Canada: British Columbia: Restoration of Antelope-Brush Aridlands in the Southern Okanagan Valley


In British Columbia’s Okanagan Valley lies a small pocket of desert constituting the northernmost extension of the Great Basin Desert and Canada’s only such ecosystem. A mix of shrub-steppe and bungrass grassland habitats, this arid region is increasingly threatened by development in the valley and the resultant fragmentation of critical habitats. In 1998, the Osoyoos Desert Society acquired a 50-hectare site and began implementing restoration experiments aimed at evaluating different techniques for seeding native grasses, as well as strategies for eliminating several prolific non-native species. By gaining a more thorough understanding of ecosystem processes and factors influencing vegetative composition, researchers hope to create a foundation for the restoration and conservation of larger areas of the Okanagan Valley.

Quick Facts

Project Location:
Okanagan Valley, BC, Kootenay Boundary E, BC, Canada, 49.781264, -118.828125

Geographic Region:
North America

Country or Territory:

Desert/Arid Land


Area being restored:
50 hectares

Project Lead:
Osoyoos Desert Society

Organization Type:
NGO / Nonprofit Organization


Project Stage:

Start Date:

End Date:

Primary Causes of Degradation

Agriculture & Livestock, Fragmentation, Invasive Species (native or non-native pests, pathogens or plants), Urbanization, Transportation & Industry

Degradation Description

Development for housing, roads, agriculture, and other uses has greatly reduced natural ecosystems in the South Okanagan, particularly in the valley bottoms (BC MoE 2001). Some ecological communities, such as the antelope-brush – needle-and-thread grass community, have been reduced to less than 40% of their historical extent, and are continually being threatened with even greater habitat loss (Dyer and Lea 2003). Riparian ecosystems have been very affected by development and by the channelization of the Okanagan River, and only 15% of these ecosystems remain (BC MoE 2001).

This human encroachment is such that habitat fragmentation has become a serious problem for wildlife. Wetlands and ponds are often separated from other habitat types by roads, fields, orchards, or developments, making it very difficult for species that require multiple habitat types (e.g. the spadefoot toad, which must migrate from pond to grassland and back throughout its life) (BC MoE 2001). Natural areas containing similar ecosystems are also segregated, making it difficult or impossible for populations to inter-breed or to colonize new habitat. Species richness and genetic diversity within species decrease with a reduction in habitat patch size (Duerksen et. al. 1997); and this reduction in patch size, caused by fragmentation, is what has contributed to the long list of threatened or endangered species in the South Okanagan. Species that have suffered as a result of this fragmentation include: coyotes (Canis latrans), mule deer (Odocoileus hemionus), American badger (Taxidea taxus), black bears (Ursus americanus), Great Basin gopher snake (Pituophis catenifer deserticola), Nuttall’s cottontail (Sylvilagus nuttallii), and many bird species [e.g. western meadowlark (Sturnella neglecta), mountain bluebird (Sialia currucoides), and lark sparrow (Chondestes grammacus)].

Cattle grazing is another problem in the South Okanagan, as grazing decreases the density of native bunchgrasses (Daubenmire 1970), destroys the cryptobiotic crust (an integral component of healthy soil) (Cannings and Durance 1998), and has driven the spread of non-native species such as cheatgrass (Daubenmire 1970). In fact, cheatgrass and Japanese brome (Bromus japonicus), two invasives, represent the largest threat to the health of the Osoyoos Desert Centre site’s natural communities. These species are capable of forming their own climax communities, to the exclusion of more desirable native species (Daubenmire 1970). In the case of cheatgrass, there is no evidence that it ever relinquishes an area to indigenes once it has become established (Daubenmire 1970).

