Serpentine Vegetation Management Project

2005 REPORT

Stuart B. Weiss, Ph.D. and David H. Wright, Ph.D. Creekside Center for Earth Observation 27 Bishop Lane, Menlo Park, CA 94025 tel: 650-854-9732, fax: 650-644-3355 e-mail: [email protected] http://www.creeksidescience.com

FWS Grant Agreement No. 814205G240

Weiss and Wright 2005 Serpentine Vegetation Management Report

Cover photo: David Wright trimming a 2-inch early trim, grazing exclusion plot on March 15, 2005. The PVC quadrat shows the size of the sample area; an additional 2 dm on all sides of the plot are trimmed to minimize edge effects. The wood blocks are temporarily placed to standardize the height of cut. The wooden stake at upper left marks and identifies the plot. The fence in the background excludes cattle grazing.

CONTENTS: Page

a. Abstract iii

b. Introduction 1

c. Description of study area 3

d. Methods and materials 3

e. Results and Discussion 5

f. Summary and conclusions 14

g. Literature cited 15

h. Summary of expenditures 15

h. Data appendices CD included

ii Weiss and Wright 2005 Serpentine Vegetation Management Report

ABSTRACT

We have begun a 3-year project to evaluate vegetation management methods and grazing impacts on serpentine habitat in Santa Clara County. In the first year of our project, funded by Habitat Restoration Program grant 814205G240, we established an experimental vegetation management trial, collecting initial vegetation data from that trial, conducted observations on serpentine vegetation characteristics across different grazing and other vegetation management regimes (fenceline and burn study), investigated herbivore damage to setchellii, and recorded the spatial distribution and other observations on Streptanthus albidus peramoenus and S. a. albidus. Developing vegetation management information and applying it in adaptive land management are priority tasks for recovery of the bay checkerspot butterfly, Santa Clara Valley dudleya, Metcalf Canyon jewelflower, and other species.

Vegetation management trial: Cattle exclusion fence and seven vegetation-management treatments have been successfully established and maintained. Initial percent cover data for all species present have been collected and archived. Visible differences among treatments are emerging. Plantago erecta densities were very low in spring 2005, but we expect future increases. This trial will continue for two additional years.

Grazing and burn comparisons: We found that ungrazed areas supported little suitable habitat for Bay checkerspot butterflies, as they were dominated by annual grasses and provided little host plant and nectar resources. Annual grass cover was lower in the Winter-Spring than in the Summer–Fall regime, but both supported similar densities of Plantago and had adequate nectar sources to support dense Bay checkerspot populations. A 2004 late-spring wildfire on Tulare Hill allowed for higher Plantago cover and lower annual grass cover. The major difference between burned areas that had been grazed and those that had not was reduced annual grass cover in the grazed areas.

Dudleya setchellii and Streptanthus albidus: Herbivory on dudleya appears strongly dependent on location, i.e. on local populations and activity of herbivores, including the timing of their activity. Animals that damage Santa Clara Valley dudleya appear to include cattle and rodents (ground squirrels, gophers) or rabbits. In some locations damage by cattle may be significant; in other areas it appears not to be. Some of our observations suggest a benefit to some dudleya populations from limited grazing, possibly through reduction of competition for light, space or other resources, or through reduced cover for small herbivores. The subspecies S. a. peramoenus, and probably the endangered subspecies albidus as well, appears to be somewhat flexible in the face of moderate damage simulating herbivory. On the other hand, naturally occurring severe damage was seen in the field. More work is planned quantifying response of the most- beautiful jewelflower to rather severe damage. We found its reproductive effort to be strongly related to plant size, setting the stage for such a response. Bombus melanopygus was documented to readily visit S. a. peramoenus. There is a transition zone between the two subspecies near the Field Sports Park, with mixed populations of pink (ssp. peramoenus), white (ssp. albidus), and intermediate flower colors across a 2 km transect.

iii Weiss and Wright 2005 Serpentine Vegetation Management Report

INTRODUCTION

We have begun a project to evaluate vegetation management methods and grazing impacts on serpentine habitat in Santa Clara County – funded in the first year by the Habitat Restoration Program. Developing this information and applying it in adaptive land management are priority tasks for recovery of the bay checkerspot butterfly, Santa Clara Valley dudleya, Metcalf Canyon jewelflower, and other species. Work is in progress atop Coyote Ridge, in a conservation area in the Kirby recovery unit for bay checkerspot butterfly, and at other significant serpentine preserves in the county. We applied to the CVPCP/HRP and were approved for funding to continue this project for two additional years (2006-2007).

