Paleoethnobotanical inquiry of early euro-American and Ojibwa Gardens on Grand Island, Michigan

Northeastern Naturalist, 1998 by Silbernagel, Janet, Martin, Susan R, Landon, David B, Gale, Margaret R

Both the pollen and phytolith remains supplement macroremains but are less direct in their use for determining cultural-plant interactions for several reasons. Taxa identified to family level or above, and unidentified or fragmented pollen remains were removed from the data set to facilitate quantitative analysis. But in doing so, the size of the pollen data set was reduced. Many taxa represented by pollen remains were not culturally important. The number of CI remains from WL1-F was influenced by many Betula grains. This can be partly attributed to the plant and fruit forms typically represented by pollen remains compared to macroremains. We tend (as this research supports) to find more edible fruit and nut bearing plants represented by macroremains than by pollen. However, at least four taxa were found in the pollen sum that are potential food plants but were not found in the historical accounts: Carya (hickory/ pecan), Castanea (chestnut), Alnus (alder), and Corylus (hazlenut).

Pollen results suggested a somewhat forested overstory around WL1-F, and an open, herbaceous vegetative community at FC (Table 2). The palynomorph taxa found were largely represented in the current flora except for the genera Carya, Castanea, and Carpinus. We found no pollen representation of F. grandifolia in the samples analyzed, a species now abundant on Grand Island. Beech is a relatively recent arrival to the area (500-3000 BP), and Grand Island is at the edge of its present northwest range (Davis 1978). The lack of F. grandifolia pollen in the sample may indicate the sample profiles predate the arrival of beech to Grand Island, or more likely, that beech pollen is not as durable or prevalent as pollen of other species (Pearsall 1989).

Over-and under-representation of archaeological pollen must be considered in interpretation of results. Wind-pollinated taxa contribute differentially to the pollen rain of a region. Generally, in forested areas, the overstory is more represented in pollen rain than are understory plants (Pearsall 1989). Overstory trees with light, buoyant grains such as pine, which are transported great distances, are typically over-represented in the pollen sum. Our pollen data were dominated by a few taxa, especially among trees and shrubs. The potential for over-representation by these taxa (Betula, Ostyra or Carpinus, Pinus, Quercus, and Salix) should be considered when comparing the counts and percentages. Also because many pollen grains are wind-transported, microvariation is likely to be less apparent than with macroremains.

We cannot tell how many pollen grains are ancient vs. contemporary. However, our intent was to sample from archaeological contexts (WLl, WL1-F, MBI, MB2), thereby uncovering plant remains consistent with the archaeological period. These findings were then compared via our CI:CP ratio to assess similarities with modern floral communities.

Phytolith data seem to complement data obtained from pollen and macroremains and strengthen interpretations of site variation. Because soil phytolith analysis is a relatively new paleobotanical technique, analytical procedures and classification have not attained the maturity of palynology. Application of phytolith analysis to identification of New World crops dates to the 1960s, with a dramatic increase in the 1970s and 1980s. A critical need in phytolith research is a database of plant phytolith types (Pearsall 1989). Like other botanical remains, plants are not equally represented in phytolith identifications. Many plants do not deposit silica, or very little. Not all phytoliths are preserved equally well in soil. Similar silica bodies may be produced by very different plant groups, reiterating the need for phytolith taxonomic research. Nevertheless, analysis of Grand Island phytoliths has revealed some interesting phytolith forms and assemblages. For instance, phytolith forms suggested forest cover with sparse Pooid grass on WLl and greater Pooid presence, or more open sites for MB1 and MB2. Samples WL1-F and FC contained less than 50% grass silica bodies but had two rondels, which have also been cited as indicators of Pooid grasses. Saddles occurred in WL1 and WL1-F and are generally indicative of Chloridoids. Rectangle ! square forms, reflecting an unknown plant contributor, occurred in a much larger proportion in the WL1-F sample than any of the others Sample FC, taken from a disturbed upland, demonstrated greater phytolith deposition from inflorescences than the other samples.