Species Status Assessment for San Clemente Island Lotus

Species Status Assessment Report for the San Clemente Island Lotus (Acmispon dendroideus var. traskiae)

Version 1.0

Photo courtesy of USFWS

March 2020

U.S. Fish and Wildlife Service Pacific Southwest Region Sacramento, CA ACKNOWLEDGEMENTS

This document was prepared by the Texas A&M Natural Resources Institute in cooperation with the U.S. Fish and Wildlife Service and the United States Navy as part of the Service’s San Clemente Island Species Status Assessment Team.

We would also like to recognize and thank the following individuals who provided substantive information and/or insights for our SSA: Kimberly O’Connor, Bryan Munson, Melissa Booker, Dawn Lawson, Sula Vanderplank, Sandy Vissman, and Andrew Bridges.

Additionally, valuable input into the analysis and reviews of a draft of this document were provided by Mitchell McGlaughlin. We appreciate his input and comments, which resulted in a more robust status assessment and final report.

Suggested reference:

U.S. Fish and Wildlife Service. 2020. Species status assessment report for the San Clemente Island Lotus (Acmispon dendroideus var. traskiae), Version 1.0. March 2020. Sacramento, CA.

SSA Report – San Clemente Island Lotus ii March 2020

EXECUTIVE SUMMARY

This Species Status Assessment (SSA) provides an analysis of the overall species viability for the San Clemente Island Lotus (Acmispon dendroideus var. traskiae). To assess the viability of this taxon, we, the U.S. Fish and Wildlife Service (Service), used the conservation biology principles of resiliency, redundancy, and representation (3 Rs). Specifically, we identified the taxon’s ecological requirements and resources needed for individual survival and reproduction. We described the stressors (threats) influencing these resources and evaluated current levels of population resiliency and taxon redundancy and representation using available metrics to forecast the ability of this taxon to sustain populations into the future. Acmispon dendroideus var. traskiae is a semi-woody, flowering subshrub, endemic to San Clemente Island (SCI) and is one of five taxa in the genus Acmispon found on the island. There are no other varieties of A. dendroideus found on the island. At listing, nonnative herbivores were the primary threat to Acmispon dendroideus var. traskiae. As a result of their removal, habitat conditions improved and led to changes in the cover of native and nonnative plants on the island, further evidenced by the increases in A. d. var. traskiae and several other threatened and endangered taxa since the feral animals were removed. In the absence of the primary threat, additional threats to A. d. var. traskiae that have been identified include: (1) land use, (2) erosion, (3) nonnative plants, and (4) fire. Additionally, we looked at the potential threat due to hybridization and the emerging threat of climate change. SCI is owned by the U.S. Department of the Navy (Navy) and, with its associated offshore range complex, the island is the primary maritime training area for the Pacific Fleet and Sea Air and Land Teams (SEALs) and supports training by the U.S. Marine Corps, the U.S. Air Force, and other military organizations. As such, portions of the island receive intensive use by the military and can involve the movement of vehicles and troops over the landscape and can include live munitions fire, incendiary devices, demolitions, and bombardment. Altogether, 34.8% of the island’s area is located in designated training areas, and much of the island is void of any infrastructure. Most of the population of Acmispon dendroideus var. traskiae falls outside of these designated training areas; thus, direct impacts to the population are minimal. Erosion, neither naturally occurring nor that induced by human activities, has affected any documented occurrence of A. d. var. traskiae to date. While the full impact of invasive species on A. d. var. traskiae is unknown, the effects are likely minimal or localized, given the expansion of A. d. var. traskiae on the island despite the presence of invasive species. Future impacts from fire remain uncertain. Fires are typically small, of low severity, and infrequent, and given they are most often ignited due to training, their typical locations are somewhat predictable. However, an increase in the frequency or severity of fires in the future, potentially due to short-term impacts of climate change, increased training, or an increase in invasive grasses, could have additional impacts to A. d. var. traskiae. We found that 50% of watersheds and 66% of individuals are located in areas where no quantifiable threats exist. Only 22% of watersheds and 10% of individuals are associated with a threat that could potentially adversely impact 50% or more of the locations, individuals, or area within the watershed. To help further ameliorate these remaining threats, the Navy implements a wildland fire management plan (US Navy 2009) to address fire-management. The Navy addresses erosion and targeted removal of invasive species, in general, through the Integrated Natural Resources Management Plan (INRMP), addresses training-related erosion through the Erosion Control Plan, and addresses further introduction of invasive species through implementation of the

