Monitoring Prairies in San Juan Island National Historical Park to Protect Natural and Cultural Resources
Regina M. Rochefort, John R. Boetsch, Natalya Antonova, Shay Howlin, Jason Mitchell, Theodore Owen
Please cite this publication as:
Rochefort, R.M., J.R. Boetsch, N. Antonova, S. Howlin, J. Mitchell, and T. Owen. 2026. Monitoring Prairies in San Juan Island National Historical Park to Protect Natural and Cultural Resources. Science Report NPS/SR—2026/437. National Park Service, Fort Collins, Colorado. https://doi.org/10.36967/2318382
Abstract
San Juan Island National Historical Park (SAJH) protects some of the last remaining prairies in Washington state. Over 60% of the American Camp unit is dominated by grasslands that have persisted since the end of the glacier era, approximately 11,700 years ago. In 2007, the National Park Service initiated prairie monitoring at SAJH to document status and trends in cover and ecological condition of these vegetation communities using repeatable, rapid assessment of physiognomic class and vegetation quality along line-intercept belt transects. Sampling was designed to address a hierarchy of monitoring questions linked to management objectives, including the maintenance of broad landscape structure (e.g., forest versus herbaceous) associated with the cultural landscape and the continued ecological integrity of plant communities. We found that while physiognomic cover was similar today to the historic reference period (36% forest:61% non-forest), the ecological condition of grasslands or prairies was threatened by introduced grasses and shrubs. Additionally, between 2007 and 2017, we found significant declining trends in the proportion of area with herbaceous types with an estimated trend of −0.58% per year. The estimated trend in high-quality native herbaceous vegetation also decreased at a rate of −0.08% per year, but the trend was not significant. Transects proved to be an economical and reliable method for monitoring long-term trends in landscape cover with transient crews, but detailed vegetation monitoring (i.e., species frequency and relative abundance) is needed to better assess prairie condition over time.
NPS
Table of Contents
Acknowledgments
Data collection and analysis was funded by the National Park Service through the North Coast and Cascades Inventory and Monitoring Network. We thank Mignonne Bivin, Steve Hahn, Steve Acker, Jenny Shrum, Jerald Weaver, Catherine Copass, Carla Wise, Laurie Kurth, Michelle Toshack, Rebecca Evans, and Mark Huff for frequent assistance with surveying transects. We also thank Lise Grace for field and GIS assistance, and Steve Gibbons, Sara Dolan, Gilbert Moreno, and Beth Fallon for reviewing earlier versions of this manuscript.
NPS
Introduction
Background
Grasslands are iconic landscapes in San Juan Island National Historical Park (SAJH); their open vistas provide perceptible reminders of past geologic events; human activities; current habitats for native, threatened, and endangered species; and windows to future management challenges (Figure 1). The Park was established as a National Historical Landmark in 1961 and a National Park in 1966 to “interpret and preserve the sites of the American and English camps on the island and commemorate the historic events that occurred there from 1853 to 1871 in connection with the final settlement of the Oregon Territory boundary dispute…” (Avery 2004; Cannon 1997). The conflict escalated with the Pig War in 1859 when an American, Lyman Cutler, shot a pig owned by the Hudson’s Bay Company in the area of the park now known as American Camp (Avery 2004). In 1853, the Hudson’s Bay Company established Bellevue Farm on the widespread native prairies, converting many areas to managed grasslands by introducing non-native species to supply forage for sheep, cattle, and establishing extensive agricultural fields (Agee 1984, Rolph and Agee 1993).
NPS
More recently, volatile populations of non-native rabbits (Oryctolagus cuniculus L.) have presented a significant, recurring disturbance across the prairie landscape. Introduced on San Juan Island in the 1880s for hunting, rabbits present a significant challenge to future prairie management and restoration of native prairies (Avery 2004). In the early 1900s, red foxes (Vulpes vulpes L.) were introduced to control rabbits, adding more ground disturbance and little population control (Wyngaert 2023). Despite continued disturbance during and after the historic reference period, patches of native species persisted across the American Camp grassland. These patches provide an important resource as the park develops strategies to protect cultural landscapes and restore native prairies.
Today, prairies are one of the most endangered landscapes in western Washington; about 97% of historic grasslands have been lost in the Puget lowlands of Washington (Chappell et al. 2001; Dunn and Ewing 1997). The prairies or grasslands of American Camp have persisted since the end of the Pleistocene and were maintained and utilized by Coast Salish people for thousands of years prior to the arrival of Europeans or Americans (Agee 1984; Avery 2004; Graham 2014; Stein 2000; USDA-USDI 2005). Farming during the historic time period resulted in the introduction of non-native perennial grasses across the American Camp prairie. Historically, prairies covered over 50% of American Camp and park planning efforts have long focused on restoration of these communities (Cannon 1997; Rolph and Agee 1993; USDA-USDI 2005; Figure 2). However, park management has struggled to establish realistic goals for native species composition and phasing of restoration activities.
