Historical Composition and Structure

Archeological evidence indicates that Native Americans were active in the San Juans for thousands of years (Stein 2000), and shell middens confirm their presence on Waldron Island. Native Americans used fire to improve hunting conditions and to promote the growth of plants used for food, fiber, and medicinal purposes (Suttles 1974, Turner 1975, 1999, Boyd 1999a). Until the mid-1800s, Native American burning resulted in fires on Point Disney occurring on average every seven years, although fire frequency varied considerably during this period (Sprenger 2006, Sprenger and Dunwiddie 2011). Fire frequency declined as Native American populations fell by 70–80% with the introduction of smallpox in the region beginning in the 1770s (Boyd 1999b), followed by outbreaks of a malaria-like illness in 1836, measles and tuberculosis in 1847, and smallpox again in 1853–54 (Agee 1987). Our study, together with the reconstructed fire history for Point Disney (Sprenger and Dunwiddie 2011), clarifies how the vegetation composition and structure have changed at this site over the last two centuries.

Garry oak ranged in age from 27–509 yr; 16% (ten trees) were more than two centuries old (Figure 3). Douglas-fir had a similar range, from 17–454 yr; 3% (seven trees) exceeded 200 yr in age. However, heart rot was prevalent on most trees of both species, and substantial estimation was required on several; the oldest ring counts were 308 years on oak, and 281 years on Douglas-fir. Since few of the extant trees established earlier than 1800, we can only speculate about the vegetation structure that existed when the influence of Native Americans on the landscape was more pronounced. However, the fact that individuals of both species are more than 400 years old suggests that the stand had a mixed composition for centuries (Figure 3). This is in contrast to many other studies, which have described Garry oak stands as largely monospecific (Thilenius 1968, Thysell and Carey 2001, Smith 2007), though Day (2005) concluded that a stand in the Willamette National Forest was co-dominated by Garry oak, Douglas-fir, and ponderosa pine (Pinus ponderosa C. Lawson).

Changes in Stand Composition and Structure: Pre Euro-American Settlement

We reconstructed historical stand parameters based on trees that established before 1881 and were still alive at the time of our sampling (Table 1). These generally were also the only trees in the stand that exhibited the spreading form and heavy lower limbs of trees that once grew in open woodland or savanna habitats. The historical density of these older trees was 99 trees/ha, with slightly more oaks than Douglas-firs (Table 1). An unknown number of individuals—presumably Douglasfir, as the oaks had little merchantable value—may have been removed by historical logging. We believe that few trees were removed from the study area, at least in the most recent logging entries, as very few stumps were present. The historical density is almost one-tenth the density of the stand prior to the recent restoration thinning. Comparable data from other oak and oak/ Douglas-fir stands in the Pacific Northwest vary widely. Gedalof et al. (2006) report densities of trees with an open-grown form as ranging from 210 oak trees/ha in historical savannas to 390 trees/ha in woodlands. At the other extreme, Gilligan and Muir (2011) found densities of large, relict oaks averaged about 21 trees/ha in southern Oregon oak stands. The data from the older Point Disney trees fall between these values. The low stand density, together with the open-grown form of most of the older trees, suggest that the stand historically was more akin to a savanna or woodland than a forest. These habitats currently occupy about one-third of Point Disney (Sprenger et al. 2005), but appear to have been considerably more extensive prior to Euro-American settlement. Their open structure was likely maintained by the low-intensity fires that regularly burned across the droughty, south-facing slope.

Figure 3.

Tree age diameter (top) and age (bottom) distributions at Point Disney. Note several classes of Douglasfir thinned in 2007 (clear bars) are truncated.

TABLE 1.

Stand structure (trees taller than breast height) of a Garry oak/Douglas-fir stand at Point Disney, Waldron Island, Washington. Calculations of historical basal area were beyond the scope of this project. Df = Douglas-fir; Go = Garry oak.

Tree establishment on Point Disney has not been continuous, but rather occurred in several clear pulses (Figure 3). This pattern has been observed at many other sites throughout the ecoregion (Agee and Dunwiddie 1984, Agee 1987, Gedalof et al. 2006, Smith 2007, Gilligan and Muir 2011). Most of the extant oaks on Point Disney became established in a 60-year long pulse beginning in the mid-1830s. Exactly which factors triggered this wave of oak establishment, and subsequent pulses of Douglas-fir establishment, are not entirely clear. Three principle mechanisms are frequently cited for stimulating tree establishment in grasslands and savannas: changes in climate (Rochefort and Peterson 1996, Woodward et al. 1995), alterations in grazing regimes (Dunwiddie 1977, Vale 1981), and changes in fire intensity or frequency (Arno and Gruell 1986). These mechanisms may interact with one another and with factors such as seed availability, herbivory, and vegetation competition, resulting in complex patterns of tree establishment (Riege & Del Moral 2004, Dovciak et al. 2008, MacDougall 2008, Haugo et al. 2010). Furthermore, in much of North America, these changes occurred simultaneously with declines in Native American populations, and with Euro-American settlement and the associated suppression of native-set fires, introduction of livestock, and hunting of native herbivores (MacDougall 2008). Therefore, it is difficult to disentangle the separate effects of these interacting mechanisms.