SSA Report – San Clemente Island Lotus iii March 2020

biosecurity plan. Military training on SCI has been and will continue to be dynamic as it evolves to meet new requirements, and changes that may affect Acmispon dendroideus var. traskiae are unknown. For instance, future training may include introduction of new training methods, equipment, and activities that would affect fire-frequency, fire-severity, or erosion. However, changes are expected to be incremental, as they have been in the past, and impacts to federally listed and sensitive species will be addressed in environmental analyses required under the National Environmental Policy Act (NEPA) and ESA. The factors that appear to have the most potential to impact species viability in the future are land use, fire, and climate change, including potential compounded effects. However, we are unable to address the long-term impacts of climate change because how climate change will affect SCI remains unclear. Most importantly, the persistence and timing of the fog layer, which provides moisture and a refuge from the full impacts of warming, is unknown. However, we assume that climate change will not have major effects on Acmispon dendroideus var. traskiae in the next 20 to 30 years, although we account for possible short-term climate impacts. Therefore, we consider the future of A. d. var. traskiae in terms of its threats and conservation efforts over the next 20 to 30 years. Therefore, to assess the future viability of Acmispon dendroideus var. traskiae, we considered three future scenarios that encompass the uncertainty associated with fire and military training, as well as uncertainty in the levels of recruitment over time: Scenario 1, our status quo scenario, assumes fire patterns and severity continue and current training impacts are maintained. Scenario 2 assumes increased training impacts and increased fire frequency/severity (due to short-term climate change impacts or the increase in training). Scenario 3 also assumes a threat increase but assumes extreme fire frequency/severity and extreme training impacts. We present the resulting population size as a range using a low and high recruitment estimate (Table A). Our methods predict that, in the next 20 to 30 years, the number of occupied watersheds is likely to increase, assuming that the species will be able to colonize new watersheds. The numbers of watersheds considered highly or very highly resilient increases in our most optimistic scenario and decreases by 3 in our most pessimistic. However, the resulting population estimates in all three scenarios do not drop below 20,000 individuals, and the current population estimate is within the range of estimates for the scenarios where additional threats are modeled (Scenarios 2 and 3) (Table A). In the absence of major threats (Scenario 1), with no factors limiting sustained recruitment, we do not expect any stochastic impacts to affect Acmispon dendroideus var. traskiae in a significant way over the next 20 to 30 years. We therefore expect that the entire island population is likely to increase in resiliency under Scenario 1. Even under the Scenario 3, despite localized extirpations in some of the northernmost and southernmost parts of its range, the total population still only is projected to potentially see a decrease of about 1,000 individuals island-wide and may increase from current. Thus, we still expect the island population will remain resilient to normal stochastic impacts (Table A). We likewise do not expect representation or redundancy to decrease in a meaningful way. Thus, we expect that the species would be able to sustain most major catastrophic events, such as unprecedented fires, major erosion events (such as caused by periods of heavy rainfall), or an outbreak of an invasive, predatory, or pathogenic species, or a change in environmental conditions. Only an unusually severe and unprecedented catastrophic impact could threaten the viability of the species. For instance, if the fire footprint changes and more severe fires break out where the species is numerous, like along the eastern escarpment, could have major impacts to

SSA Report – San Clemente Island Lotus iv March 2020

the population size, and thus, redundancy. Also, severe or extensive droughts, coupled with other stressors, could have substantial impacts to species viability. A severe drought could impact the vegetation island-wide, although we’d expect at least some individuals would be able withstand even severe drought. Still, like all endemics, A. d. var. traskiae has a small range and is confined to SCI and would be unable to disperse elsewhere.