Reference Ecosystem Description

About 20% of the Osoyoos Desert Centre Site contains antelope-brush habitat (Atwood 1996). This habitat is generally found on sandy-loam or loamy sand-textured soils in the project site, though antelope-brush shrub-steppe on the west edge of the site contains more silt and clay (Atwood 1996). The antelope-brush shrub-steppe section of the Osoyoos Desert Centre site has an average shrub cover of 39% (dominant shrubs are antelope brush and great basin sage); a perennial native grass cover of 20% (dominant grasses are sand dropseed [Sporobolus cryptandrus] and needle-and-thread grass [Stipa comata]); a native forb cover of 11% (dominant forbs are snow buckwheat [eriogonum niveum], arrow-leaf balsam root [Balsamorhiza sagittata], brittle prickly-pear cactus [Opuntia fragilis], Indian wheat [Plantago patagonica], and yarrow [Achillea millefolium]); a micriobiotic crust cover of 25% (dominant species is rusty steppe moss [Torula ruralis]); and a non-native species cover of 57% (dominant species are cheatgrass [Bromus tectorum], diffuse knapweed, spotted knapweed [Centaurea maculosa], sweet white clover [Melilotus alba] and Japanese brome [Bromus japonicus]) (Atwood 1996).

The northern section of the Osoyoos Desert Centre site contains most of the Great Basin sage shrub-steppe, which makes up about 20% of the total site. Soil texture for this habitat type ranges from clay to silty loam (Atwood 1996). The Great Basin sage shrub-steppe component has a shrub cover of 31% (dominant shrubs are the Great Basin sage, common rabbitbrush [Chrysothamnus nauseosus] and green rabbitbrush [Chrysothamnus viscidiflorus]); a perennial native grass cover of 33% (dominant grasses are needle-and-thread grass, and sand dropseed); a native forb cover of 12% (dominant forbs are Indian wheat, pale comandra [Comandra umbellatum], long-leaved phlox [Phlox lonifolia], and yarrow); a microbiotic crust cover of 11% (dominant species are rusty steppe moss, fire moss [Ceratodon purpureus], and lichen species [Cladonia spp.]); and a non-native species cover of 42% (dominant species are cheatgrass, Japanese brome, and sweet white clover) (Atwood 1996).

Grassland habitat is located on flat areas of the Osoyoos Desert Centre site, and covers about 51% of the site (Atwood 1996). The soil texture is generally loamy sand, and a there is a large percentage (16%) of bare ground associated with the grassland sites (Atwood 1996). The bunchgrass grassland found at the site has an average shrub cover of 7% (dominant shrubs are antelope brush, common rabbitbrush, Great Basin sage, and silver sage [Artemisia cana]); a perennial native grass cover of 18% (dominant grasses are sand dropseed, and needle-and-thread grass); a native forb cover of 19% (dominant forbs are brittle prickly-pear cactus, Indian wheat, snow buckwheat, and golden aster [Heterotheca villosa]); a microbiotic crust cover of 16% (dominant species are rusty steppe moss, fire moss, and foliose and fruticose lichens [Peltigera didactyla and P. rufescens, and Cladonia spp., respectively]); and a non-native species cover of 51% (dominant species are diffuse knapweed, cheatgrass, crested wheatgrass, sweet white clover, spotted knapweed, Japanese brome, Kentucky bluegrass [Poa pratensis], quackgrass [Agropyron repens], toadflax [Linaria genistifolia], hound’s tongue [Cynoglossum officinale], and asparagus [Asparagus officinalis]) (Atwood 1996).

Riparian areas can be found in depressions and run-off areas, and their soil texture ranges from clay to silty clay loam. The riparian areas contain a few sporadic shrubs (dominant shrubs are Saskatoon [Amelanchier alnifolia], mock orange [Philadelphus lewisii], poison ivy [Rhus radicans], Great Basin sage, and antelope brush); a perennial native grass cover of 15% (dominant grasses are sand dropseed and needle-and-thread grass); a native forb cover of 2% (dominant forbs are lemonweed [Lithospermum ruderale], showy milkweed [Asclepias speciosa], shaggy daisy [Erigeron pumilus], parsnip-flowered buckwheat [Eriogonum heracloises] and yarrow); a microbiotic crust cover of 8% (dominant species are rusty steppe moss, golden curls moss [Homalothecium aeneum], and shaggy yellow sand moss [Racomitrium ericoides]); and a non-native species cover of 40% (dominant species are Japanese brome, cheatgrass, diffuse knapweed, and sweet white clover) (Atwood 1996).