NPS
Monitoring Implementation
In 2000, the National Park Service initiated a long-term inventory and monitoring program (I&M) that identified “vital signs” within each park as indicators of the status and long-term trends of ecosystem functions (Fancy et al. 2009; NPS 2001). During implementation stages of this program, an inventory of vascular plants and vertebrates was completed in network parks. A vascular plant inventory was completed in San Juan Island National Historical Park (SAJH) between 2001 and 2005 with the goal of documenting 90% of vascular species present; documentation included collection of voucher specimens or photographs (Rochefort and Bivin 2010). While previous studies had described the broad SAJH landscape as including forests dominated by native species, herbaceous areas were identified primarily as abandoned agricultural fields dominated by perennial introduced grasses (Agee 1984; Peterson 2002; Rolph and Agee 1993). During these surveys, more than 223 native herbaceous species were documented across herbaceous areas identified as prairie, sand dune, and sand flat floristic types. In addition, we located 87 patches, ranging in size from 25 m2 to 1.8 ha of native species across the prairies of American Camp (Rochefort and Bivin 2010; Rochefort et al. 2012; Figure 2). Following these surveys, the prairie ecosystem was designated as a SAJH vital sign for its importance to natural ecosystem function, but also because grassland physiognomy was important in preserving the scene of the cultural landscape (Weber et al. 2009).
Although we were developing a long-term vital sign monitoring strategy for SAJH, we realized that it was important to broadly track and report short-term changes in plant species distribution and composition to aid park resource managers in the protection of park resources. We deliberated as to whether we should concentrate on monitoring only the identified native polygons or a more comprehensive plan that would include the distribution of broad physiognomic types (e.g., trees, shrubs, herbaceous) and plant communities. Because the extent and distribution of prairies and forests have relevance to both the natural ecosystem and cultural landscapes (Hammond and Hoke 2004; NPS 2008), we selected a comprehensive, unit-wide plan with a hierarchy of monitoring questions.
Our long-term monitoring objectives included: 1) detect change in the extent of physiognomic cover types (e.g., trees, shrubs, and herbaceous vegetation), 2) detect change in the proportion of areas dominated by exotic plant species, 3) detect change in the quality of herbaceous cover types, and 4) detect changes in species composition and diversity of herbaceous cover types (Rochefort et al. 2012; Table 1). Pilot surveys of field methods were conducted in 2008 and 2009. Monitoring was implemented in 2012 and continued in 2013, 2014, 2015, and 2017. The program was implemented at the American Camp unit of SAJH since 90% of all prairies are located within this section of the park. Objective 4, related to changes in species composition and diversity, was not implemented due to funding and staffing constraints.
| Objective | Metric Type | Response Variables A | Ecological Integrity Rating B | ||
|---|---|---|---|---|---|
| Good | Caution | Significant Concern | |||
| 1. Detect change in the extent of physiognomic cover types | Landscape Structure | Cover Type: forest vs. non-forest C (38%:62%) | <10% difference between annual estimate of either tree or non-tree cover from baseline | 10–30% difference between annual estimate of either tree or non-tree cover from baseline | >30% difference between annual estimate of either tree or non-tree cover from baseline |
| 2. Detect change in the proportion of area dominated by exotic plant species | Vegetation Community Structure | Total Park Cover | ≤10% of area is dominated by exotic species | 11–30% of area is exotic | >30% of area is exotic |
| Vegetation Community Structure | Tree (forest) Cover | ≤10% of area is exotic | 11–30% of area is exotic | >30% of area is exotic | |
| Vegetation Community Structure | Shrub Cover | ≤10% of area is exotic | 11–30% of area is exotic | >30% of area is exotic | |
| Vegetation Community Structure | Herbaceous Cover | ≤10% of area is exotic | 11–30% of area is exotic | >30% of area is exotic | |
| 3. Detect change in quality of native herbaceous communities | Quality of Native Herbaceous Communities D | Exotic Cover | 90% of native-dominated areas have ≤10% cover of exotic species (high quality) | ≤50% of native-dominated areas have ≥50% cover of exotic species (medium quality) | >50% of native-dominated areas have ≥50% cover of exotic species (low quality) |
A Forests were defined for this study as areas predominantly covered with trees. Response variables are further defined in Tables 2 and 3.
B Ecological integrity rating reflects the reference status of the parameter. Trend describes whether the condition of the parameter is stable, improving, or declining.
C The baseline reference for this objective is the ratio of forest to non-forest cover that was present during the historic reference period of significance 1853–1875, 38% forest: 62% non-forest. The source of this metric is the ratio of soils that developed under forest vs. non-forest vegetation as interpreted by the soils survey (USDA-USDI 2005, Figure 2), as this reflects the cover types that were present during the historic reference period.
D Native herbaceous communities were defined as herbaceous-dominated segments of monitoring transects where ≥50% of vegetative cover was contributed by native species. The quality of areas meeting this threshold was then classified based on the cover of exotic species, which in poor quality segments could still exceed cover of native species due to overlapping cover. For instance, an herbaceous site having 52% cover of natives and 60% cover of exotic grasses would be classified as a low-quality native herbaceous segment.
Pilot Testing and Power Analysis
During the 2008 and 2009 field seasons, we evaluated line-intercept survey methods as a repeatable yet relatively rapid method to survey broad vegetation types across the entire prairie at American Camp (Figure 2). Using the GIS layer of native prairie polygons developed during the SAJH Inventory, we conducted a computer simulation to investigate the statistical power to detect trends in native cover of prairie vegetation under various levels of sampling intensity. Sampling intensity was defined by the number of transects, although total linear distance of transects varied along with the number of transects. Our goal was to determine the number of transects needed to detect trends in the cover of native prairie patches with a 90% probability (acceptable rate of Type 1 error of 10%). Our simulations tested a sampling intensity from 10 to 50 transects in increments of five. Since we were planning to use unmarked transects, transect origins were jiggled in a random direction and a random distance offset. For each sampling intensity level, 1000 sets of random transect locations were generated. The effect sizes chosen for the simulation were decreases of 1% and 0.5% of native prairie area per year, resulting in a loss of 3.2 and 1.6 ha, respectively, after ten years. Power was calculated as the percentage of replications that detected a significantly non-zero decreasing trend. Power estimates for the cover parameter, with a 99% probability, increased from 0.91 to 1.0 with a sample size of 35.