Table A. The number of watersheds considered of very high, high, moderate and low resiliency and the total estimated population as considered current and in each of our four future scenarios. Watershed numbers in parentheses represent the total watersheds assuming recruitment into new watersheds, with a range from low (5 new watersheds) to high (10 new watersheds) recruitment. Watershed Resiliency Very High High Moderate Low Total Individuals Current 9 13 17 18 57 20,743 Scenario 1 10 13 16 18 (23-28) 57 (62–67) 21,595–25,708 Scenario 2 10 11 18 18 (23-28) 57 (62–67) 20,627–24,556 Scenario 3 10 9 18 20 (25-30) 57 (62–67) 19,706–23,460

SSA Report – San Clemente Island Lotus v March 2020

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ...... ii EXECUTIVE SUMMARY ...... iii TABLE OF CONTENTS ...... vi LIST OF TABLES ...... viii LIST OF FIGURES ...... x Section 1 – INTRODUCTION and analytical framework ...... 12 1.1 Status of the Taxon ...... 13 Section 2 – TAXON BIOLOGY...... 14 2.1 Taxonomy ...... 14 Genetics...... 14 2.2 Description ...... 15 2.3 Range and Distribution ...... 15 2.4 Habitat ...... 18 2.5 Life History ...... 19 2.6 Population size and abundance ...... 20 2.6.1 Current distribution ...... 23 Section 3 – INDIVIDUAL AND POPULATION NEEDS ...... 26 3.1 Population Resiliency ...... 27 3.1.1 Individual Level ...... 28 3.1.2 Population Segment (Watershed) Level ...... 28 3.2 Taxon Redundancy and Representation...... 29 3.2.1 Taxon Level ...... 29 Section 4 – FACTORS INFLUENCING VIABILITY ...... 30 4.1 Land Use (direct effects) ...... 31 Management efforts ...... 36 Summary ...... 36 4.2 Erosion and Roads ...... 37 Management efforts ...... 38 Summary ...... 39 4.3 Invasive plants ...... 39 Management efforts ...... 41 Summary ...... 41 4.4 Fire ...... 42

SSA Report – San Clemente Island Lotus vi March 2020

Management efforts ...... 48 Summary ...... 49 4.5 Climate Change ...... 50 Summary ...... 51 4.6 Hybridization ...... 51 4.7 Summary of Factors Influencing Viability ...... 52 Section 5 –CURRENT CONDITION ...... 55 5.1 Populations and Management Units ...... 55 5.2 Methods for Estimating Current Condition ...... 55 Within individual watersheds ...... 56 Island-wide ...... 56 5.3 Current Condition Results...... 57 Within individual watersheds ...... 57 Island wide ...... 57 5.4 Current Population Resiliency ...... 58 5.6 Current Representation ...... 60 5.6 Current Redundancy ...... 61 Section 6 – FUTURE CONDITIONS AND VIABILITY ...... 61 6.1 Introduction ...... 62 6.2 Methods...... 63 Growth and recruitment: ...... 63 Fire frequency and severity:...... 64 Land Use/Training ...... 65 6.3 Models and Scenarios ...... 65 6.4 Future Resiliency ...... 66 6.5 Future Representation ...... 67 6.6 Future Redundancy ...... 68 6.7 Limitations and Uncertainties ...... 68 6.8 Conclusions ...... 69 References Cited ...... 71 APPENDIX A ...... 76 APPENDIX B ...... 79 APPENDIX C ...... 81

