Abstract. Settlement of most frontier lands in the U.S. was preceded by a federal survey. The General Land Office conducted the first survey of San Juan County in 1874 and, in order to secure the location of key survey monuments, it identified over two thousand witness trees by location, species and size. This article utilizes that database of witness trees to reconstruct aspects of that early forest.

     Maps and charts display the spatial distribution, frequency, size, and dominance (a function of frequency and size) of all the major evergreen and hardwood species of forest trees in the San Juans as they were encountered in 1874. The distinctive qualities of San Juan County's forest are further illustrated by comparisons with a forest in nearby Whatcom County. The differences between these pre-settlement forests were due to underlying geologic, soil, and hydrologic factors. The nature of the early forests is relevant to the effort to understand today's forests.
o

GLO database

     During the westward expansion in the 19th century the U.S. General Land Office was charged with documenting and describing frontier lands as a precursor to homesteading and settlement. Across vast expanses of landscape, survey crews systematically marked out the mile-square grid of sections whose identities persist even today in landowners' deeds. GLO surveys were expeditious and fairly accurate in establishing section corners, but in secondary aspects they were superficial (though not necessarily inaccurate). For example, summaries of soil quality and vegetative landcover were "broad-brush" or "coarse-grained;" that is, the assessments were highly general but lacked specificity or detail. They were based on the cursory impressions gained as the surveyors traversed only the section perimeters (section lines), while ignoring the vast interiors.

     One aspect of GLO surveys -- and a distinctively quantitative one -- were their lists of witness trees. These trees helped to locate and identify section corners and points halfway between section corners, the so-called quarter-section points, all of which were physically demarcated with short posts or rock mounds. The locations of witness trees served as backups to pinpoint the monumented reference points in case the they became disturbed. Witness trees were systematically (that is, not randomly) selected: four for each section corner (the nearest one in each compass quadrant) and two for each quarter-section point (the nearest one on each side of the section line). Accordingly, each mile-square section contained eight of such witness trees; if additional trees were intersected by section lines, they were added to the compendium as "line trees." In modern parlance, witness trees were "georeferenced" in the sense that their locations were unambiguously identified with reference to fixed points on the Earth's surface in terms of distance and compass bearing. Witness trees were also identified according to species and diameter at chest height.

     Environmental historians and geographers have extracted witness -tree data (namely species, location and diameter) from early GLO surveys as objective and quantitative evidence concerning early, pre-settlement forests. The same approach is used in this article to characterize the early forests in San Juan County, which were documented by the GLO survey in 1874. Using those data, we can answer some aspects of the question: What were the "original" forests like when, or slightly before, settlers arrived?

     The 1874 GLO Field Notes for San Juan County contain species, diameter and location data for 2074 witness trees from all reaches of the 172 square miles of land area. In the map that follows I display the distributions of 1832 witness trees (line trees could not be accommodated by the display method that I devised). For the sake of contrast and comparison, I have prepared a similar map that displays witness-tree species of an equivalent area of a nearby mainland forest (northwest Whatcom County,) which was surveyed piecemeal between 1859 and 1873. The latter database contains 1642 witness trees, of which 1498 were used to construct the map below.

Species maps

     Even at their small scale the above species distribution maps point to some obvious patterns and geographical differences. The principal colors in the San Juans are pale green (Douglas-firs), light blue (lodgepole pines), and magenta (red alder), whereas in NW Whatcom there are significant blocks of "reddish" species (red alder, bigleaf maple and cottonwood). Although the 19th-century forest of NW Whatcom exhibited many Douglas-firs (pale green) and western redcedars (dark green), they did not generally occur in such large blocks as in the San Juans.

    In general, the same species of trees existed in both geographical areas, although their proportionate representation and distribution differed. The few exceptions of non-overlapping species were all in the minor or "other" category and included: Garry oak, juniper, and Pacific yew, which occurred exclusively in the San Juans; and Pacific dogwood, beaked hazelnut, paper birch, and vine maple, which were restricted to NW Whatcom. Minor species common to both regions included Oregon ash, western cranapple, and bitter cherry.