At Point Disney, the onset of the oak establishment period coincides with one of the most widespread fires that occurred on the site (1834) and continued through a period when fires burned with regularity (Sprenger and Dunwiddie 2011). It is unclear whether these oaks regenerated from seed or resprouted from root sprouts on older genets. However, many were single-trunked trees, suggesting a seed rather than resprout origin. Douglas-fir also established during this period, though the numbers of extant individuals are lower than for Garry oak, possibly because conditions were more favorable to the establishment of oaks or because of later selective logging of Douglas-fir.

Several other studies in the region describe similar waves of tree establishment in the early-mid 1800s. At Rocky Point on Vancouver Island, ca. 50 km southwest of Waldron Island (Gedalof et al. 2006), Garry oak and, to a lesser extent, Douglas-fir, invaded an open prairie in an 80-year pulse beginning in the 1850s, two decades after establishment began on Point Disney. At Young Hill on San Juan Island, 13 km southwest, Agee (1987) documented a period of oak regeneration from 1830–60 into an oak/Douglas-fir woodland. In an oak-dominated stand on Blakely Island, 22 km southeast, Batiuk (personal communication 2009) reported that 83% of the oaks regenerated between 1835 and 1900, with the main pulse occurring from 1850–1895. In southern Oregon, considerable Garry oak regeneration began at multiple sites starting in the 1830s, with peak establishment generally occurring between 1850–1890, and ceasing with increasing suppression of fires (Gilligan and Muir 2011). Both Gedalof et al. (2006) and Agee (1987) attributed the pulse of tree establishment during this time period to the cessation of Native American burning. Thilenius (1968) also implicated fire suppression as resulting in the infilling of Garry oak savannas in Oregon over the last century. Smith (2007) concluded that both fire and disturbance associated with Euro-American settlement, rather than climate, contributed to patterns of tree establishment that began in the latter 1800s.

Changes in Stand Composition and Structure: Post Euro-American Settlement

As in other regions, dramatic reductions in fire frequency followed settlement in the Pacific Northwest (Agee 1993, Boyd 1999a, Turner 1999). At Rocky Point, tree encroachment coincided with Euro-American settlement, the initiation of grazing, and the cessation of frequent fires (Gedalof et al. 2006). At eight sites in the Gulf Islands and Vancouver Island, Smith (2007) also concluded that a pulse of tree recruitment typically occurred at or shortly after Euro-American settlement. The scenarios at Point Disney and Young Hill are more complex as establishment began several decades before Euro-American settlement, although Native American populations had already been declining. The 1834 fire on Point Disney occurred after a relatively lengthy firefree period, and may have stimulated establishment. The end of the pulse occurred several decades after Euro- American settlement and roughly coincides with the commencement of livestock grazing on Point Disney, and may reflect oaks' low tolerance of trampling and livestock herbivory (Jackson et al. 1998). Gilligan and Muir (2011) also found little regeneration of Garry oak during the 1900s in southern Oregon.

The wave of Garry oak establishment on Point Disney ended as a century-long period of Douglas-fir regeneration began (Figure 3). Extensive infilling of Douglas-fir during the 20^th century has been reported from many other sites in the region (Agee and Dunwiddie 1984, Agee 1987, Peterson and Hammer 2001, Spurbeck and Keenum 2003, Gedalof et al. 2006, Smith 2007). At Point Disney, this occurred in three increasingly larger waves: 1882–1896, 1937–1951, and 1962–1976. Only a few more fires were documented on the site once this period began, with the last fire occurring in 1906 (Sprenger and Dunwiddie 2011). Young Douglas-fir trees are readily killed by burning, and once fires were removed, significant infilling by this species converted much of the savannas and woodlands to forest. Increasing conifer density in the 20^th century precluded further establishment of oaks, which can germinate but do not establish successfully under a Douglas-fir canopy (Thysell and Carey 2001). During this period, the primary disturbances on the site were livestock grazing and, to some extent, selective logging from the 1940s to the early 1970s. However, we have insufficient detail on the nature of these events to closely evaluate the causal relationships with the observed pulses in tree infilling, with the exception of the most recent wave. An enormous and rapid pulse of Douglas-fir regeneration began ca. 1962, with thousands of young trees forming dense stands in the forest understory and in many savanna and woodland openings (Figure 3). This last wave was particularly concentrated from 1972–1976, and was exceptionally large; it accounted for 73% of the trees in the pre-restoration stand.