Serpentine grasslands in Santa Clara County support multiple listed and rare species, including the bay checkerspot butterfly (Euphydryas editha bayensis), Santa Clara Valley dudleya (Dudleya setchellii), Metcalf Canyon jewelflower (Streptanthus albidus albidus), and Opler’s longhorn moth (Adela oplerella). Table 1 provides a list of species of interest that may benefit from this project. Many are endemic to serpentine sites, and the federally listed species are endemic to Santa Clara County or nearly so.

A critical need in order to recover species and restore habitat on serpentine soils in Santa Clara County is to build adaptive vegetation management strategies. This need is identified in the USFWS 1998 final Recovery Plan for Serpentine Soil Species of the as Priority 1 tasks, number 5.16 and 5.24 (p. IV-27 to 28, 33, 35), and the results of strategy development are needed for habitat management of preserved serpentine sites, Priority 1 task 3.1 (p. IV-16) (USFWS 1998). Serpentine soil habitats supporting endemic species in Santa Clara County consistently have been identified by the CVPCP and HRP as a high priority for attention and funding in recent years. Our ultimate objective is to provide land managers with information they need on serpentine vegetation response to management alternatives, so that they can more effectively implement their own vegetation management programs for conserving rare species.

In the first year of our project we established an experimental vegetation management trial, collecting initial vegetation data from that trial, conducted observations on serpentine vegetation characteristics across different grazing and other vegetation management regimes (fenceline and burn study), investigated herbivore damage to Dudleya setchellii, and recorded the spatial distribution and other observations on Streptanthus albidus peramoenus and S. a. albidus.

1 Weiss and Wright 2005 Serpentine Vegetation Management Report

Table 1. Special Status Species and Species of Interest in the Area of the Project

Common Name Scientific Name Status1 Comments bay checkerspot Euphydryas editha FT largest core population, butterfly bayensis critical habitat

Opler=s longhorn Adela oplerella FSC moth

Santa Clara Valley Dudleya setchellii FE, CNPS rocky serpentine areas dudleya 1B

Metcalf Canyon Streptanthus albidus FE, CNPS white flowers jewelflower albidus 1B

Amost beautiful@ Streptanthus albidus FSC, pink-purple flowers jewelflower peramoenus CNPS 1B smooth lessingia Lessingia micradenia FSC, var. glabrata CNPS 1B serpentine linanthus Linanthus ambiguus CNPS 4 dwarf plantain Plantago erecta food plant of bay checkerspot purple owl=s clover Castilleja densiflora food plant of bay checkerspot exserted paintbrush Castilleja exserta food plant of bay checkerspot

California cream cups Platystemon food plant of Opler’s californicus longhorn moth tidy-tips Layia platyglossa nectar plant, bay checkerspot

California goldfields Lasthenia californica nectar plant, bay checkerspot desert-parsley Lomatium spp. nectar plant, bay checkerspot scytheleaf onion Allium falcifolium nectar plant, bay checkerspot sea muilla Muilla maritima nectar plant, bay checkerspot false babystars Linanthus androsaceus nectar plant, bay checkerspot intermediate Amsinckia intermedia nectar plant, bay checkerspot fiddleneck

1FE=federally endangered, FT=federally threatened, FSC=federal species of concern. CNPS codes refer to California Native Plant Society lists.

2 Weiss and Wright 2005 Serpentine Vegetation Management Report

DESCRIPTION OF STUDY AREA

All work was conducted in serpentine grassland habitats in Santa Clara County, California, USA. The vegetation is essentially treeless, dominated by grasses and forbs, with unusually high levels of native and endemic species relative to non-serpentine California grasslands. The soils of our study areas typically are mapped as Montara series, a distinctive serpentine series in Santa Clara County. A general description of this type of habitat can be found in the introduction of the Recovery Plan for Serpentine Soil Species of the San Francisco Bay Area (USFWS 1998).