Sampling Design
Based on the results of these simulations, on the time required for past field surveys, and on known personnel and time constraints, we developed and field-tested a monitoring approach that included surveying a total of 25 transects positioned across the prairie, figuring that we could reasonably expect to survey this many transects per year.
We also decided to use a rotating panel design to balance our goals of estimating status and trends of vegetation cover and ecological condition (i.e., predominance of native species). Using this approach and revisiting 20 transects annually helped to reduce the overall variance and through the incorporation of five rotating panels, we were able to sample a larger area and better represent spatial variation across the American Camp landscape.
Sampling was conducted along 1-meter-wide line-intercept belt transects using a serially-alternating augmented panel design (Urquhart and Kincaid 1999) composed of 45 transects separated into an annual panel containing 20 transects plus five alternating panels containing five transects each. Alternating panels were rested for four years before each was re-surveyed. Thus, in each survey year we sampled 25 transects, consisting of the full annual panel and one alternating panel that is sampled once every five years. All transect origins were drawn and assigned to panels at random using a Generalized Random-Tessellation Stratified sample (GRTS, Figure 3). Transects were oriented along the cardinal direction from south to north and extended across the American Camp unit from either the park boundary or in cases of bluffs or shoreline, the first safely accessible point along the transect.
NPS
As we conducted the field surveys, we found ourselves tempted to add detailed vegetation/floristic types and realized this was not reproducible by multiple observers. Ultimately, we revised our vegetation types to physiognomic cover types with modifiers (Tables 2 and 3).
| Cover Type | Definition |
|---|---|
| Trees | Any segment where the tallest vegetation is a tree species. In areas where trees are becoming established in herbaceous vegetation, the cover type is tree if the tree canopy is at least 1 m wide. |
| Shrubs | Shrubs are defined by species rather than growth form. All willows are shrubs. |
| Herbaceous | All areas with ≥10% cover that are predominantly herbaceous (non-woody) vegetation. This includes prairies, sand dunes, sand flats, and mowed lawns. |
| Unvegetated | Areas with <10% cover that are not roads, trails, or buildings. This category includes open bodies of water. |
| Developed | Roads, trails, and buildings that meet the minimum resolution for documentation (i.e., 1 m of the line transect). In some cases, small social trails or animal trails were not documented. |
| Category | Term | Definition |
|---|---|---|
| Modifier (for use in Vegetated cover types) |
Native | Any cover type where vegetative cover contributed by native species is ≥50%. In trees or shrubs this applies to the overstory not the understory. So, if native trees are the tallest canopy class the cover type is called native even if there is an understory of exotic shrubs. |
| Exotic | Any cover type that does not have ≥50% cover by natives. For example, an herbaceous site having 52% cover of natives and 60% cover of exotic grasses would be classified as a low-quality native herbaceous segment. | |
| Rabbit Grazed | Additional yes/no modifier indicating that the area has evidence of rabbit grazing, pellets, or burrows that appear to have been made during the season of the survey. | |
| Substrate (for use in Unvegetated cover types) |
Sand | Fine sand |
| Gravel | Loose rocks from pebble size to cobbles | |
| Rock | Large or embedded rock substrate | |
| Soil | Soil substrate other than sand | |
| Logs | Beach logs | |
| Water | Open water such as lagoons or ponds |
Field Surveys
Surveys were conducted by two people walking along 1-meter belt transects and documenting the start and end points of segments represented by different physiognomic classes and vegetation quality modifiers (Tables 2 and 3). Surveyors used a Trimble GeoExplorer GPS unit with transect lines and other background files loaded to navigate along the unmarked transect and record changes with a 1-meter resolution. Physiognomic type and vegetative cover were recorded for each segment using ocular estimation calibrated among observers. The documented physiognomic type classes were trees, shrubs, herbaceous, unvegetated, and developed (Table 2). Vegetation modifiers included native or exotic cover and rabbit grazed (Table 3). Vegetation was labeled as native if ≥50% of the vegetative cover was contributed by species identified as native; exotic vegetation segments had <50% native cover. Herbaceous vegetation in each category (native or exotic) was further assessed for quality based on the percent of exotic cover in native areas or native vegetation in exotic dominated areas. For use in summaries and analysis, we defined high-quality native herbaceous vegetation (i.e., having ≥50% cover of native species) as having <10% exotic cover. Medium-quality native herbaceous vegetation was defined as having 11–49% exotic herbaceous cover, and low-quality native was defined as having more than 50% exotic herbaceous cover (due to the possibility of overlapping cover of different species). In unvegetated areas, we listed the substrate: sand, gravel, rock, soil, logs, or water (see Table 3).