This recent wave of Douglas-fir establishment was previously documented on Point Disney by Salstrom (1989), and closely resembles similar pulses that have been reported at several nearby locations. We have observed similarly-aged young cohorts of Douglas-fir that have colonized grassy balds, bluffs, and headlands at numerous other locations in the San Juans, including Sentinel Island, multiple sites along the south shore of Lopez Island, on the slopes of Mt. Constitution on Oreas Island, and American Camp on San Juan Island, to name but a few. Agee and Dunwiddie (1984) described a wave of Douglas-fir regeneration into open grasslands on Yellow Island, 9 km south of Point Disney. The oldest tree in this cohort germinated in 1963; the invasion peaked in the early 1970s but continued into the 1980s, when managers began to remove trees from the grasslands on the island. Increased summer moisture was suggested as a probable factor that may have helped initiate this event (Agee and Dunwiddie 1984). Agee (1987) documented a similar period of recent tree establishment on Young Hill on San Juan Island. Douglas-fir was also the dominant species in this recruitment pulse, which began about 1965. Regeneration virtually ceased in the late 1970s in plots with an oak/Douglas-fir overstory, but continued in grasslands into the 1980s, particularly in a plot that burned in 1973. Agee (1987) found that favorable climatic conditions and good seed years were only weakly correlated to years of peak conifer recruitment, and attributed its initiation largely to the cessation of grazing around 1966. On Rocky Point, Gedalof et al. (2006) described a large wave of Douglas-fir establishment in the 1960s and 70s. They concluded this recruitment was “probably ongoing,” but were uncertain what circumstances initiated it. It is noteworthy that at all of these sites where sufficient data exist, this recent recruitment event was by far the largest of any in well over a century.

Fire is not mentioned as a likely triggering factor for these recent events at any of the sites, with the exception of a small portion of Young Hill (Agee 1987). Although it has been suggested that changes in the abundance of native ungulates may contribute to tree encroachment events in some areas (Peterson and Hammer 2001, MacDougall 2008), the absence of such grazers on Waldron today is noteworthy. As on Young Hill, the beginning of the most recent conifer pulse on Point Disney closely coincided with the cessation of livestock grazing. Selective logging prior to the 1960s may have played a role on the Waldron site as well. Similar pulses of tree establishment occurring in response to release from grazing pressure have been reported in other grassland systems (Dunwiddie 1977, Vale 1981, Bakker and Moore 2007, Beschta and Ripple 2009). But, livestock grazing was never a factor on Yellow Island, which also experienced a simultaneous episode of Douglas-fir recruitment. Similarly, Gedalof et al. (2006) reported no changes in grazing at Rocky Point in the 1960s.

The widespread occurrence of similar events that are relatively simultaneous and almost unprecedented in magnitude across the region is striking. That this has occurred even in sites with no history of livestock grazing, such as Yellow Island, points toward causal factors that have common, widespread impacts across the region. Agee and Dunwiddie (1984) suggested an anomalous series of moister years occurred in the 1960s that were particularly favorable for the establishment of trees in xeric grassland habitats. Invasion continued to occur, even after drier conditions resumed, because the recently established trees ameliorated the microenvironment and created conditions that favored additional establishment (Halpern et al. 2010). More detailed analyses by Agee (1987) detected only a weak relationship between moist summers and with years of tree establishment. More detailed comparisons of climatic records, land use history, and tree recruitment data are needed at sites across the region to resolve these uncertainties. On Point Disney, we suggest the recent large wave of Douglas-fir recruitment most likely resulted from a combination of interacting factors, which included the cessation of livestock grazing, climatic factors, and perhaps logging disturbance.

By the early 1980s, this period of conifer invasion had virtually ceased. The apparent pulse of oak establishment in 1962–1966 and 1972–1976 (Figure 3) is an artifact of a single plot center situated in an opening amongst a few small oaks. Even though oaks are regenerating on Point Disney, reproduction occurs almost entirely in more open areas rather than under the conifer canopy. We recorded no tree regeneration of oak or Douglas-fir since 1987. Although this is partially a result of our sampling trees taller than breast height, it also reflects a real lack of establishment; less than a half dozen saplings were encountered during our sampling. However, oaks are regenerating well in more open areas of Point Disney, presumably where competition with Douglas-fir is less. This is in noted contrast to many other sites we have observed in the San Juans, where extensive deer browse has prevented any oaks from reaching sapling size.