We established our vegetation management trial on Coyote Ridge, in the Kirby Conservation Area north of the Kirby Canyon Landfill. The location of the vegetation management trial is on top of the ridge several hundred meters north-northwest of the intersection of the conservation area access road with the ridge road; approximately UTM Zone 10, 617,600 easting, 4,117,500 northing (WGS84 datum; USGS Morgan Hill 7.5” quad). The area is grazed in winter and spring (rancher: Tony Pierce), with some oversight by Waste Management, Inc. consultants to ensure good habitat conditions for bay checkerspot. This grazing generally results in low vegetation heights and low residual dry matter that favor the butterfly’s low-stature host . Most-beautiful jewelflower studies were carried out in part along the Kirby Conservation Area access road, in serpentine slopes, gravels, and outcrops.

Fencelines in serpentine grassland in the same general area (Kirby) were used as part of the fenceline study and the dudleya herbivory observations. Ungrazed areas behind certain fencelines at Kirby have been ungrazed since 1986. Ungrazed areas at Tulare Hill have been fenced off since 2001. Fencelines separating different grazing management regimes (winter-spring vs. summer) also are present at Kirby.

Tulare Hill, rising from the Santa Clara Valley between Coyote Ridge and the Santa Teresa Hills, was another important study area (USGS Morgan Hill and Santa Teresa Hills quads). Vegetation studies, including the burn comparisons, and dudleya herbivory observations were conducted there. The burned area on Tulare Hill experienced an uncontrolled burn in late May 2004.

MATERIALS AND METHODS

Vegetation management trial

Seven treatments were applied in a randomized block design. Two grazed treatments were: (1) ambient grazing (winter-spring), and, (2) partial grazing exclusion (fencing adjusted to try to reach 4-inch vegetation height). Within complete grazing-exclusion fencing were the five remaining treatments: (3) untrimmed, (4) trimmed to 2 inches early (winter-spring), (5) trimmed to 2 inches late (summer season), (6) trimmed to 4 inches early, and, (7) trimmed to 4 inches late

3 Weiss and Wright 2005 Serpentine Vegetation Management Report

We fenced the grazing exclusion plots (Figure 1) in late fall of 2004 before substantial annual plant growth, and added the partial grazing exclusion fencing on March 15, 2005, when cattle were present. Two separate blocks (each about 40 by 40 feet) of plots were established to avoid pseudo-replication. Each half-meter by half-meter plot was staked and numbered. In the grazing exclusion area, plots were randomly assigned to vegetation management treatments.

Our first trimming treatments took place on March 15, 2005, with the trimming of the simulated early-grazing plots, with subsequent trimmings on April 14 and May 16, to maintain target vegetation height. Simulated late-grazing plots were trimmed on July 27, after most grass seed heads had shattered and dispersed their seed. We trimmed plots using battery-powered hedge-trimmers (cover photo), and trimmed 20 cm outside each plot to reduce edge effects.

We quantified plant percentage cover in each plot by plant species, and also recorded percent cover of bare ground and rocks. Plant cover was recorded on April 13-14, when many species were flowering and most recognizable. Cover estimates used a scale of: 0, 1, 2, 5, 10, 20, 30, 40, …, 100%.

Cross-fenceline vegetation study

We established transects to sample species composition across different grazing regimes. Transects were 50 m long, and ten 0.5 x 0.5 m quadrats spaced 5 m apart were sampled for percent cover of plant species and bare ground, as described above. Grazing regimes included ungrazed, winter-spring, and summer-fall. All data were log10 transformed for one-way ANOVA. Tabular data report untransformed values.

Dudleya herbivore damage

We quantified damage to Dudleya plants across fencelines separating different grazing regimes at four locations, and measured plants within a recent burn area in one location (Figure 3). The samples on either side of the fences were within 50 m of one another in visually similar habitat; and plants within 1 m of the fence were not included, to avoid areas of potential heavy trampling by cattle. We counted and measured inflorescences, and counted flowers on selected inflorescences to relate inflorescence size to reproductive potential. Damage to inflorescences was recorded, and observations of evidence of potential herbivore activity were made.

Streptanthus studies

One of us (SBW), with Allison Green (UC Berkeley, Department of Environmental Science, Policy, and Management), surveyed the distribution of white-flowered (the endangered Metcalf Canyon jewelflower, Streptanthus albidus albidus) and pink- flowered (most-beautiful jewelflower, Streptanthus albidus peramoenus) plants along Coyote Ridge, from north of Metcalf Canyon to the Kirby Canyon Landfill. At each site,

4 Weiss and Wright 2005 Serpentine Vegetation Management Report

a minimum of 50 plants were examined for flower color and other characters and classified as “white,” “pink,”, or “intermediate.”