SSA Report – San Clemente Island Lotus vii March 2020

LIST OF TABLES

Table 1. Vegetation types that support Acmispon dendroideus var. traskiae on SCI [using vegetation data from the Integrated Natural Resources Management Plan (US Navy 2013, pp. 3–59) and current distribution (US Navy 2017)]. From Vanderplank et al. 2019, p. 11...... 19 Table 2. Increasing population trend in Acmispon dendroideus var. traskiae. (USFWS 1984, p. 59; Junak and Wilken 1998, p. 261-266; Junak 2006 in USFWS 2008 p. 115; Vanderplank et al. 2019, p. 10). Survey extents are unknown. Taken from Vanderplank et al. 2019, p. 10...... 22 Table 3. Number of watersheds known to be occupied by Acmispon dendroideus var. traskiae recorded by decade. These represent only surveyed watersheds where A. d. var. traskiae were found in that decade and do not account for watersheds that may have remained populated but were not surveyed in subsequent decades...... 22 Table 4. Summary of training areas, their size, use, and the threats to Acmispon dendroideus var. traskiae within each...... 33 Table 5. The numbers of locations and total individuals that occur within each of the training area types based on the distribution of Acmispon dendroideus var. traskiae considered current...... 35 Table 6. Locations and individuals by watershed located within 30 m (100 ft) of a road, and what percent of the total locations and total individuals that represents that could potentially be affected...... 38 Table 7. Fire severity classes and definitions, reproduced from the US Navy 2009 Fire Management Plan for SCI, with severity classes adapted from the National Park Service (1992)...... 42 Table 8. Location points and individuals counted at points where a fire had burned within the past 10 years...... 45 Table 9. Numbers and percentages of watersheds and individuals assessed to have varying levels of threats: none, low (threats that could potentially affect <50% of the locations, individuals, or area within the watershed), or medium (threats that could potentially affect ≥50% of the locations, individuals, or area within the watershed). Threats identified include locations or individuals near the AVMR or a road or in the TARs, and percent of the watershed area that burned once or >1 time in the past 20 years (1999–2018)...... 53 Table 10. Occupied watersheds that may have lost Acmispon dendroideus var. traskiae during the 2012 and 2017 fire seasons where fires burned at a severity that can kill shrubs. Percentages are given for the numbers of individuals that could have been affected in each severity class, as well as the total percent of individuals that may have experienced negative effects of the fire, and the resulting adjusted estimate of the total individuals. . 57 Table 11. Total locations and individuals considered current, broken down into survey points retained by year. Our methodology estimates approximately 21,251 individuals at 249 locations...... 58 Table 12. The number of watersheds that fall into each of our resiliency categories, the numbers of individuals the watersheds in each category accounts for, and the percent of the total island wide population represented...... 59 Table 13. The number of watersheds considered of very high, high, moderate and low resiliency and the total estimated population as considered current and in each of our three future

SSA Report – San Clemente Island Lotus viii March 2020

scenarios. Watershed numbers in parentheses represent the total watersheds assuming recruitment into new watersheds, with a range from low (5 new watersheds) to high (10 new watersheds) recruitment...... 66 Table 14. Occupied watersheds, including the current number of locations and individuals present (adjusted for the 2017 fire season), the percent of locations and individuals near roads, the percent of each watershed that burned in the last 20 years and more than once in that timespan, and the projected individuals (within a range given low versus high growth) that will occur in that watershed in 20-30 years under each of three scenarios. Threats are represented as low (yellow) and moderate (pink). Numbers of individuals under current and future scenarios are represented as low (pink), moderate (yellow), high (light green), very high (dark green) and extirpated (red), depending on population size and number of locations (see Section 5.4 and Section 6.4)...... 76 Table 15. Conservation measures for terrestrial plants on San Clemente Island (SCI) as relevant to Acmispon dendroideus var. traskiae, were taken from the Biological Opinion (BO; USFWS 2008) and Table 3-48 of the Integrated Natural Resources Management Plan (INRMP; US Navy 2013). Taken from Vanderplank et al. 2019, p. 14...... 79 Table 16. Names referencing the 29 occurrences used in the 2013 downlisting rule, including the corresponding element occurrences contained in each, and the overlapping Watershed IDs used as units of resiliency in this SSA...... 83