A small portion of the Osoyoos Desert Centre site is comprised of rock outcrop habitat, which contains very shallow soils of a silty clay loam texture. The rock outcrop sites contain a shrub cover of 12% (dominant shrubs are Great Basin sage, Saskatoon, and prickly phlox, and antelope brush); a perennial native grass cover of 35% (dominant species are bluebunch wheatgrass and sandberg bluegrass [Poa secunda]); a native forb cover of 79% (dominant forbs are compact selaginella [Selaginella densa], low pussytoes [Antennaria dimorpha], thread-leaved daisy [Erigeron filifolius], snow and parsnipflowered buckwheat, round-leaved alumroot [Heuchera cylindrica], and bitterroot [Lewisia redeviva]); a microbiotic crust cover of 2% (dominant species are rusty steppe moss, a foliose lichen [Peltigera rufescens], and a fruticose lichen [Cladonia spp.]); and a non-native species cover of 9% (dominated by cheatgrass, Japanese brome, and diffuse knapweed) (Atwood 1996).

Project Goals

– Restore the community structure, species composition, and function of the natural habitats
– Reduce the weedy species component
– Increase the cover and recruitment rate of native grasses and forbs
– Replace areas of early successional species with climax species
– Maximize species diversity within each habitat type
– Build local stewardship for the landscape as a means of promoting habitat conservation


The project does not have a monitoring plan.