We began a clipping experiment on S. a. peramoenus along the Kirby ridge access road, to examine its response to simulated herbivore damage. Quartets of plants near each other in similar microhabitat were identified and tagged, and each plant randomly assigned to one of four clipping treatments: no clipping; 1/3 of leaves clipped at base; 2/3 of leaves clipped; and 2/3 of leaves plus apex (apical meristem) clipped. While this experiment did not yield statistical data because many of the plant tags were lost (strings disintegrated or were eaten), useful information was discovered (see Results). We plan to repeat the experiment with modifications (e.g., aluminum tags/ties!) next year.

We made general observations of Streptanthus albidus peramoenus reproductive biology, including pollinators, fruit capsule number, sizes and number of seeds. After counting, seeds were given to the Santa Clara Valley chapter of the California Native Plant Society for their work in educating people about native plants and native plant gardening.

RESULTS AND DISCUSSION

Vegetation Management Trial

Grazing exclusion fencing was successfully established and has held up well. At the end of the growing season, vegetation in untrimmed portions of the grazing exclusion trial area was noticeably taller than surrounding grazed areas, with dense growth of the non- native grass Lolium multiflorum. Vegetation height in trimmed plots was markedly lower than ungrazed areas (Figure 1). Periodically opening or closing the fencing of the partial grazing exclusion areas worked fairly well, resulting in intermediate vegetation removal, though it was difficult to control to achieve a precise vegetation height.

One immediate result observed in trimming treatments was that early trimming removed some of the flowering stalks of several taller plant species, including Muilla maritima, Layia platyglossa, Sanicula bipinnatifida, and Lolium. Some flowers/inflorescences do not appear to be grazed by cattle (e.g., Muilla), so this eventually could result in differences between our trimming and grazing treatments. Grazing cattle do remove inflorescences of Lolium.

Densities of the bay checkerspot host plant Plantago erecta were unusually low compared to past years in the area where we established our trial plots, averaging less than 1 % cover. Patches of low Plantago densities were observed across Coyote Ridge (SBW, pers. obs.). One possible explanation is that extremely high densities of bay checkerspot larvae (>1/m2) in this area in previous years decimated the crop of the annual Plantago in 2005. Other anecdotal reports exist of unusually high densities of bay checkerspot larvae apparently reducing their host Plantago. We anticipate increasing Plantago densities in 2006 and 2007.

5 Weiss and Wright 2005 Serpentine Vegetation Management Report

Figure 1. Vegetation inside the grazing exclusion fencing was visibly higher toward the end of the growing season (July 27, 2005; compare vegetation in background). Trimmed trial plots are visible as shorter patches amid the high grass.

With such low Plantago densities, we did not detect any differences among vegetation treatments. Next spring will be the first growing season where germinating Plantago and other native plants will be subjected to tall residual vegetation and thatch due to trial vegetation treatments, and we may expect more changes in the coming season. In reviewing the data, however, we did discover significant differences in Lolium percent cover in different areas. The west grazing exclusion block had more plots with Lolium absent, resulting in a lower average cover of Lolium than in the east block (west: 13%; east: 53%). This difference between blocks is fine, because it allows us to investigate a wider range of initial conditions and generalize our results to a wider range of sites. The ambient grazing and partial grazing plots also had higher Lolium cover (55% and 53%, versus 20% in all grazing exclusion plots). Since these plots are spatially separate from the grazing exclusion plots (necessarily outside the fence), they are readily subject to spatial differences in vegetation cover. They could not be completely randomized with the grazing exclusion plots and are not strictly part of that experimental design, but they are intended as a qualitative comparison to the 2” trimmed and 4” trimmed plots. In one sense this raises the bar for the untrimmed, ungrazed plots: if they become grassier than the ambient grazing plots – as we expect they will -- despite having started less grassy,

6 Weiss and Wright 2005 Serpentine Vegetation Management Report

our confidence in the importance of the observed effect of grazing exclusion will be even stronger. The differences in Lolium cover may sound considerable, but cover of this non- native annual grass varies widely from year to year, and responds very rapidly to changes in vegetation management.

No bay checkerspot larvae were observed during any project activities to date, and none were directly impacted. We observed bay checkerspot adults flying in the project vicinity on Coyote Ridge in the Kirby habitat area during field work on March 14 and continuing to May 16, 2005, a rather extended flight season.