Data Analysis
Status Estimates
Data were analyzed to estimate both annual status and trends of identified metrics (Table 1), each of which are expressed as the proportion of total transect length represented by a particular physiognomic type and quality rating. Status, defined as the annual mean for a given metric, was estimated as the weighted mean of the estimates along each transect. All transects sampled in any year were used for status estimates. Transect data were weighted to account for the different length of transects to ensure inference to the American Camp unit of SAJH. Confidence intervals were calculated with the variance of a ratio of two random variables (Cochran 1977). Individual transects were also examined to identify locations and cover changes of introduced shrub species, native vegetation patches of high or medium quality, or rabbit grazing.
Trend Estimates
Trends were fit with a linear mixed model for each of the physiognomic types over all transects and years (Piepho and Ogutu 2002). The linear mixed model fit by restricted maximum likelihood is suitable for trend analyses using repeated measures temporal data (Piepho and Ogutu 2002). We followed the approach of Pinheiro and Bates (2000) and implemented by Starcevich (2013) in coupling model assumptions with data fits.
We evaluated the ability of the logit and arcsine square-root transformations to aid in model-fitting when the proportion data were either less than 20% or greater than 80% of the total transect length. The arcsine square-root transformation has the benefit of helping to stabilize variance, though the benefits of this transformation have been questioned due to difficulty in interpretation of the modeled parameters (Warton and Hui 2011).
All models were fit using the R statistical software (Pinheiro and Bates 2000; R Core Team 2022). When the untransformed data lacked normality for a given percent cover metric, the logit and arcsine square-root transformations were evaluated. In addition, we estimated an epsilon value as defined by Warton and Hui (2011) to adjust data values that were either 0 or 100 percent.
To evaluate trends, the significance of the slope parameter was tested with a two-sided Wald t-test and Satterthwaite degrees of freedom (Kenward and Roger 1997; Satterthwaite 1946).
Results
Field Surveys
Surveys were conducted from 2007–2009, 2012–2015, and in 2017. Our goal was to sample 25 transects each year (20 annual transects and five alternating transects). Monitoring was generally conducted in May to maintain focus on a desired plant phenology. To that end, we aimed for the sampling period to coincide with early summer flowering forbs and the onset of flowering by grasses. We worked in crews of 2–3 people and generally completed surveys within 5–7 days. During the first year of sampling (2007), we tested and refined field methods and were only able to sample a total of 10 transects; in 2014, weather and other constraints limited our monitoring of the annual panel to 16 transects. Across the 11-year time series, only 15 of the 25 transects in the alternating panels had been visited a second time and could be used in the estimation of trend (Table 4).
| Transect Type | Number of Transects Surveyed (Alternating Panel Number) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 2007 | 2008 | 2009 | 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | 2017 | |
| Annual | 5 | 20 | 20 | – | – | 20 | 20 | 16 | 20 | 20 |
| Alternating | 5 (2) | 5 (3) | 7 (4 + 3) | – | – | 5 (5) | 5 (6) | 5 (2) | 5 (3) | 5 (4) |
Status Estimates
Status estimates are summarized for each metric by year in Tables 5–7. Our first objective monitored landscape structure as the ratio of forest percent cover (defined here as vegetated areas predominantly covered with trees) to non-forest vegetation. We compared this ratio to a reference ratio of 38%:62%, derived from the proportion of soil survey types that developed under forest vegetation and grassland vegetation, respectively (USDA-USDI 2005), as this reflects the cover types present during the historic reference period (1853–1875). Our monitoring data indicate an empirical range of 32%:67% to 36%:63% across the study period. Based on this, we categorized the status of physiognomic cover types over the study period as “Good” on our management scale as both cover types remained within 10% of the baseline.
| Objective | Metric | Response Variable | 2007 | 2008 | 2009 | 2012 | 2013 | 2014 | 2015 | 2017 | Ecological Integrity Rating |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1: Detect change in extent of physiognomic cover types | Landscape Structure | Cover Type: forest vs non-forest A | 36:63 | 34:65 | 36:63 | 35:64 | 33:65 | 32:67 | 36:62 | 36:61 | Good |
A The baseline for comparison is 38% forest vs 62% non-forest (grassland).
| Objective | Cover Type | Percent Exotic Vegetation Cover by Each Physiognomic Cover Type | Ecological Integrity Rating |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 2007 | 2008 | 2009 | 2012 | 2013 | 2014 | 2015 | 2017 | |||
| 2: Detect change in proportion of each vegetation type dominated by exotic plant species | All Physiognomic Cover Types | 42.7 | 49.4 | 44 | 47 | 46.1 | 48.6 | 43.9 | 44.3 | Significant concern |
| Trees | 1.1 | 0.1 | 0.5 | 0 | 0 | 0.4 | 0.4 | 0.4 | Good | |
| Shrubs | 9.7 | 10.2 | 10.4 | 12.3 | 10.2 | 17.8 | 14.3 | 11.6 | Good in 2007, Caution all other years | |
| Herbaceous | 82.8 | 87.1 | 79.1 | 85.4 | 82.7 | 82.3 | 82.0 | 85.4 | Significant concern | |
| Objective | Exotic Cover | Percent of Exotic Vegetation in Native Herbaceous Communities A | Ecological Integrity Rating B | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 2007 | 2008 | 2009 | 2012 | 2013 | 2014 | 2015 | 2017 | |||
| 3: Detect change in quality of native herbaceous communities | <10% | 37.5 | 58.4 | 39.8 | 41.8 | 31.2 | 35.7 | 36.6 | 38.8 | Caution |
| 11–49% | 38.8 | 34.9 | 44.2 | 46.4 | 58.0 | 43.6 | 41.2 | 48.5 | Caution | |
| 50–100% | 23.7 | 6.7 | 15.9 | 11.8 | 10.8 | 20.8 | 22.2 | 12.7 | Caution | |
A Native herbaceous communities were defined as herbaceous-dominated segments of monitoring transects where ≥50% of vegetative cover was contributed by native species. The quality of areas meeting this threshold was then classified based on the cover of exotic species, which in poor quality segments could still exceed cover of native species due to overlapping cover.