SSA Report – San Clemente Island Lotus ix March 2020

LIST OF FIGURES

Figure 1. Species Status Assessment Framework ...... 13 Figure 2. Southern California Channel Islands off the coast of California...... 17 Figure 3. Distribution of locations of Acmispon dendroideus var. traskiae per decade, with occupied watersheds in gray. Known locations occupied 4 watersheds between 1980 and 1989, 6 watersheds between 1990 and 1999, 32 watersheds between 2000 and 2009, and 50 watersheds between 2010 and 2014. Only records documented in each decade are shown, though they may have remained extant in subsequent decades when they were not surveyed or located...... 23 Figure 4. Distribution of Acmispon dendroideus var. traskiae considered current as per methodology described in Section 4.1...... 26 Figure 5. Habitat and population factors that influence the viability of Acmispon dendroideus var. traskiae throughout its range...... 28 Figure 6. Factors that affect population resiliency in Acmispon dendroideus var. traskiae. This is not a complete compilation but represents the most important factors...... 31 Figure 7. Locations of Acmispon dendroideus var. traskiae as considered current in relation to the training areas on SCI, including the Impact Areas, the Training Areas and Ranges (TARs), the Assault Vehicle Maneuver Areas (AVMAs), the Infantry Operations Area (IOA), and the Shore Bombardment Area (SHOBA), which occupies the southern third of the island. Current Restricted Access Areas (RAAs) are also shown, but these change as unexploded ordnance are removed...... 32 Figure 8. Total acres that have burned annually in wildfires and acres that have a moderate to high severity (severity classes 1, 2, or 3)...... 44 Figure 9. Acres burned annually for years where fires were estimated since listing...... 46 Figure 10. Locations of Acmispon dendroideus var. traskiae (ACDET) points considered current in relation to areas where fires have burned in the last 20 years (1999–2018, after the initiation of fire management), including number of fires in that time...... 47 Figure 11. Locations of Acmispon dendroideus var. traskiae (ACDET) points considered current in relation to areas where fires where severity data is known have burned (2007–2018). Severity categories 1, 2, and 3 have the potential to burn shrubs where they will not resprout; severity categories 4 and 5 have little to no effect on shrubs...... 48 Figure 12. Representation of locations of watersheds where no threats exist to Acmispon dendroideus var. traskiae (ACDET), a low level of threats exist to the watershed (threats could potentially affect <50% of the locations, individuals, or area within the watershed), or a moderate level of threats exist to the watershed (threats could potentially affect ≥50% of the locations, individuals, or area within the watershed). Threats identified include locations or individuals within 100 ft of a road or the AVMR, in the TARs, and percent of the watershed area that burned once or >1 time in the past 20 years...... 54 Figure 13. Current resiliency of Acmispon dendroideus var. traskiae (ACDET) (based on estimated number and distribution of individuals) by watershed...... 60 Figure 14. Resiliency estimates by watershed (based on number of individuals) currently as well as under each of our three scenarios. Extant watershed counts do not account for recruitment into new watersheds...... 67 Figure 15. Approximate boundaries of the 29 occurrences used in the 2013 downlisting rule; polygons represent the bounding geometry (minimum convex polygons) around the the

SSA Report – San Clemente Island Lotus x March 2020

point locations and element occurrences used to define each occurrence, and canyon names (used to reference each occurrence) are provided...... 82

SSA Report – San Clemente Island Lotus xi March 2020

SECTION 1 – INTRODUCTION AND ANALYTICAL FRAMEWORK

The San Clemente Island lotus (Acmispon dendroideus var. traskiae) is a flowering subshrub in the legume or pea family (Fabaceae) that is endemic to San Clemente Island (SCI), the southernmost of the California Channel Islands. Acmispon dendroideus var. traskiae was federally listed as endangered on August 11, 1977 (USFWS 1977, p. 40682) and reclassified as threatened in 2013 (USFWS 2013, p. 45437). The Species Status Assessment (SSA) framework (USFWS 2016, entire) is intended to support an in-depth review of the taxon’s biology and threats, an evaluation of its biological status, and an assessment of the resources and conditions needed to maintain long-term viability. The intent is for the SSA Report to be easily updated as new information becomes available and to support all functions of the Endangered Species Program from Candidate Assessment to Consultations to Recovery. This SSA for Acmispon dendroideus var. traskiae is intended to provide an update on the taxon’s biological condition and level of viability. For the purpose of this assessment, we generally define viability as the ability of Acmispon dendroideus var. traskiae to sustain populations in their natural ecosystem up through and beyond a biologically meaningful timeframe, in this case, 20 to 30 years. We chose 20 to 30 years because beyond 20 to 30 years, the level of uncertainty associated with the impacts of climate change (specifically, the persistence and timing of the fog layer; see Section 5.5) becomes very high, making predictions unreliable. The available climate model projections for SCI are uncertain, but the impacts are more likely to be minimal within a 20 to 30-year timeframe. Using the SSA framework (Figure 1), we consider what the taxon needs to maintain viability by characterizing the status of the taxon in terms of its resiliency, redundancy, and representation (Wolf et al. 2015, entire).

• Resiliency describes the ability of populations to withstand stochastic events (arising from random factors). We can measure resiliency based on metrics of population health; for example, birth versus death rates and population size. Highly resilient populations are better able to withstand disturbances such as random fluctuations in birth rates (demographic stochasticity), variations in rainfall (environmental stochasticity), or the effects of anthropogenic activities.

