Skip common site navigation and headers
United States Environmental Protection Agency
Great Lakes Ecosystems
Begin Hierarchical Links EPA Home > Great Lakes EcosystemsUpland Ecosystems > 1994 Oak Savanna Conferences > J. Drake
Aquatic Ecosystems
EPA Region 5 Critical Ecosystems
Ecosystem Funding
Great Lakes Basin Ecosystem
Great Lakes Biological Diversity
Green Landscaping
Rivers and Streams
Upland Ecosystem

1994 Proceedings
North American Conference on Savannas and Barrens


J. Drake
Conservation Biology Program
University of Minnesota
St. Paul, MN.

D. Faber-Langendoen, Regional Ecologist
Midwest Regional Office
The Nature Conservancy
Minneapolis, MN

Living in the Edge: 1994 Midwest Oak Savanna Conferences


The pine barrens of Wisconsin were described by Curtis (1959) as a community distinct from forests or other barrens but the type was described broadly in part because of lack of information (Curtis 1959, p.327). Curtis also used only presence-absence data from 26 stands located in central and northern Wisconsin. Brown (1950) and Vogl (1974) also studied the pine barrens but their analyses were limited to parts of Wisconsin; they did not consider the state as a whole. As part of a regional classification of plant communities developed by The Nature Conservancy (Faber-Langendoen 1993), all of Curtis' wooded stands were re-analyzed. Fifty-four stands were separated based on these analyses and investigated in more depth. The relationship between floristics and structure of these stands was examined. These pine barrens stands were also compared to communities in southern Wisconsin to determine their similarity.


Part of the data, gathered primarily by J. T. Curtis and his students in the 1940s and 1950s using the point-quarter method, were stored at the Wisconsin Plant Ecology Laboratory (PEL). Data from a second set of 40 stands were collected but not used by Curtis. They were made available in 1990 in electronic form by Charles Umbanhowar (Umbanhowar 1990; data available from author upon request). Trees were defined as woody stems with a diameter of 10 cm or more at a height of at 1.47 m (Curtis 1959, p.75). Quantitative groundlayer data were subjected to multivariate analytical techniques to arrive at floristically similar groups. The methods included two ordination methods, detrended correspondence analysis (DCA) (Ter Braak 1986) and non-metric multidimensional scaling (MDS) (Minchen 1990), and two clustering methods, two-way indicator species analysis (TWINSPAN) (Hill 1979) and unweighted pair group method using arithmetic averages (UPGMA) (McKune 1991). Clusters derived from the latter two methods were overlaid on the ordinations to help define community types. Final assignment of stands to community types was made by identifying congruent patterns of tree overstory and groundlayer composition; to a lesser extent, structural data and geographic location were used.


All of the upland forest and barrens stands from northern Wisconsin were first analyzed to determine major divisions between communities. The primary division was between the more open stands, dominated chiefly by Pinus banksiana (jack pine) or jack pine-oak, and all other forest types. There were 54 stands in this group.

These 54 stands were grouped into 6 communities (Fig. 1). The communities were Jack Pine-Oak Sparse Woodland (savanna) (8 stands), Jack Pine-Oak Woodland (11 stands), Jack Pine Forest (14 stands), Quercus ellipsoidalis (pin oak) Forest (8 stands), Populus spp. (aspen) Forest (3 stands), and Mixed Pine-Oak Forest (10 stands). Preliminary analyses showed that the Mixed Pine-Oak Forest was distinct from the other communities. These 10 forest stands are therefore not included in the final analyses of the Pine Barrens. Of the remaining 5 Woodland and Sparse Woodland communities, jack pine was the leading dominant in the overstories of 3, and pin oak was the leading dominant in 1. Pin oak was the second leading dominant in 2 other types. Jack pine, Pinus strobus (white pine), pin oak, and Populus grandidentata (big-tooth aspen) were the only tree species that occurred in all 5 of the communities (Table 1). Several species of herbaceous plants and woody shrubs typical of dry, open northern areas were common in all of the types. Among these were Vaccinium angustifolium (blueberry), Gaylussacia baccata (huckleberry), Gaultheria procumbens (wintergreen), Pteridium aquiline (bracken fern), Lysimachia quadrifolia (whorled loosestrife), and Maianthemum canadensis (Canada mayflower) (Table 2).

Only one tree species was 100% constant in any of the types. Jack pine was in all 11 stands of the Jack Pine-Oak Woodland. No other tree species was in all of the stands of any one community. The relative basal area (RBA) contributed by any single species was variable. In the Jack Pine-Oak Sparse Woodland, the RBA of jack pine and pin oak averaged 41.5% and 35.0%, respectively, but in individual stands it ranged from 0% to almost 100% for each species. This trend towards diversity in the canopy layer composition was present in all of the communities.

The structural variation within and between communities was similar to the compositional variation. The communities' structural characteristics were as follows:

  1. Jack Pine-Oak Sparse Woodland-basal area (BA) 14.1 m2/ha, 512 stems/ha, canopy cover (cover) 34%.
  2. Jack Pine-Oak Woodland-BA 15.6 m2/ha, 638 stems/ha, cover 34%.
  3. Jack Pine Forest-BA 25.5 m2/ha, 730 stems/ha, cover 54%.
  4. Pin Oak Forest-BA 19.0 m2/ha, 626 stems/ha, cover 60%.
  5. Aspen Forest-BA 24.7 m2/ha, 1329 stems/ha, cover 59%.