B Ecological Integrity Rating: Caution across all years except 2008 due to the high percentage of exotic vegetation cover in native vegetation communities.
Our second objective estimated the proportion of each vegetation type dominated by exotic vegetation. Approximately 42–49% of all vegetated portions of transects were dominated by exotic or non-native species across all years, and therefore this objective was determined to be of “Significant Concern.” Notably, few tree species detected during the monitoring were considered exotic, and thus tree cover (forest) status was rated as “Good” across all years. Non-native shrub areas exceeded 10% of coverage in all years except 2007 (9.7%) and therefore rated as “Caution”. Herbaceous vegetation status was rated as “Significant Concern” because the proportion of non-native species exceeded 80% coverage in herbaceous areas.
Objective 3 focuses on the quality of herbaceous communities, determined by the predominance of native species present. Under our operational definition, “Good” condition was indicated if >90% of all native herbaceous areas contained <10% of exotic species. “Significant Concern” was defined as >50% of all native herbaceous areas have 11–50% exotic cover (low quality herbaceous). Native herbaceous communities in our surveys were rated as “Caution” because herbaceous areas were predominantly composed of 11–50% exotic cover (medium quality).
Locations of introduced shrub species of management concern (e.g., Crataegus monogyna Jacq.), rabbit grazing, and native herbaceous vegetation that decreased in quality (high to medium) were identified and mapped. Data on individual transects were examined for locations of Crataegus monogyna (common hawthorn) and indications of interannual increases in cover that might inform annual control plans. Shrubs (native and non-native) were documented on 19 of the 20 annual panel transects and 22 of the 25 alternating panel transects. A close look revealed that non-native shrubs were located on 19 of the annual transects and 15 of the alternating transects. Crataegus monogyna was only found on 12 of the annual transects and 9 of the alternating transects. While both native and non-native shrubs were distributed across the American Camp prairie, Crataegus monogyna occurred predominantly in the western portion of American Camp and increased in extent along annual transects from 138.5 m in 2008 to 151.64 m in 2017 (Table 8, Figure 4). Two years had reduced values recorded for linear extent, 2012 with 46.59 m and 2014 with 28.3 m, which may have resulted from inadequate documentation of the species occurrence. There were only two transects on the eastern side of the prairie (Transects 2-1 and 3-2) where Crataegus monogyna was documented. Transect 2-1 was only surveyed in 2014 and the small length of the segment observed (3.6 m) may represent a recent establishment of the species.
| Transect A | Combined Segment Length (m) by Year | ||||||
|---|---|---|---|---|---|---|---|
| 2008 | 2009 | 2012 | 2013 | 2014 | 2015 | 2017 | |
| AC.1-18 | – | – | 4.54 | 6.80 | NS B | 23.11 | 5.33 |
| AC.1-6 | – | – | – | – | NS | 39.57 | 6.73 |
| AC.1-12 | 9.37 | 5.4 | – | 14.79 | – | 45.98 | 66.77 |
| AC.1-4 | – | 62.56 | – | – | 5.18 | 4.21 | 5.38 |
| AC.2-2 C | NS | NS | – | NS | 7.45 | NS | NS |
| AC.4-1 C | – | – | – | – | – | – | 54.33 |
| AC.1-15 | – | 1.75 | – | – | NS | 12.82 | 13.2 |
| AC.6-3 C | NS | NS | – | 2.24 | NS | NS | NS |
| AC.1-2 | 129.13 | – | 42.05 | – | 5.95 | 34.37 | – |
| AC.1-9 | – | – | – | – | 2.49 | 3.54 | – |
| AC.2-5 C | NS | NS | – | NS | 1.24 | NS | NS |
| AC.1-13 | – | – | – | 73.64 | 4.37 | 11.37 | 19.15 |
| AC.2-3 C | NS | NS | – | NS | 91.42 | NS | NS |
| AC.3-4 C | – | NS | – | NS | NS | 98.13 | NS |
| AC.4-5 C | NS | NS | – | NS | NS | NS | 20.16 |
| AC.1-19 | – | – | – | 2.12 | 2.76 | 10.74 | 2.9 |
| AC.1-7 | – | – | – | – | 1.45 | 13.23 | 12.09 |
| AC.3-1 C | – | NS | – | NS | NS | 3.58 | NS |
| AC.1-16 | – | 32.54 | – | – | – | – | – |
| AC.1-10 | – | 4.82 | – | – | 6.00 | 9.35 | 8.77 |
| AC.2-1 C | NS | NS | – | NS | 3.6 | NS | NS |
| AC.3-2 C | – | – | – | – | – | 5.98 | – |
| Total Length (m) – Annual Transects | 138.5 | 107.07 | 46.59 | 97.35 | 28.2 | 204.75 | 151.64 |
| Total Length (m) – All Transects |
138.5 | 107.07 | 46.59 | 99.59 | 131.91 | 312.44 | 226.13 |
A Transects are listed in order from west to east.