• Representation describes the ability of a taxon to adapt to changing environmental conditions. Representation can be measured by the breadth of genetic or environmental diversity within and among populations and gauges the probability that a taxon is capable of adapting to environmental changes. The more representation, or diversity, a taxon has, the more it is capable of adapting to changes (natural or human caused) in its environment. In the absence of taxon-specific genetic and ecological diversity information, we evaluate representation based on the extent and variability of habitat characteristics across the geographical range.

• Redundancy describes the ability of a taxon to withstand catastrophic events. Measured by the number of populations, their resiliency, and their distribution (and connectivity), redundancy gauges the probability that the taxon has a margin of safety to withstand or can bounce back from catastrophic events (such as a rare destructive natural event or episode involving many populations).

SSA Report – San Clemente Island Lotus 12 March 2020

Figure 1. Species Status Assessment Framework. From USFWS 2016.

1.1 Status of the Taxon Acmispon dendroideus var. traskiae was one of the first plants to be listed pursuant to the Endangered Species Act (ESA). The major threat to this taxon at the time of listing was herbivory and habitat destruction from non-native mammals but since removal of these animals in 1992, its abundance and distribution has increased such that A. d. var. traskiae was downlisted to threatened in 2013 (USFWS 2013, p. 45437). The regulatory history of A. d. var. traskiae is as follows:

• Acmispon dendroideus var. traskiae was listed as federally endangered on 11 August 1977 (USFWS 1977). • Acmispon dendroideus var. traskiae was listed as state endangered in California in April 1982 (California Code of Regulations, Title 14, §670.2). • A Recovery Plan for Channel Islands species, including Acmispon dendroideus var. traskiae, was finalized in 1984 (USFWS 1984). • 5-year status reviews were completed in 2007 (USFWS 2007) and 2012 (USFWS 2012). These status reviews recommended reclassification of Acmispon dendroideus var. traskiae from endangered to threatened. • On May 18, 2010, USFWS received a petition dated 13 May 2010, from the Pacific Legal Foundation to downlist Acmispon dendroideus var. traskiae from endangered to threatened under the ESA.

SSA Report – San Clemente Island Lotus 13 March 2020

• On 19 January 2011, a 90-day finding was published announcing the initiation of a status review of Acmispon dendroideus var. traskiae (USFWS 2011). • On 16 May 2012, a proposed rule to reclassify Acmispon dendroideus var. traskiae from Federally Endangered to Threatened was issued (USFWS 2012). • On 26 July 2013, the final rule to reclassify Acmispon dendroideus var. traskiae from Federally Endangered to Threatened was published (USFWS 2013).

While most survey data used to support the analyses in this SSA were collected in 2011- 2013 before the 2013 downlisting rule (USFWS 2013), this SSA will differ from and build upon the 2013 downlisting rule by both reassessing the level of threats perceived in 2013, including incorporating new data on those threats and on the taxon’s response to threats, as well as assessing the data using different methods and following the SSA framework. Additional information relating this SSA’s methods and approach to that of the 2013 downlisting rule can be found in Appendix C.

SECTION 2 – TAXON BIOLOGY

In this section, we provide biological information about Acmispon dendroideus var. traskiae, including its taxonomic history, morphological description, historical and current distribution and range, and known life history.

2.1 Taxonomy Acmispon dendroideus var. traskiae is in the legume, or pea, family (Fabaceae). It has undergone taxonomic reclassification several times since the 1977 listing. The name used for this taxon when it was listed in 1977 (USFWS 1977, p. 40682) was Lotus scoparius (Nutt.) Ottley subsp. traskiae (Abrams) Raven. Subsequently, Isely (1978, p. 467) separated this and two other Channel Islands endemic taxa (L. scoparius var. veatchi Ottley and L. scoparius var. dendroideus (Greene) Ottley) from mainland Lotus scoparius. He recognized them as varieties (considered equivalent to subspecies in plants) of a single species, Lotus dendroideus, which was the oldest name among the three taxa. The name, Lotus dendroideus var. traskiae, was published by Isely in 1978 (p. 467), and recognized in floristic (Isely 1993, p. 619) and systematic treatments (Isely 1998, p. 646). Recent morphological (Sokoloff 2000, p. 128) and molecular (Allan and Porter 2000, p. 1876) data support recognition of a separate genus, Acmispon, from Lotus. The required nomenclatural combination Acmispon dendroideus (Greene) Brouillet var. traskiae (Noddin) Brouillet was made in 2008 (Brouillet 2008, p. 389). This name is recognized and accepted by the scientific community in major floristic works, such as the Jepson Manual (Brouillet 2012), and the continental Flora of North America, as well as by the California Native Plant Society (CNPS 2011). Moreover, this taxon has been referred to by other common names (such as Trask’s Island lotus and San Clemente Island broom) (Isely 1993, p. 619; 76 FR 3069, January 19, 2011; USFWS 1977, p. 40682).