Description of Project Activities:
Eight restoration experiments were conducted on the Osoyoos Desert Centre site between 1998 and 2002. Approximately 40% of the native seed required for these experiments was collected during the 1998 field season. As seed matured, seed heads and plant stalks were cut from four native bunchgrasses: Aristida longiseta (red three-awn), Hesperostipa comata (needle and thread grass), Sporobolus cryptandrus (sand dropseed), and Pseudoroegneria spicata (bluebunch wheatgrass). The plant material was collected from natural shrub-steppe communities within the South Okanagan Basin Ecosection. Because extensive stands of native bunchgrasses are exceedingly rare in the South Okanagan, and access to them is difficult, the seed-collection process required a great deal of time and resources. Most of the remaining natural areas are on private land or are protected, and many landowners and land managers are reluctant to allow native seed harvesting. In addition to a poor supply, few seed collectors are trained, and seed cleaning techniques are rudimentary. The time required to gather and clean native seed varied by species, but producing 1 kg of pure seed (pick and clean) ranged from 6 person-hours for sand dropseed (Sporobulus cryptobulus) to 57 person-hours for bluebunch wheatgrass. Three hundred and five 100-m2 experimental plots were then delineated, staked, and tagged (60 plots each at 5 experiment replicates at the same study site, plus 5 plots selected solely to assess the effect of control of dalmation toadflax). Baseline vegetation data was collected from 3 1-m2 quadrats randomly located within each 100-m2 plot (n = 915). Percent cover, distribution and vigour of the vascular plant species, along with percent cover of moss, lichen, litter, and bare ground, were collected from each quadrat. Soil texture data were also collected from the 100-m2 plots. 3,075 kg of noxious weeds were hand-pulled, bagged, and removed from areas around the experimental plots. The targeted species were diffuse knapweed (Centaurea diffusa), dalmatian toadflax (Linaria genistifolia ssp. dalmatica), hound's tongue (Cynoglossum officinale), and mullein (Verbascum thapsus). Following weed removal, bare soils were covered with native hayseed. This hayseeding experiment, intended to assess variables influencing species recruitment, was initiated on two plots per replication (10 plots total) in September 1998. Approximately 200 litres of plant material (50 litres from each species) was distributed evenly over each 100-m2 plot. In another experiment conducted in October 2000, broadcast seeding was initiated to evaluate the effect on native species establishment of different soil-disturbance conditions and different seed rates. 100-m2 plots were double split, producing four 25-m2 subplots. One-half of the plots were tilled to mimic soil disturbance that would be associated with development projects. Shrubs remained, but existing herbaceous vegetation was cut and removed from the plot before tilling and the soil was packed after tilling. Standing herbaceous vegetation was also cut and removed from the no-till plots. Non-native species remaining in the plots were spot treated with the herbicide glyphosate (Roundup), applied at the full label rate. The seed mix consisted of four perennial native bunchgrasses (A. longiseta, H. comata, S. cryptandrus, and P. spicata) and one annual agronomic Lolium multiflorum (annual ryegrass). The native grasses were combined evenly in the mix (25% live seed per species), and seed rates were 28 kg / ha (1027 seeds / m2) and 41 kg/ha (1504 seeds / m2). Application rates were adjusted to account for the germination rate of the collected seed. Each seed rate was broadcast on one-half of the 100-m2 plot and the soil was rolled after seeding. Also in the fall of 2000, an experiment was conducted to determine how the addition of native vesicular arbuscular mycorrhizae (VAM) fungi might affect natural grass species establishment. One 100-m2 plot was divided into four 25-m2 subplots, and two treatments (Nurse plant inoculant and Soil-Root inoculant) and two control plots (no inoculant) were randomly established in each plot. Native VAM was produced on site for the inoculant treatments. Mature bunchgrasses on the site where tested and found to be colonized with vesicular arbuscular mycorrhizae hyphae, ranging from an average percent colonization of 36.19% + 2.28% for A. longiseta, 38.91% + 1.69% for S. cryptandrus, 53.98% + 1.13% for H. comata, and 55.4% + 4.08% for P. spicata. Thus, bunchgrasses from the site were used as nurse plants (Nurse plant treatment) and also used to produce VAM beds that were harvested for the Soil-Root treatment. The harvested material was added to trenches in the experimental plots, spaced 15-cm apart. The plots were tilled, inoculated (or not), and then seeded with the native grass seed mix at 28 kg/ha. Percent cover data for the seeded native grasses were collected in June 2001 and 2002. Besides experiments aimed at optimizing seeding techniques, project activities also encompassed experiments designed to evaluate different strategies for the elimination of invasive species. The first, begun in 1998, assessed the efficacy of solarization as a passive weed-removal technique. Ten 100-m2 plots were covered with transparent polyethylene sheets in 1998 and the plastic was removed from 5 of the plots in April 2000. Vegetation data were collected from the plots before the plastic was put down and again following its removal (June 2000 and 2002). Besides solarization, manual and chemical weed removal was tested. One 100-m2 plot per replication was randomly chosen for the manual weed control, and two plots were treated chemically. The manual control of C. diffusa experiment was to determine the most effective time to hand weed C. diffusa, and whether weed density was related to the timing of the manual control. The first hand-pulling was scheduled for early May, after which monthly treatments were scheduled if weed density was > 25% of the original C. diffusa cover. The experiment for the chemical control of C. diffusa was implemented in 2000. C. diffusa plants in two 100-m2 plots per replication (10 plots) were spot sprayed with an over-the-counter broadleaf herbicide, Killex, at the recommended label rate of 1.85 kg active ingredient per hectare in May 2000. Killex, an over-the-counter combination of 2,4-D, mecoprop, and dicamba was used in the chemical control experiment to determine if adequate control of C. diffusa could be obtained using a less expensive broad-leaf herbicide with less residual than the commonly used Tordon 22K (picloram). Finally, in 1998, grazing exclusion was tested as a means of eliminating invasives. Two randomly-chosen, 100-m2 plots in each replication were established as control plots (10 plots). Species identity and percent cover data were collected annually, in June, from 1998 to 2002. The control plots were monitored to document changes in the plant community without livestock grazing or restoration activity.