     Notice that each of the two regional maps can be enlarged for closer examination.

Quantitative charts

     Witness-tree data lend themselves to quantitative analysis in addition to distributional display. The following three charts compare tree parameters species-by-species between the San Juans and Whatcom, and for them the full data sets have been utilized (2074 and 1498 trees, respectively).

     The first chart (relative abundance aka frequency) addresses the question, "What proportion of all trees was represented by each species?" The second chart (average tree diameter) documents reported tree diameters, as averaged by species. The third chart (relative dominance) is derived indirectly (see Appendix 2) from the data underlying the previous two charts; dominance of a given species in a forest addresses the question, "What proportion of a forest's total wood volume is represented by each species?"

     According to the above chart of relative abundance Douglas-firs were encountered in the San Juans in 1874 much more frequently than any other species: above 60% of the time. Of all other species in the San Juans, only alder rose above 10%, although shore pine was also fairly abundant. In NW Whatcom, Douglas-fir, redcedar and alder were roughly equal in abundance at 18-20% each, and a much wider diversity of species was represented at intermediate levels, except shore pine and madrone, and were proportionally better represented there than in the San Juans.

    When average tree diameter is considered, most species in NW Whatcom were equal in size or substantially larger than in the San Juans, except for grand fir (and madrone, which did not occur). To the ordinary observer tree diameter is often the parameter that makes the strongest first impression of "tree size," but of course tree volume is a square function of diameter (assuming a fixed height). That is, a tree that is twice the diameter of another has a volume four times greater. Relatively small diameter differences have large effects upon volume differences.

     Stark differences between the two regional forests show up in the chart of relative dominance. In San Juan County Douglas-firs accounted for the vast majority of "wood volume" (85%), to such a degree that redcedar was the only other species whose dominance rose above 4%. In other words, forests of the San Juans in 1874 were compositionally nearly monocultures in which other species were present but contributed very little volume. In contrast, in NW Whatcom County top dominance was shared by Douglas-fir and redcedar (25-30% each), and alder, hemlock, spruce, and maple were also moderately represented. That is, Whatcom's early forests were more diverse. It may be presumed that many of these distinctions persist in today's forests.

Discussion: Ecological Perspectives

     The two geographical areas compared in this article differ in many respects yet they are separated by less than 25 miles and share overall climates, since they both lie within the Olympic rainshadow that reduces precipitation and increases sunshine. The San Juans, however, are largely rocky islands with shallow and poorly developed soils; their valleys are composed of glacial drift, and there are essentially no year-round streams. Most of NW Whatcom, in contrast, is a low-lying flood plain of the Nooksack River and its tributaries; its soils are alluvial and well hydrated.

     The most successful forest species in the San Juans, in 1874 as well as currently, is the Douglas-fir. This tough species requires direct sunlight, establishes well in rocky terrain, and does not thrive in water-saturated conditions such as poorly drained soils in winter or areas of high water table. Such wet areas are more conducive to red alder, so long as direct sunlight is present, which explains the presence of alder in the lowlands of Lopez Island and in Crow and San Juan Valleys. Pacific madrone and Garry oak are exclusively present in the San Juans because they require dry soil in summer, which the Nooksack floodplain cannot provide.

     In the Nooksack River valley sun-loving red alder, bigleaf maple, black cottonwood and various other hardwood species prosper despite the low summer rainfall because the soils are well watered by the Nooksack. Redcedar and hemlock are similaly comfortable in such a setting; they have the added advantage of being able to germinate in the shade of hardwoods but eventually grow taller. Lowland conditions are genereally too wet for Douglas-fir or lodgepole pine, although these species grow vigorously on the slightly elevated margins of the floodplain.