Cross-Fenceline Vegetation Study

Grazing regimes had substantial effects on plant composition (Table 2). Plantago did not differ between under Winter-Spring (WS) and Summer-Fall (SF) grazing, but was substantially lower in the ungrazed (UG) areas. Castilleja of either species did not differ among grazing regimes, but these plants have a highly patchy distribution, and was totally absent from the ungrazed areas. Lasthenia, Layia, and Muilla – all bay checkerspot nectar plants – were highest in WS, followed by SF, and lowest in UG. Lolium and total annual grass cover were highest in UG, and lowest in WS. Species richness at the quadrat level was highest in WS, followed by SF, and lowest in UG.

Table 2. Key plant species responses to grazing and different grazing regimes. Means followed by different letters within rows are significantly different by Tukey-Kramer HSD Winter-Spring Summer-Fall Ungrazed Probability Plantago erecta 7.9 a 7.5 a 4.1 b 0.004 Castilleja exserta 0.15 0.06 0.04 n.s. Castilleja densiflorus 0.04 0.04 0 n.s. Lasthenia californica 8.8 a 3.1 b 0.9 c <0.0001 Layia platyglossa 0.7 a 0.5 a 0 a 0.01 Muilla maritima 0.6 a 0.2 b 0.1 b <0.0001 Allium falcifolium 0.04 0.02 0.04 n.s. Lolium multiflorum 11.5 a 34.5 b 48.0 c <0.0001 Annual Grass 23.8 a 44.8 b 64.3 c <0.0001 Species Richness 12.4 a 9.3 b 6.4 c <0.0001

The effects of the burn on Tulare Hill were substantial (Table 3). Plantago, Castilleja densiflorus, and species richness were higher in the burned area. Lolium and total annual grass cover were higher in the unburned areas, and the other nectar sources did not differ. The grazed areas that burned had lower Lolium and total annual grass cover than the ungrazed-burned areas. Ungrazed-burned areas had higher Castilleja cover than the grazed-burned areas.

7 Weiss and Wright 2005 Serpentine Vegetation Management Report

Table 3. Effect of burning and grazing on Tulare Hill. The unburned area was ungrazed; not enough grazed area remained unburned for comparison. Means followed by different letters within rows are significantly different by Tukey-Kramer HSD. Unburned Burn-Ungrazed Burn-Grazed Probability Plantago 2.1 a 15.8 b 16.7 b <0.0001 Castilleja densiflorus 0 a 0.62 b 0.21 a <0.0001 Lasthenia 1.5 2.6 4.1 n.s. Muilla 0.17 0.05 0.35 n.s Allium 0.07 0.62 0.17 n.s. Lolium 48.4 a 18.8 b 6.6 c <0.0001 Annual Grass 72.1 a 33.3 b 24.5 c <0.0001 Species Richness 6.8 a 10.4 b 9.8 b <0.0001

Dudleya Herbivore Damage

Substantial numbers of inflorescences of the endangered Santa Clara Valley dudleya showed damage in the July 27-29 sampling, many apparently from herbivory (Figure 2). The average rate of damage of inflorescences across all sites was approximately 40% (n = 778). We observed evidence of several herbivores potentially contributing to the damage in the immediate vicinity of the dudleya plants: cattle, ground squirrels, pocket gophers, and leporids (hares and rabbits). Other species, such as woodrats, voles, deer, and elk cannot be excluded. The location of sampling sites is shown in Figure 3.

Figure 2. A dudleya cluster in middle upper right shows numerous inflorescences severed at approximately the same level. To the left, against the rock outcrop, is a single intact inflorescence, and another from the same base has fallen over and is lying to the far left. There are a number of dudleya rosettes in the picture that lack any inflorescence. The yellow strip is 17 cm long.

8 Weiss and Wright 2005 Serpentine Vegetation Management Report

Figure 3. Project area showing dudleya sampling locations, in red. The map spans the Santa Clara Valley between San Jose and Morgan Hill; USGS 7.5” quadrangle names are shown in yellow. Scale: from side to side the map portrays an area approximately 10.5 km wide. Source: http://casil.ucdavis.edu/mapsurfer/ .

Table 4 shows the frequency and proportion of broken inflorescences on grazed and ungrazed sides of fencelines. In two of three comparisons where testing was possible (chi-squared tests), the proportion damaged was significantly different on opposite sides of the fence, and in the third case the difference was suggestive.