B NS = Not Surveyed, Cells with “–” indicate zero cover by Crataegus monogyna.
C Indicates alternating panels (also with gray shading).
NPS
Reviewing coverage of herbaceous vegetation along survey transects revealed that native and non-native herbaceous patches were well-distributed across the American Camp prairie: 19 of 20 annual transects and 22 of 25 alternating transects had coverage consisting of both native and non-native herbaceous vegetation. The transect surveys confirmed the location of native “polygons” documented in the earlier plant inventory (Rochefort and Bivin 2010) but also located numerous additional areas with a predominance of native species (Figure 5), supporting our focus on unit-wide monitoring.
NPS
Trend Estimates
Trends were calculated for the proportions of transect length represented by the physiognomic cover types (Objective 1), the proportion of exotic-dominated areas (Objective 2), and native herbaceous quality (Objective 3). Metrics for the five broad cover types in Objective 1 (herbaceous, shrub, tree, developed, and unvegetated) were modeled with each of the three possible transformations: untransformed, logit, and arcsine square-root. Herbaceous and tree areas were modeled with an untransformed response. The models for shrub, unvegetated, and developed types did not satisfy the assumptions of normality with either the untransformed data or the two transformations, likely because most of the observed values for shrubs were less than 20% and most of the observed values for developed areas were at or near zero. Therefore, final models were not interpreted for the shrub, unvegetated, or developed types.
The estimated trend in herbaceous cover decreased at a rate of 0.58% per year (95% CI: −0.78 to −0.37). This trend was significantly different from zero (p < 0.0001). The estimated trend in tree cover increased at a rate of 0.25% per year (95% CI: −0.02, 0.52); however, this trend was not significantly different from zero (p = 0.0638).
The four metrics in Objective 2 (exotic herbaceous, exotic shrub, exotic tree, and the combined exotic vegetation) were modeled with each of the three possible transformations: untransformed, logit, and arcsine square-root. The herbaceous, shrub, and composite measure of exotic vegetation cover was modeled with an untransformed response. The models for exotic tree areas did not satisfy the assumptions of normality with either the untransformed data or the two transformations, likely due to the predominance of zero values, and so final models were not interpreted for this metric.
The estimated trend in the proportion of exotic herbaceous areas increased at a rate of 0.08% per year (95% CI: −0.81, 0.96), though the estimate was not significantly different from zero (p = 0.8417). Exotic shrub proportions increased at a rate of 0.24% per year (95% CI: −0.45, 0.92); however, this rate was not significantly different from zero (p = 0.4443). The estimated trend in exotic vegetation overall decreased at a rate of 0.31% per year (95% CI: −0.75, 0.13), though the estimate was not significantly different from zero (p = 0.1459).
Three cover metrics in Objective 3 (high-quality native herbaceous, medium-quality native herbaceous, and low-quality native herbaceous) were modeled with each of the three possible transformations: untransformed, logit, and arcsine square-root. The logit-transformed response appeared to demonstrate the best fit for high-quality areas. The estimated trend in high-quality native herbaceous cover decreased at a rate of 0.08% per year (95% CI: −0.22, 0.05). This trend was not significantly different from zero (p = 0.1393).
The model for medium-quality herbaceous vegetation did not satisfy the assumption of independent slope and intercept random effects with either untransformed data or using the two transformations. The model for low-quality native herbaceous vegetation did not satisfy the assumptions of normality with either the untransformed data or using the two transformations, likely because many values were less than 20%. Therefore, final models were not interpreted for these two metrics.
Discussion
The prairie monitoring program at SAJH was initiated with broad monitoring questions regarding the ratio of forested to non-forested areas because broad, physiognomic landscape patterns are of primary importance to the cultural landscape and indicative of long-term vegetation change. Additionally, landscape vistas are important to many park visitors. Status estimates from the present study confirmed that today’s visitor to SAJH sees the overall park landscape much as it was during the historic reference period—predominantly as open prairie (61%) with forest cover on north-facing slopes and leeward aspects (36%). However, a closer examination of vegetation composition reveals that the native composition and ecological integrity of the park landscapes is at risk. While native species dominate tree- and shrub-dominated areas, exotic perennial grass species predominate in herbaceous communities (85%). Our findings indicate that documented shifts in SAJH vegetation communities are likely to continue in the future. Specifically, woody encroachment is likely to serve as a persistent driver of change and, as a result, the proportion of the park with herbaceous vegetation will continue to decrease. In addition, continued expansion of introduced species is likely to further alter the condition of native prairie patches. While our results do not clearly link changes in herbaceous communities to a specific agent, status estimates for Objectives 2 and 3 align with our observations that high-quality prairies are diminishing in some areas to lower quality with continued invasion by introduced shrubs (e.g., Crataegus monogyna, Daphne laureola) and grasses (e.g., Poa pratensis L., Bromus tectorum L., Elymus repens (L.) Gould, Holcus lanatus L.).