Ecological Outcomes Achieved

Eliminate existing threats to the ecosystem:
On the Osoyoos Desert Centre research site, locally collected natural grasses established as a result of both hayseeding and broadcast seeding, although after the first growing season, soil condition (disturbed versus undisturbed) and soil type [high sand content (Rep 5) versus higher silt content (Rep 1 to 4)], coupled with ecological preferences of the grass species, had more of an effect on overall cover of seeded natural grasses than the seed rate. Broadcast seeding was more effective than hayseeding on undisturbed soils, and seed rate did not affect establishment. Rates of greater than 1000 s/m2 are thought to be high (Jacobs et al. 1999), and it appears applications that exceed that amount are unnecessary. The average cover of 28% that was reported on site would indicate that roughly 290 of the 1027 seeds planted germinated. Further work is required to determine if the level of plant establishment is a reflection of the carrying capacity of the local soils, given their low moisture and nutrient availability (Wicklow- Howard 1994), or the result of self-induced seed dormancy, which has limited germination in the harsh environmental conditions (Halvorson 1989, Allan et al 1994, O'Keefe 1996). The hayseed appeared to repress one of the most common grass species on the site, S. cryptandrus. The average cover of S. cryptandrus fell slightly in Rep 1 to 4 over the four years as compared to an 18% increase in cover in the control plots. S. cryptandrus cover increased in Rep 5, although the control plots increased at a higher rate (90% vs. 84%). When the hayseeding experiment was initiated, Rep 5 contained twice the amount of bare ground as Reps 1 to 4. Light availability was likely higher in Rep 5, even with the hayseed cover. Sabo et al (1979) reported germination of S. cryptandrus increased with light availability. In contrast, the hayseed cover enhanced P. spicata and H. comata establishment. H. comata only established in the sandy soils of Rep 5 when covered with hayseed mulch, and P. spicata, which was not found in the research plots before seeding, only established in areas that received the hayseed mulch or VAM inoculant. All of the seeded grasses are mycorrhizal (Trappe 1981) and VAM is particularly critical for the establishment of warm season grasses (Clapperton and Ryan 2001), which would include A. longiseta and S. cryptandrus. To date, VAM colonization levels that will improve grass establishment are unknown. Solarization was not an effective weed control method for the primary weeds on the Osoyoos Desert site. C. diffusa and V. thapsus germinated readily following the removal of the plastic, indicating the 75ºC recorded under the plastic during treatment was not sufficient to kill the seeds. In addition, solarization resembles broadcast herbicide treatment, exposing large expanses of bare soil after treatment. Revegetating solarized areas with only locally collected native species will also require a consistent and long-term weed control program. Manual and chemical control of C. diffusa did reduce the weed component; however, livestock removal was the most effective weed control measure. In one year, weed cover in plots monitored for the effect of livestock removal dropped an average of 74% (71% in Rep 1-4 and 77% in Rep 5), and over five years there was a significant increase in native grass and herb cover. The rapid decline of C. diffusa is puzzling, however, since the species is known to have an extensive and long-lived seed bank. Reduced soil disturbance is a factor because C. diffusa did germinate in the tilled plots. However, Clements et al (unpublished) also found few viable knapweed seeds on the site, which may be the successful result of Sphenoptera jugoslavica (biological control agent - beetle), which occurs throughout the area. By 2002, A. cristatum was the dominant non-native species on the site.

Factors limiting recovery of the ecosystem:
Species establishment was influenced by soil type. Soil texture differed over the site, with textures in replications (Reps) 1 - 4 being similar, but Rep 5 soils containing more sand and less silt. On average, Rep 5 also contained a higher cover of litter, bare ground, weeds, and native herbs, but fewer native grasses than Reps 1 - 4. A. longiseta had higher establishment in sandier soils (Rep 5), while H. comata did best in siltier soils (Rep 1 to 4). The percent cover of H. comata was almost 7 times higher in Reps 1 to 4 than in Rep 5. A. longiseta and H. comata are both promoted as drought tolerant species and yet, the limited establishment of A. longiseta, and preferential establishment of H. comata, suggests they were affected by the droughty conditions experienced in the South Okanagan over the past four years. Indeed, Weaver (1968) did find that A. longiseta decreased in extended droughts. It is also possible that the high heat requirement for A. longiseta germination was not met prior to the June data collection (Evans and Tisdale 1972).

Socio-Economic & Community Outcomes Achieved

Key Lessons Learned

The experiments conducted thus far on the Osoyoos Desert site have demonstrated that seeding technique and seed mix composition must be considered and matched to local environmental conditions. That being said, some techniques can aid in the establishment of species in environments they would not normally colonize.

Long-Term Management

This phase of the Osoyoos Desert Centre restoration plan focused on reducing the weedy species component and increasing the cover and recruitment rate of native grasses and forbs. Future efforts could be directed at replacing areas of early successional species with climax species and maximizing species diversity within each of the identified habitat types.

Sources and Amounts of Funding

Funding for this project has been provided by the Osoyoos Desert Society, Environment Canada, University of British Columbia, Human Resources Development Commission – Canada, and the Habitat Conservation Trust Fund for research on “Ecological Restoration and Habitat Renewal of the South Okanagan Shrub-steppe.”

Other Resources

Osoyoos Desert Centre

Primary Contact

Organizational Contact