All of the communities had wide ranges in density, basal area, and canopy cover values. In two of the communities, there were single stands whose presence greatly increased the averages of these values. By eliminating one stand of the Jack Pine-Oak Sparse Woodland when figuring the above averages the results were BA 12.5 m2/ha, 455 stems/ha, 29% canopy cover. Following the same procedure for the Jack Pine Forest changed the averages to 24.4 m2/ma, 681 stems/ma, and 52% canopy cover.

The Jack Pine-Oak Sparse Woodland and Jack Pine-Oak Woodland were similar to each other, as their names suggest. They shared a canopy composition in which greater than 75% of the RBA was contributed by jack pine and pin oak with approximately 10% of the RBA from Quercus velutina (black oak). Many understory species were common to both communities, although the frequency and constancy differed markedly in some cases. There were stands in each that were extremely similar to stands in the other community, as similar as they were to stands in their assigned group. Therefore, the separation of these stands was somewhat arbitrary. However, the difference between the communities was significant. They represented two points along a gradient from open to closed canopy communities.

As the ordination shows (Fig. 1), the Jack Pine Forest, Aspen Forest, and Pin Oak Forest had understories of similar composition. They were not clearly separated by any of the multivariate analytical techniques, although the overstory composition was substantially different. When separated on the basis of overstory composition it was found that the resulting groups were weakly clustered by multivariate analysis of the understory data. The Jack Pine Forest was dominated by conifers. Jack pine made up 55% of the RBA and Pinus resinosa (red pine) and white pine combined were 26% of the RBA. The Aspen Forest and Pin Oak Forest were dominated by deciduous trees. Less than 12% of the RBA of either of these communities was due to conifers. Big-tooth aspen and Populus tremuloides (quaking aspen) made up 80% of the RBA of the Aspen Forest. Pin oak made up 58% of the RBA of the Pin Oak Forest. Quercus rubra (red oak) and Q. alba (white oak) added another 20%.

To test whether other communities in northern upland forests might be similar to the pine barrens, further comparisons were made. Neither the southern oak barrens and woodlands nor the bracken grasslands proved to be similar using multivariate analyses. In all of the methods used, the first division was between the stands in the pine barrens and those in the other communities.


Thirty-nine of the 44 stands were located on the central sand plains of Wisconsin. They were within an imaginary rectangle approximately 40 km by 90 km with the long axis running NW-SE. The soils of this area are derived from glacio-lacustrine or outwash sands, and there was some tendency for communities to occur on soils of different origin. The Jack Pine forest had 10 stands on sands derived from glacial outwash or pitted outwash, while 4 stands were on glacio-lacustrine sands. The Jack Pine-Oak Woodland had 10 stands on glacio-lacustrine sands and only 1 on outwash sands. The Pin Oak Forest stands were distributed with 6 on glacio-lacustrine sands and 2 on outwash stands. The other 2 communities were more evenly spread between the 2 types of sands.

There was a floristic gradient from Jack-Pine-Oak Sparse Woodland through Jack Pine-Oak Woodland to Jack Pine Forest, Pin Oak Forest, and Aspen Forest. This gradient appears to reflect a progression from relatively open to relatively closed overstory canopy. The stands used by PEL researchers had greater basal areas and densities than would be expected for dry, open communities. Brown (1950) also found extremely high densities in his work on central Wisconsin pine barrens. He found that areas that had 5-20 trees/ha in 1850 had 450-625 trees/ha in 1950. Formerly open stands were closing after effective fire suppression began in the early 1900s. The groundlayer in Brown's studies and those of PEL workers reflected a more open condition with many shade intolerant grasses, herbs, and shrubs (Table 2). However, it would appear that the historic pine barrens condition has not been properly sampled yet.

Multivariate analyses confirmed the distinctiveness of a pine barrens community or suite of communities. We believe that Curtis' naming of only one pine barrens type was too coarse. Vogl (1974) distinguished 2 pine barrens communities in northern Wisconsin. One had a groundlayer with many prairie species and one had more shrubby species. Vogl did not consider the central sand plains in his study. The Jack-Pine-Oak Sparse Woodland described in this analysis is similar to Vogl's pine barrens with a prairie understory. We found no sparse woodland community with a non-prairie understory.

The multivariate methods employed in this study allowed comparison of a large number of stands from across Wisconsin. The methods identified groups of stands with similar floristic features. Final placement into community types was also based on overstory and structural characteristics. It is hoped that by delimiting communities in this way the vegetation can be viewed within geographic and environmental gradients and be viewed in a more dynamic perspective.


Partial support for the senior author was provided by the Environmental Protection Agency's Wisconsin Site Project. We thank Lance Perry with the Wisconsin Department of Natural Resources for preparing a stand location map.