Trend estimates are useful indicators for park managers projecting the condition of future landscapes, communicating park conditions to the public, developing strategic planning documents, and preparing funding proposals for project work. However, annual work plans and efforts to protect native plant communities are often better served by annual status reports and maps aimed at informing rapid response. Data collected during annual surveys can easily be imported into Geographic Information Systems that support spatial analyses or to visualize the information on maps (see Figure 4 for an example). This information helps land managers make informed decisions by highlighting areas where specific resource management concerns such as new or expanded introduced (exotic) plant invasions, rabbit grazing, or a decline in quality of native prairie patches are documented.
Park managers often have in-depth local knowledge of recent weather conditions or events that may inform interpretation of composition changes documented along the transects. For example, in some years there was a noted decline in native shrubs or a regrowth of dying Himalayan blackberry (Rubus armeniacus). The decline in native shrubs was a result of park mowing and would not have necessitated a quick management response. However, the regrowth of Himalayan blackberry would signal the need for retreatment of areas where exotic plant management had not been successful in permanent removal of the non-native shrub. New patches of Crataegus monogyna and expansion of existing patches would also indicate a need for rapid response.
Our broadest goal for long-term monitoring of prairies at SAJH was to develop a statistically robust, repeatable program that could address cultural and ecological concerns regarding vegetation cover and ecological integrity and to provide these data to managers to inform future condition assessment and management efforts. Based on current and projected budgets, we also needed a method for rapid assessment that could be implemented economically by a field crew that would rely on the participation of volunteers or practitioners from nearby parks (i.e., limited experience with SAJH vegetation).
We found that transect monitoring methods using broad physiognomic types was relatively easy for a transient field crew to implement. Over the 11-year time series, there were three people who participated most years; the remainder participated for anywhere from a single day in one year to one or two days for multiple years. Despite revolving crews, survey results demonstrated consistency year-to-year and crew members appeared comfortable with survey methods after the first full day of field training. Competency on vegetation identification was more difficult to attain than GPS operation or general survey methodology. In addition, each survey crew always had one person who could identify most species and since we were focused on dominant cover, learning a small subset of species (20–30) permitted accurate characterization of cover origin (i.e., native vs. exotic). However, more expertise would have been required had there been funding to allow implementation of protocols to monitor species composition and diversity (i.e., Objective 4).
Revisions to existing monitoring methods may support improvements in data utility and interpretation. One recommendation would be to add a standardized list of species of special concern and a drop-down menu on GPS input screens. This would be an improvement over current practices of recording the presence of species of concern in a Notes field on the data form, where the level of detail can vary considerably across observers. Second, existing procedures only require observers to record evidence of rabbit browse, phenologic stage of indicator species, and physiognomic type. This could be expanded to record the presence or absence of a larger set of species of management concern such as non-native invasive species and host plants for species of concern. Specifically, Lepidium virginicum L., Brassica rapa L., and Sisymbrium altissimum L. are hosts for the federally endangered island marble butterfly (Jones et al. 2026) and Abronia latifolia Eschsch. is a host for the sand verbena moth (a Washington state candidate species; WDFW 2026).
Sampling design revisions could include the addition of transects in areas of special management concern where there are currently no transects. For example, there are very few transects located within a 2 ha (5 ac) revegetation area where introduced grasses were removed and Festuca roemerii and native forbs were planted (Figure 5). New transects could be located at random within these (subjectively identified) areas and vegetation plots could be established providing more detailed species composition, cover, and frequency data. Finally, we recommend the continuation of annual transect surveys to document status and trends in vegetation cover, with an emphasis on the rapid data summaries to inform annual work plans. Analysis of long-term trends could be conducted on a longer rotation such as every ten years.
Literature Cited
- Agee, J.K. 1984. Historic landscapes of San Juan Island National Historical Park. National Park Service, Cooperative Studies Unit, College of Forest Resources, University of Washington, Seattle, Washington.
- Avery, C. 2004. San Juan Island National Historical Park: An environmental history. National Park Service, Pacific West Regional Office.
- Cannon, K.J. 1997. Administrative history of San Juan Island National Historical Park. Prepared for the National Park Service. https://irma.nps.gov/DataStore/Reference/Profile/2186262 (accessed April 25, 2021).
- Chappell, C., M.S. Gee, B. Stephens, R.C. Crawford, and S. Farone. 2001. Distribution and decline of native grasslands and oak woodlands in the Puget lowland and Willamette valley ecoregions, Washington. Pages 124–239 in S.H. Reichard, P.W. Dunwiddie, J.G. Gamon, A.R. Kruckeberg, and D.L. Salstrom (eds.). Conservation of Washington’s rare plants and ecosystems. Washington Native Plant Society, Seattle, Washington.
- Cochran, W.G. 1977. Sampling techniques. Third Edition. John Wiley & Sons. New York, New York.
- Dunn, P., and K. Ewing, eds. 1997. Ecology and conservation of the South Puget Sound prairie landscape. The Nature Conservancy of Washington, Seattle, Washington.
- Fancy, S.G., J.E. Gross, and S.L. Carter. 2009. Monitoring and condition of natural resources in US National Parks. Environmental Monitoring and Assessment 151:161–174.
- Graham, J.P. 2014. San Juan Island National Historical Park: Geologic resources inventory report. Natural Resource Report NPS/NRSS/GRD/NRR—2014/835. National Park Service, Fort Collins, Colorado.
