Germination trials and determination of cold stratification requirements found that chilling for 6 weeks resulted in about 80% germination (St. Hilaire 1988). A three-year greenhouse experiment examined seed dormancy and germination requirements. Cold stratification was required for laboratory germination. In field and laboratory trials, germination rates were higher among year old seed than two year old seed, and no germination occurred among three year old seeds (Wentworth 2000).

Many populations have destroyed by the conversion of its native prairie habitat to agricultural, residential, and commercial uses. The decline of golden paintbrush is also correlated with fire suppression. Fire disturbance is an integral component of the prairie ecosystem, maintaining grassland by preventing the successional encroachment of woody shrubs and trees. As a direct consequence of these land-use changes, golden paintbrush has not been seen in Oregon for over 40 years and is now endangered in Washington.

A rare regional endemic, historically known west of the Cascades from southern British Columbia to central Oregon, but now extirpated from many of its historic localities, and known from only 11 populations. The extant populations face ongoing threats including development, herbivory by native animals and introduced rabbits, invasion of non-native plants, and habitat modification due to succession in the absence of fire.

Prairie destruction due to residential, commercial or agricultural use is a threat at five of the six privately owned sites (USFWS 2000). Encroachment by native species, often resulting from fire suppression: Pseudotsuga menziesii (Douglas fir), Pteri

Castilleja levisecta is known from eleven total sites, including nine in Washington and two on small islands in British Columbia (USFWS 2000): 1 - 4000+ flowering plants 3 - 1000+ flowering plants 3 - 100-1000 flowering plants 1 - <100 flowering plants 3 - unknown numbers, one of which may have been extirpated

Germination trials and determination of cold stratification requirements found that chilling for 6 weeks resulted in about 80% germination (St. Hilaire 1988). A three-year greenhouse experiment examined seed dormancy and germination requirements. Cold stratification was required for laboratory germination. In field and laboratory trials, germination rates were higher among year old seed than two year old seed, and no germination occurred among three year old seeds (Wentworth 2000). In greenhouse experiments, Castilleja levisecta was grown in pots with three suspected host plants: Fragaria virginiana, Chrysanthemum leucanthemum, and Festuca idahoensis. Castilleja levisecta did not require a host to survive and reproduce. These results suggest that C. levisecta is a facultative parasite. When conditions are optimal, it does not need the additional nutrients and so does not need to form parasitic connections (Wentworth 2000). Pollinator exclusion experiments between bagged and unbagged inflorescence revealed that fruit set was nearly 5 times greater among unbagged individuals (Wentworth 2000). A demographic study from 1991 to 1994 of over 800 individuals indicated that plants have the ability of grow quickly or regress from year to year. Regression to smaller stage was greater than growth and particularly common in large reproductive adults. This tendency towards regression complicates assessment of demographic trends. It also reveals that reproductive plants may not produce flowers one year but not the next year, decreasing reproductive potential. The medium stage class made the greatest proportional contribution to population growth, and implying that conservation efforts should focus on this stage (Wentworth 2000). Censuses of three populations in Washington all indicate rapid population declines over 10-20 years. Trends are difficult to discern in two other populations, but suggest that at least portions of the populations have undergone significant declines. The two populations in British Columbia appear to be stable over the short term. One last population in Washington shows no directional trend, but there are reports that plants used to grow in locations in this site where they are no longer found. It is likely this population suffered a long-term decline (Gamon et al. 2000). A lack of data from periods of significant population decline, and data too sparse to be certain of population fluctuations, presently make long-term viability predictions difficult. Analysis of available data reveals that sudden declines can occur in healthy populations within a few decades and in the absence of habitat destruction. Due to a general lack of data and little information on non-flowering individuals, little is known about the causes and extent of natural variability. Preliminary models suggest that no present Castilleja levisecta site has a sufficient number of plants to make it immune to drastic declines. Available data indicates that while some current populations will likely persist for 10-20 years without active management, there is a high probability that all populations are susceptible to decline. The species could go extinct within 10-50 years without aggressive steps to increase population size and establish new populations (Gamon et al. 2000). Three different burn-effect studies at one site in Washington determined that burning does not increase Castilleja levisecta mortality when averaged over several years. Plants that survived the burn tended to be of similar height and have comparable numbers of flowering stems as those plants in the unburned controls. Two of the three studies indicated that plants that establish following a burn had about a 20% higher survival rate. In one study, both vegetative and reproductive plants were more likely to flower the year following the burn. More seedlings were found in burned plots than in controls, but it is possible that the difficulty in finding seedlings in unburned vegetation skewed the results. Castilleja levisecta population size increased for a number of years following burning (Dunwiddie et al. 2000). Researchers digitally mapped unplowed grassland and oak dominated/co-dominated communities in the Puget Lowland and Willamette Valley Ecoregions. These maps were used to compare the existing distribution of these communities to coverage thought to exist prior to European settlement. Co-occurrence with three rare plant species was analyzed. Two of the three rare plants showed considerable correspondence with either the grassland or oak canopy communities. The remaining rare plant, Castilleja levisecta, had low co-occurrence with mapped vegetation and soil units. This low level of co-occurrence is likely due to a small sample size combined with scale issues. Many C. levisecta sites are smaller than the minimum map unit, and in areas that did not have any mapped prairie soils, reflecting small and ephemeral grasslands (Chappell et al. 2000).

Tree removal and prescribed burning at one Washington site and mowing at three other sites are used to control shrub encroachment. Most of these activities have only been undertaken for 2-3 years and none have resulted in large enough changes to be apparent in monitoring data (Gamon et al. 2000). Monitoring and management at Whidbey Island Naval Air Station. A large population is monitored and managed by The Nature Conservancy (Pratt Preserve). Sites in British Columbia are in designated ""Ecological Reserve"" land. Entry is restricted and plant collection and resource destruction are not allowed (USFWS 2000).

Determine breeding system and pollinators (Evans et al. 1984). Calculate recruitment rates and investigate effect of host presence on seedling survival (Wentworth 2000) Determine minimum viable population size (Wentworth 2000). Annual or biannual monitoring in order to create more reliable models and determine the causes and extent of natural population fluctuations. When individual plants can be distinguished, censuses should include all flowering plants. It is recommended that the total number of flowering stems be tallied at the same time as individual plants are counted. Additionally, the number of seedlings and vegetative plants should be accounted for by estimates based on careful searches in randomly placed, small scattered plots (Gamon et al. 2000) Long term experiments incorporating cohorts of tagged individuals are needed to clarify how Castilleja levisecta responds to burns under differing conditions. The effect of burn timing and frequency also needs to be explored (Dunwiddie et al. 2000). Monitoring to determine if plants germinate from seed banks following burns. Also, sowing seed into burned areas to investigate if post-burn conditions enhance germination/survival would assist restoration efforts (Dunwiddie et al. 2000). Castilleja levisecta will be considered for de-listing when there are at least 20 stable populations distributed throughout the historic range. To be considered stable, a population must maintain a 5 year running average population size of at least 1000 individuals. At least 15 of these populations must be located on protected sites (USFWS 2000). Conduct analysis of genetic variation within and among populations (USFWS 2000). Study pollination biology and long term seed viability (USFWS 2000).

Collect and store seeds from all populations across the species' range. Determine optimum germination requirements. Determine propagation and re-introduction protocols.