Table 4. Broken vs. intact dudleya inflorescences across fencelines at 4 sites

Kirby Middle Kirby 2nd Gate Intact Broken Intact Broken Ungrazed 54 39 (42%) Ungrazed none found Grazed 41 16 (28%) Grazed 80 59 (42%)

Χ2 = 2.92 df=1 , P~0.09

Kirby Lower Tulare Hill Intact Broken Intact Broken Ungrazed 73 68 (48%) Ungrazed 49 7 (12.5%) Grazed 70 29 (29%) Grazed 14 91 (87%)

Χ2 = 8.65 df=1 , P<0.005 Χ2 = 84.3 df.=1 , P<0.001

However, the direction of the difference varied with location. At both Kirby locations there was less damage on grazed side, while at Tulare Hill there was more damage on the grazed side. Both sites are grazed winter-spring, but there might be slight differences in how late cattle have access to the sites we sampled. Dudleya inflorescences develop and

9 Weiss and Wright 2005 Serpentine Vegetation Management Report

may be attractive as forage in late spring. We observed patterns of damage suggestive of cattle or ungulate grazing (e.g., clusters of inflorescences all severed at the same level with a similar angle of cut; Figure 2, above). Other damage was consistent with small mammals, and we observed rodent burrows among rock outcrops, pocket gopher mounds and pocket gopher activity in soil around rock outcrops, ground squirrel or rabbit trails, and rabbit scat. We did observe some physically caused breakage, e.g., by wind, usually leaving the inflorescence lying nearby, but there was no reason to expect such breakage to differ across fencelines.

Did herbivore damage actually impair dudleya reproduction? Figures 4 and 5 show that broken inflorescences usually bore no flowers, whereas most intact inflorescences had more than 10 flowers and the average was between 25 and 35 flowers. We looked for but saw no evidence of branching of inflorescences below the break, or of new inflorescences developing to replace broken ones. The plants are perennial, however, and might compensate somewhat with additional reproductive effort in a future year.

16 160

14 140

12 120

10 100

8 80

6 60

4 40 Number ofInflorescences

Number of Inflorescences 2 20

0 0 0204060 0 0 0 0 0 0 0 10 20 3 4 5 6 70 8 9 Number of Flowers Num be r of Flow e r s More

Figure 4. Histogram of number of flowers on Figure 5. Histogram of number of flowers on intact inflorescences in 2005. The frequency broken inflorescences. The great majority bore axis indicates the number of inflorescences no flowers; the average number was 1.1 bearing the number of flowers on the x-axis. flowers. This figure is based on a non-random sample used to construct Figure YY, and over- represents large and small inflorescences, but the central tendency is clear (average between 25 to 35 flowers per intact inflorescence), as is the contrast with the much reduced reproductive potential of damaged inflorescences (Figure Xx/2).

Larger inflorescences produced more flowers (Figure 6), and our visual observations indicated they also produced more fruit. We did not dissect fruits to count seeds.

10 Weiss and Wright 2005 Serpentine Vegetation Management Report

Dudleya setchellii : no. of flowers vs. inflor. length, 2005

90

80 R2 = 0.918 70

60

50

40

30 Number of Flowers 20

10

0 0 5 10 15 20 Inflorescence Length (cm)

Figure 6. Flower production increased significantly with increasing inflorescence length in Dudleya setchellii. The regression equation is log10(n flowers) = 1.1652*log10(inflor cm) + 0.3752.

We measured inflorescence lengths across fencelines and in a burned area in two locations, at Kirby and at Tulare Hill (Table 5). Inflorescences were significantly smaller on the grazed side at Tulare Hill, and did not appear adversely affected in the burn area. There was no significant difference in inflorescence length on grazed and ungrazed sides of the fence at the Kirby middle site. The difference at across the Tulare Hill fenceline corresponds, according to the equation of Figure 6, to a difference of about 9 flowers per inflorescence (23 grazed vs. 32 ungrazed) – potentially a reduction of about 28% in reproductive output.