Ambrose, D., J. Drake, and D. Faber-Langendoen. 1994. Rare Plant Communities of the United States: Midwestern Region. In D. H. Grossman, K. L. Goodin, and C. L. Reuss, editors. Rare Plant Communities of the Conterminous United States: An Initial Survey. The Nature Conservancy, Arlington, VA.

Brown, R. T. 1950. Forests of the Central Wisconsin Sand Plains. Bulletin of the Ecological Society of America 22:217-234.

Curtis, J. T. 1959. The Vegetation of Wisconsin: An Ordination of Plant Communities. University of Wisconsin Press, Madison.

Faber-Langendoen, D. (editor). 1993. Midwest Regional Community Classification-Preliminary Draft. Midwest Heritage Task Force, Minneapolis, MN.

Hill, M. O. 1979. TWINSPAN-A FORTRAN Program for Arranging Multivariate Data in an Ordered Two-way Table by Classification of the Individuals and Attributes. Cornell University, Ithaca, NY.

McKune, B. 1991. Multivariate Analysis on the PC-ORD System. Department of General Science, Oregon State University, Corvallis, OR.

Minchin, P. R. 1990. DECODA-Database for Ecological Community Data, vers. 2.04. University of Melbourne, Parkville, Victoria, Australia.

Ter Braak, C. J. F. 1986. CANOCO-A FORTRAN Program for Canonical Community Ordination by [Partial] [Detrended] [Canonical] Correspondence Analysis, Principal Components Analysis, and Redundancy Analysis, vers 2.1. TNO Institute of Applied Computer Science, Wageningen, The Netherlands.

Umbanhowar, C. E., Jr. 1990. Regional Lists of Prairie Herbs and Forest Trees for Highway Rights-of-way Restoration and Management: Extracted from the Wisconsin Plant Ecology Laboratory Files. Final Report, Project No. 0072- 07-10, to Wisconsin Department of Transportation.

Vogl, R. J. 1974. Fire and the Northern Wisconsin Pine Barrens. Proceedings of the Tall Timbers Fire Ecology Conference 10:175-209.

Figure 1. Detrended Correspondence Analysis of 54 Pine Barrens stands. Meanings of symbols are: Closed rectangle, Mixed Pine-Mixed Oak Forest; open circles, Jack Pine-Oak Woodland; Closed square, Jack Pine-Oak Sparse Woodland; open triangles pointing down; Jack Pine Forest; closed triangle, Pin Oak Forest; open triangle pointing up, Aspen Forest.

Drake1.gif (18803 bytes)







Table 1. Overstory components of Pine Barrens communities. RBA = Relative basal area. RFre = Relative frequency (percentage of stands in each community in which a species occurs). Meanings of acronyms are: JPW-OS = Jack Pine-Oak Sparse Woodland; JPW-OW = Jack Pine-Oak Woodland; JPF = Jack Pine Forest, AspenF = Aspen Forest; PinOakF = Pin Oak Forest.

Species JPW-OS JPW-OW JPF AspenF PinOakF
Pinus banksiana 42 88 60 100 55 93 6 33 6 63
Pinus resinosa 1 25 3 36 14 79 0 0 3 50
Pinus strobus 6 25 0 18 11 57 3 33 2 50
Populus grandidentata 1 25 0 27 1 43 45 76 1 25
Populus tremuloides 0 0 1 18 3 50 37 67 4 50
Quercus alba 3 63 3 46 0 21 0 0 9 63
Quercus ellipsoidalis 35 75 21 73 9 57 8 67 58 75
Quercus velutina 14 25 10 36 2 7 0 0 0 0


Table 2. Understory components of Pine Barrens communities. Freq = Frequency. RFre = Relative Frequency (percentage of stands in each community in which a species occurs). Meanings of acronyms are: JPW-OS = Jack Pine-Oak Sparse Woodland; JP-OW = Jack Pine-Oak Woodland; JPF = Jack Pine Forest, AspenF = Aspen Forest; PinOakF = Pin Oak Forest.


Freq RFre Freq RFre Freq RFre Freq RFre Freq RFre
Andropogon gerardii 31 88 3 27 0 7 0 0 0 0
Antennaria spp. 20 100 6 36 1 29 0 0 3 25
Aster macrophyllus 1 25 3 27 17 79 5 67 19 88
Comtonia peregrina 11 63 12 73 7 50 2 33 8 38
Corylus cornuda 10 50 11 45 12 64 4 100 10 63
Gaultheria procumbens 7 63 19 100 70 100 62 100 57 100
Gaylussacia baccata 7 63 23 91 27 57 47 67 41 100
41 100 10 82 0 0 0 0 4 38
Krigia beflora 18 75 5 64 0 14 7 67 6 38
Lysimachia quadrifolia 27 75 11 64 10 64 37 100 24 88
Maianthemum canadense 8 75 12 64 60 93 18 33 23 88
Pteridium aquilinum 12 63 15 82 62 100 65 100 57 100
Schizachyrium scoparium 13 50 1 18 0 0 0 0 0 0
Vaccimium angustifolium 44 100 54 100 71 100 63 100 67 100



Begin Site Footer

EPA Home | Privacy and Security Notice | Contact Us