30 17.92 1 cm 22.64 24.21 58.81 67.61

Seedling 66.80 44.50 37.50 124.00

Page  17 2016 THE MICHIGAN BOTANIST 17 Size class Asimina triloba Picea pungens Dirca palustris Cornus spp. 9 cm 0.00 0.00 0.00 0.00 7 cm 0.31 0.00 0.00 0.00 5 cm 0.94 0.00 0.00 0.63 3 cm 1.89 0.00 0.00 2.20 1 cm 60.38 0.00 3.14 15.41 Seedling 23.50 0.63 0.00 22.90 Juniperus Size class virginiana Viburnum lentago Crataegus sp. Carya cordiformis 9 cm 0.00 0.00 0.31 0.00 7 cm 0.00 0.00 3.14 0.00 5 cm 0.00 0.00 0.94 0.00 3 cm 0.00 0.31 0.94 0.00 1 cm 0.31 16.67 0.63 0.31 Seedling 0.00 7.60 0.00 0.00