Table 5. Dudleya inflorescence length comparisons (average, confidence interval, in cm)

Kirby Middle Tulare Hill Average (95% C.I.) Average (95% C.I.) Ungrazed 7.3 (6.33 – 8.18) Burn 10.0 (8.97 – 10.94) Grazed 8.1 (7.02 – 9.22) Ungrazed 9.3 (8.32 – 10.33) Grazed 6.9 (5.76 – 8.02) Ungrazed vs. grazed t-test: P = 0.24* Ungrazed vs. grazed t-test: P = 0.0033*

* two-tailed, unequal variances

11 Weiss and Wright 2005 Serpentine Vegetation Management Report

At one fenceline sample location, we found no dudleya on the ungrazed side in apparently similar soil, slope and aspect (Kirby 2nd Gate; see Table 4, above). While it is possible that some abrupt but invisible physical change coinciding with the fenceline is responsible for this sharp break in the dudleya population, the spatial pattern is suspicious. We suggest the possibility that the grazing regime at this location has some benefit to dudleya rosette establishment or maintenance. Potential mechanisms for such an effect include: reduced competition for light, space, water, nutrients, or other resources from grasses and other plants in the grazed area; increased germination gaps due to grazing or trampling disturbance; or even reduced small mammal herbivory due to reduced cover making small mammal populations vulnerable to predation, or making them reluctant to forage due to lack of cover and therefore less active. We plan to survey additional locations in the future to see if this pattern is repeated elsewhere.

Streptanthus Studies

The surveys identified a transition zone between ssp. albidus and ssp. peramoenus around the Santa Clara County Field Sports Park (see Figure 3). All plants north of Metcalf Canyon were white (ssp. albidus). Between Metcalf Canyon and the Field Sports Park, most plants were white, but there were a substantial fraction of intermediate plants and a few that ranged into pink. South of the Field Sports Park, the ratios were reversed, and white plants were relatively rare. At Kirby Canyon, virtually every plant was pink, with only a couple of individuals intermediate:

White Intermediate Pink Metcalf North 100.00% 0.00% 0.00% Field Sports North 80.36% 15.36% 4.29% Field Sports South 7.17% 20.77% 72.06% Kirby Canyon 0.00% 0.10% 99.90%

In observations associated with the preliminary Streptanthus albidus peramoenus clipping experiment (see Methods), we learned that the most-beautiful jewelflower appears to be flexible in the face of physical damage. The plants readily branch and, unlike the dudleya, send out new inflorescence branches in response to loss of the apex. Stems and leaves are similarly colored, the leaves are rather small and sparse, and the stems may be photosynthetically active, all of which mean that clipping even two-thirds of the leaves is not a very severe event for S. a. peramoenus. On the other hand, we observed that jewelflower stems nipped off essentially to ground level – by whatever herbivore – were somewhat common in the field at Kirby. As in the case of dudleya, the cast of potential culprits among herbivores is quite large, but this kind of severe damage did occur in ungrazed and extremely steep areas (presumed inaccessible to cattle) as well as in grazed areas.

We also observed an insect herbivore – or seed predator – on most-beautiful jewelflower: a small green moth larva (possibly Geometridae) living within the seed capsules and apparently feeding on the seeds.

12 Weiss and Wright 2005 Serpentine Vegetation Management Report

We plan to repeat a clipping experiment to simulate herbivory and measure effects on growth, survival, and reproductive output. The follow-up experiment will use aluminum plant tags, more extreme clipping (including a near-ground clipping treatment), and will stratify by initial plant size rather than in groups of nearby plants, as initial size appears a better correlate of relative growth rate than proximity.

Capsule measurements and seed counts were conducted on the 10 plants harvested as part of the clipping experiment. These plants were growing close to the dirt access road at Kirby and may not have typical soil or water conditions, and so their size should not be taken as representative of the subspecies – nor was the growing season over for all plants at the time of harvest. However, these plants usefully demonstrate relationships between size (height) and fruit production (Figure 7), and between fruit size and seed number (Figure 8) such as are typical of many plants, especially those like Streptanthus with indeterminate growth. There was no correlation between plant size and fruit size (P>0.5, n = 9). The average number of seeds per fruit was 34.3 (s.d. = 13.47, n = 47 siliques from 9 plants). Combining fruit production and seeds per fruit, larger plants have greater seed output than smaller plants.

90 70 80 60 y = 0.7553x - 17.313 70 R2 = 0.8757 50 60 50 40 40 30 30 20 20 Number of Seeds

Number of capsules Number 10 10 0 0 0 20406080100 20 40 60 80 100 Height (cm) Silique Length (mm)

Figure 7. Fruit production vs. height of 10 Figure 8. Number of seeds per fruit vs. fruit Streptanthus albidus peramoenus plants in size in 47 fruits (siliques) from 9 Streptanthus 2005. There was a significant positive albidus peramoenus plants. Number of seeds relationship (Spearman rank correlation = 0.68, appears strongly and essentially linearly related P~0.02 two-tailed). to fruit length.