Oak Release Activities

By the early 2000s, the historically open vegetation on Point Disney had been replaced with dense forests: the stand contained 927 trees per ha, 87% of which were Douglas-fir, before the oak release treatment was applied (Table 1). This density is nearly double the number reported in conifer forests on Oreas Island (Peterson and Hammer 2001). Basal area was 50.6 m² per ha, of which 78% was Douglas-fir. Most trees were in the 5–10 cm and, to a lesser extent, 10–15 cm dbh classes (Figure 3).

Douglas-fir is able to establish under a light-moderate oak canopy, and once established, grow significantly more rapidly than Garry oak at Point Disney. As a result, Garry oaks were older than Douglas-firs of the same size (Figure 4). This rapid growth, together with the large wave of conifer recruitment over the last century, explains the overtopping and crowding that is contributing to the decline and mortality of many of the oaks.

The extensive degradation and loss of oak-dominated communities throughout the Puget Lowland and Willamette Valley adds urgency to efforts to restore these endangered systems (Hanna and Dunn 1996, Vesely and Tucker 2004, Vellend et al. 2008). At many sites, oaks are witnessing considerable canopy die-back, alteration in form, and high rates of mortality due to competition for resources by fast-growing firs. On Point Disney, oak establishment has been largely unsuccessful for more than 100 years. However, there are virtually no examples of dry, lowland Garry oak/ Douglas-fir systems that continue to be maintained by frequent fire in the Pacific Northwest, and the remnant oak stands we have are not representative of the full range of its distribution historically (Vellend et al. 2008). We therefore do not have good examples of reference conditions to guide the restoration of stand structure in these communities.

Restoring communities that have been significantly altered for extended periods presents large challenges. However, mechanical removal of competing conifers is a feasible and effective approach for ensuring survival of canopy oaks (Vesely and Tucker 2004, Devine & Harrington 2006, Harrington & Devine 2006). Beginning in 2006, large numbers of Douglas-fir that were less than 20 cm dbh, and select trees less than 30 cm dbh within one tree height of targeted oaks, were cut from the stand. A handful of individuals that could not be easily removed without damaging oaks were girdled. Overall, 55% of the trees were killed; the remote location of this site meant that all cut trees were burned on site. Since attention focused primarily on small diameter Douglas-fir (Figure 3), most of the cut trees were from the cohort established after 1970 (Figure 3), and only 10% of the overall basal area was cut (Table 1). Upon completion, Douglas-fir still accounted for 72% of the trees and 76% of the basal area.

The restoration work carried out on the site so far appears to have succeeded in accomplishing its primary objective of maintaining the mature oaks. We have not observed any increased mortality or significant wind damage to the oaks since their release. In addition, epicormic sprouting is beginning to reinvigorate the oaks with increased foliar cover. However, the post-treatment composition, density, and basal area data all differ significantly from historical values. Douglas-fir density in all diameter classes remains high, and basal area has been only slightly reduced. Although some smaller trees can still be felled and burned, removal of larger-diameter trees in numbers sufficient to return the stand to a woodland or savanna structure poses significant logistical challenges. Our best options may be to create patches and openings within the stand that will allow restoration of an understory more characteristic of an open canopy. However, the understory in the savannas and woodlands that still persist on Point Disney is currently dominated in many areas by invasive, non-native species. Experimental treatments are currently underway to identify the most effective strategies for controlling these invasives, as well as for re-establishing herbaceous grasses and forbs in areas where the forest canopy has been opened and where slash piles have been burned.

Figure 4.

Relationships between DBH and age for Garry oak (top) and Douglas-fir (bottom). Equations are based only on those samples that hit the pith or where the number of missing rings was estimated using Applequist's method (shaded symbols). Samples whose ages were estimated with the early radial growth method are shown as open symbols (N = 20 for Garry oak, 22 for Douglas-fir). For reference, each graph also contains as a lighter line the best fitting line from the other species. Axes are on the same scale in both graphs.

Numerous questions remain regarding the consequences of the restoration actions taken to date, and determining what future actions are most feasible. Have sufficient canopy gaps been created to ensure oak seedling survival at a rate sufficient to maintain the stand? Should competition from more of the larger Douglas-fir trees be removed, perhaps by girdling? How will invasive species respond to the increased light levels in the understory? How can significant accumulations of duff, needle litter, and woody debris be reduced to create conditions more suitable for supporting native herbaceous species? What is the desired stand density and composition? Has the removal of small-diameter Douglas-fir created conditions that will trigger a new wave of fir regeneration? What type of management should be initiated to maintain this stand as a mixed-species stand? Can fire be returned to the site as a significant ecological process? Future efforts will focus on answering these questions, and on developing strategies that will be most successful at enhancing the ecological attributes and long-term viability of this endangered system. Additional work is also needed to restore the substantial acreage of oak/ Douglas-fir habitats that exists nearby (Ellison 2010).