     Overall, differences in the early forests in the San Juans and Whatcom illustrate that all forests in northwest Washington are not equivalent. Species composition, distribution, and tree size are affected by local differences in environmental conditions.

Appendices

1. How the maps were constructed.

     In an actual forest (left), trees of various species (colored dots) are scattered irregularly across the landscape. The GLO established a gridwork of 1 x 1 mile sections, which were designated numerically; they recorded witness trees exclusively around the perimeter and ignored all internal one (center). Four trees (one per compass quadrant) nearest to a section corner (+) were recorded as witness trees, as were the two nearest trees astride the centerpoints of each section line (+ and -). For the graphic display used in this article, each section was subdivided by a 3-by-3 grid whose nine areas were filled with species-coded colors representing the section's eight witness trees and the section number (right). This display method did not accommodate line trees, which occurred irregularly with rrespect to the grid system.

2. How Relative Dominance was calculated

     A species' relative dominance in a forest is a proxy value for that species' contribution to wood volume. Ideally, it is a function of abundance and tree volume. Strictly speaking, since GLO data did not include tree height, tree volume is impossible to calculate. However, the square of tree diameter (a function of trunk cross-sectional area or basal area) offers an acceptable proxy of volume. For the charts, the relative dominance of a species was calculated as the proportion of that species' aggregate "volume" (actually the aggregate of diameters squared in the present case) relative to the forest's total "volume," expressed as a percentage.

     Therefore, the relevant data bases from the GLO sources were processed as follows: 1) a data base for each geographical region was constructed for all trees and associated diameters; 2) data bases were sorted by species; 3) each diameter was squared; 4) for each species the squared diameters were summed; 5) these sums for all species were also summed to obtain the forest "volume;" 6) the proportion of the total forest "volume" of each species' sum of squared diameters was expressed as a percentage; 7) charts were constructed from these values.

3. Sources

     a. Data for San Juan County. Copies of the GLO Field Notes for San Juan County (1874) are available for reference at San Juan County Department of Public Works in Friday Harbor. I photocopied this lengthy document and entered all witness-tree data into a spreadsheet.

     b. Witness-tree data for NW Whatcom County was obtained indirectly. Prof. David Wallin of Western Washington University generously provided me with an electronic file of a spreadsheet of witness-tree data for the six townships north of Bellingham, Washington. The file had been prepared a few years earlier by two of his students, but it contained some flaws: 1) when extracting tree-species data from the GLO records (which are in longhand script), the term "do." was consistently misinterpreted as signifying Pacific dogwood, when in actuality it was the 19th-century way of writing "ditto;" 2) data from a large block of T40N, R3E, Sections 4-9, 16-21 and 28-29 were unaccountably missing; 3) some outdated species names such as "balm" (for Balm of Gilead, otherwise known as black cottonwood) and "balsam" (otherwise known as grand or white fir) were treated as distinct species rather than pooled; and 4) the species of forty one trees were left undeciphered and formed a class of "unknowns."

     For the misinterpretation of "ditto" I noted that it exclusively occurred after "fir" (otherwise known as Douglas-fir) at section corners, where four witness trees were listed. Accordingly, I interpreted the term "do." to signify Douglas-fir.

     I filled the missing data in the WWU spreadsheet by acquiring species information from a copy of the GLO Field Notes. However, I neglected the diameter data (at the time I only foresaw the species maps, not the quantitative charts). As a result, out of the total database for NW Whatcom of 1642 witness trees some lacked species identifications, 1498 of which were incorporated into the species distribution map (including 185 without diameters as well as all "unknowns," whose species identities were temporarily assigned by their nearest line tree, which otherwise could not have been utilized). The Whatcom charts were constructed from 1457 witness trees whose species and diameters were specified (including line trees, but excluding all "unknowns").

    For this analysis, trees originally identified as "balm" were combined with cottonwoods, and trees identified as "balsam" were combined with grand firs.