In addition to these contributions to the reproductive biology of Streptanthus albidus peramoenus, on June 6, 2005, along the Kirby Conservation Area access road between the first (west) and second gates, one of us (DHW) observed several bumblebees busily visiting a large number of open S. a. peramoenus flowers. One was captured and identified as Bombus melanopygus (“edwardsii” color form [black rather than reddish hair on abdominal tergites 2 and 3]; thanks to Dr. Robbin Thorp, UC Davis, for help confirming the i.d.). Bumblebees often are considered to be effective pollinators. Thorpe et al. (1983) reported B. “edwardsii” visits species in the Streptanthus, as well as Platystemon and Ceanothus (the endangered Ceanothus ferrisae occurs in this vicinity). Bombus vosnesenskii also is reported to visit Streptanthus. Conservation of S. a.

13 Weiss and Wright 2005 Serpentine Vegetation Management Report

peramoenus, and S. a. albidus by extension, may well depend on adequate conservation of surrounding nesting and foraging habitat for pollinators such as B. melanopygus.

SUMMARY AND CONCLUSIONS

Vegetation management trial: Cattle exclusion fence and seven vegetation- management treatments have been successfully established and maintained. Initial percent cover data for all plant species present have been collected and archived. Visible differences among treatments are emerging. Plantago erecta densities were very low in spring 2005, but we expect future increases. This trial will continue for two additional years.

Grazing comparisons: Ungrazed areas supported little suitable habitat for Bay checkerspot butterflies, as they were dominated by annual grasses and provided little hostplant and nectar resources. Annual grass cover was lower in the Winter-Spring than in the Summer–Fall regime, but both supported similar densities of Plantago and had adequate nectar sources to support dense Bay checkerspot populations.

Burn comparison: The late-spring burn on Tulare Hill allowed for higher Plantago cover and lower annual grass cover. The major difference between burned areas that had been grazed and those that had not was reduced annual grass cover in the grazed areas.

Dudleya setchellii: Herbivory on this species appears to be strongly dependent on location, i.e. on local populations and activity of herbivores, including the timing of their activity. Animals that damage Santa Clara Valley dudleya appear to include cattle and rodents (ground squirrels, gophers) or rabbits. In some locations damage by cattle may be significant; in other areas it appears not to be. Some of our observations suggest a benefit to some dudleya populations from limited grazing, possibly through reduction of competition for light, space or other resources, or through reduced cover for small herbivores.

Streptanthus albidus: There is a transition zone between the two subspecies near the Field Sports Park, with mixed populations of pink-flowered (ssp. peramoenus), white- flowered (ssp. albidus), and intermediate colors across a 2 km transect.

The subspecies S. a. peramoenus, and probably the endangered subspecies albidus as well, appears to be somewhat flexible in the face of moderate damage simulating herbivory. On the other hand, naturally occurring severe damage was seen in the field. More work is planned quantifying response of the most-beautiful jewelflower to rather severe damage. We found its reproductive effort to be strongly related to plant size, setting the stage for such a response. Bombus melanopygus was documented to readily visit S. a. peramoenus.

14 Weiss and Wright 2005 Serpentine Vegetation Management Report

LITERATURE CITED

Thorp, R.W., D.S. Horning, Jr., and L.L. Dunning. 1983. Bumble bees and cuckoo bumble bees of California (Hymenoptera, Apidae). Bulletin of the California Insect Survey Vol. 23. University of California Press, Berkeley, 79 pp.

USFWS. 1998. Recovery Plan for Serpentine Soil Species of the San Francisco Bay Area. United States Fish and Wildlife Service, Sacramento, California.

Weiss, S.B. 1999. Cars, cows, and checkerspot butterflies: nitrogen deposition and management of nutrient-poor grasslands for a threatened species. Conservation Biology 13:1476-1486.

CONTACT INFORMATION

Stuart Weiss 27 Bishop Lane Menlo Park, CA 94025 (650) 854-9732 [email protected]

David Wright 1573 49th Street Sacramento, CA 95819 (916) 739-8906 (425) 696-3265 fax [email protected]

EXPENDITURE SUMMARY

Allocated funds for the project were $32,400.

Personnel: $30,075.00 Travel : $ 1,849.71 Supplies : $ 382.95

No major property was purchased during the project.

15 Weiss and Wright 2005 Serpentine Vegetation Management Report

DATA APPENDICES

Files provided in electronic format on CD.

16