- Hammond, J., and A. Hoke. 2004. National Park Service Cultural Landscape Inventory 2004: American Camp, San Juan Island National Historical Park. Cultural Landscapes Inventory Report, 400105. National Park Service, Pacific West Regional Office, Pacific West Regional Office/CLI Database. https://irma.nps.gov/DataStore/Reference/Profile/2185604 (accessed April 6, 2023).
- Jones, K.S., A.W. Aunins, C.C. Young, R.J. Johnson, and C.L. Morrison. 2026. Population genetics of the endangered narrowly endemic Island Marble butterfly (Euchloe ausonides insulanus). Conservation Genetics 27:5.
- Kenward, M.G., and J.H. Roger. 1997. Small sample inference for fixed effects from restricted maximum likelihood. Biometrics 53(3):983–997.
- National Park Service (NPS). 2001. Park Vital Signs Monitoring – a commitment to resource protection. National Park Service, Natural Resource Program Center, Fort Collins, Colorado.
- National Park Service (NPS). 2008. San Juan Island National Historical Park final general management plan and environmental impact statement. National Park Service Pacific West Region, Park Planning and Environmental Compliance, Seattle, Washington. https://parkplanning.nps.gov/projectHome.cfm?parkID=340&projectID=11187 (accessed April 26, 2021).
- Peterson, D.L. 2002. Developing a vegetation and fuels data base for San Juan Island National Historical Park. Unpublished report. USGS Forest and Rangeland Ecosystem Science Center, Cascadia Field Station, Seattle, Washington.
- Piepho, H.P., and J.O. Ogutu. 2002. A simple mixed model for trend analysis in wildlife populations. Journal of Agricultural, Biological, and Environmental Statistics 7(3):350–360.
- Pinheiro, J.C., and D.M. Bates. 2000. Mixed-effects models in S and S-PLUS. Springer, New York, New York. https://doi.org/10.1007/b98882
- R Core Team. 2022. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/
- Rochefort, R.M., and M.M. Bivin. 2010. Vascular plant inventory of San Juan Island National Historical Park. Natural Resource Technical Report NPS/NCCN/NRTR—2010/350. National Park Service, Fort Collins, Colorado.
- Rochefort, R.M., M.M. Bivin, J.R. Boetsch, L. Grace, S. Howlin, S.A. Acker, C.C. Thompson, and L. Whiteaker. 2012. Prairie vegetation monitoring protocol for the North Coast and Cascades Network. Natural Resource Report NPS/NCCN/NRR—2012/538. National Park Service, Fort Collins, Colorado.
- Rolph, D.N., and J.K. Agee. 1993. A vegetation management plan for the San Juan Island National Historical Park. Natural Resource Technical Report NPS/PNRU/NRTR—93/02. National Park Service, Pacific Northwest Region.
- Satterthwaite, F.E. 1946. An approximate distribution of estimates of variance components. Biometrics 7(2):110–114.
- Starcevich, L.A.H. 2013. Trend and power analysis for National Park Service marine data from 2006 to 2010: Pacific Island Network. National Resource Technical Report NPS/PACN/NRTR—2013/771. National Park Service, Fort Collins, Colorado.
- Stein, J.K. 2000. Exploring the Coast Salish history, the archaeology of San Juan Island. The Burke Museum of Natural History and Culture, University of Washington Press, Seattle, Washington.
- Urquhart, N.S., and T.M. Kincaid. 1999. Designs for detecting trend from repeated surveys of ecological resources. Journal of Agricultural, Biological, and Environmental Statistics 4(4):404–414.
- USDA Natural Resources Conservation Service and USDI National Park Service (USDA-USDI). 2005. Soil survey of San Juan Island National Historical Park, WA. National Park Service TIC No. D-87. National Park Service, Fort Collins, Colorado. https://irma.nps.gov/DataStore/Reference/Profile/2186033
- Warton, D.I., and F.K. Hui. 2011. The arcsine is asinine: the analysis of proportions in ecology. Ecology 92(1):3–10.
- Washington Department of Fish and Wildlife (WDFW). 2026. Washington Department of Fish and Wildlife, threatened and endangered species. https://wdfw.wa.gov/species-habitats/at-risk/listed?state_status=25402 (accessed April 22, 2026).
- Weber, S., A. Woodward, and J. Freilich. 2009. North Coast and Cascades Network vital signs monitoring report (2005). Natural Resource Report NPS/NCCN/NRR—2009/098. National Park Service, Fort Collins, Colorado.
- Wyngaert, W. 2023. Island Histories: San Juan Island’s foxes and rabbits. https://www.islandhistories.com/items/show/11 (accessed April 6, 2023).
About the National Park Service Science Report Series
The National Park Service Science Report Series disseminates information, analysis, and results of scientific studies and related topics concerning resources and lands managed by the National Park Service. The series supports the advancement of science, informed decisions, and the achievement of the National Park Service mission.
All manuscripts in the series receive the appropriate level of peer review to ensure that the information is scientifically credible and technically accurate.
Views, statements, findings, conclusions, recommendations, and data in this report do not necessarily reflect views and policies of the National Park Service, U.S. Department of the Interior. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by the U.S. Government.
The Department of the Interior protects and manages the nation’s natural resources and cultural heritage; provides scientific and other information about those resources; and honors its special responsibilities to American Indians, Alaska Natives, and affiliated Island Communities.
This report is available in digital format from the National Park Service DataStore and the Natural Resource Publications Management website. If you have difficulty accessing information in this publication, particularly if using assistive technology, please email irma@nps.gov.