Genetics Allan (1999, pp. 1–105) analyzed 10 California mainland and Channel Island taxa of Lotus (all of which are now in the genus Acmispon and referred to as such here), including Acmispon dendroideus var. traskiae. Allan (1999, pp. 50–53) sampled A. d. var. traskiae on SCI

SSA Report – San Clemente Island Lotus 14 March 2020

from Wilson Cove only. The Acmispon island populations, including A. d. var. traskiae, tended to have lower genetic variability than mainland populations (Allan 1999, p. 63). Allan’s (1999, p. 61) analysis of genetic diversity also found that the majority (67 percent) of A. d. var. traskiae’s variability is found among, rather than within, occurrences. However, more recent genetic work (McGlauglin et al. 2018, p. 754) has shown moderate levels of genetic diversity in A. d. var. traskiae, with gene flow between neighbor populations. Wallace et al. (2017, p. 747-748) found that the genetic diversity of A. d. var. traskiae is equal to or higher than that of A. d. var. dendroideus, and A. d. var. traskiae also contains unique and highly divergent genotypes. Acmispon dendroideus var. traskiae has been known to hybridize with A. argophyllus var. argenteus in disturbed areas in Wilson Cove (Liston et al. 1990, pp. 239–240; Allan 1999, p. 86). Based on intermediate characteristics, the hybrid plants appear to be first generation (F1 generation) plants from a cross between the two varieties. It is not known whether these plants are capable of producing viable seeds by backcrossing between the hybrids or with the putative parent plants (Allan 1999, p. 86). Plants of intermediate morphology were first observed by R.M. Beauchamp in 1986 (Liston et al. 1990, p. 239). In April 1989, Liston et al. (1990, pp. 239–240) noted a small number of suspected hybrids in the same area as the largest known occurrence of A. d. var. traskiae in Wilson Cove. A smaller group of nonhybrid A. argophyllus var. argenteus was found approximately 80 ft (24.4 m) upwind; the two taxa were separated by a road. No documented evidence of hybridization has been recorded anywhere else on the island (Allan 1999, p. 86), although there are unconfirmed reports in other areas (e.g., Warren Canyon; A. Braswell 2011, pers. obs. in USFWS 2012, p. 29095). Wallace et al. (2017, p. 747-748) found strong differentiation between the three varieties of A. dendroideus, with var. traskiae being the most distinct and basal to the clade. McGlaughlin et al. (2018, p. 750-751) further confirmed these findings. Acmispon dendroideus var. traskiae likely did not evolve from A. d. var. dendroideus on Santa Catalina Island but is believed to have originated from a direct independent colonization event from mainland California ancestors (Wallace et al. 2017, p. 747-748).

2.2 Description Acmispon dendroideus var. traskiae is a semi-woody, flowering subshrub. It is endemic to SCI (Isely 1993, p. 619) and is one of five taxa in the genus Acmispon found on the island (Tierra Data Inc. 2005, p. C–8; Brouillet 2008, p. 388–392). There are no other varieties of A. dendroideus found on the island. This variety can be distinguished from other varieties of A. dendroideus by its bushy habit and elongated fruits (Allan 1999, p. 88). Acmispon dendroideus var. traskiae is typically less than 4 ft (1.2 m) tall with slender erect green branches (Munz 1974, p. 449–450; USFWS 1984, p. 59; Allan 1999, p. 82). Each leaf has three to five leaflets, each approximately 0.2 to 0.3 in (5 to 9 mm) long and uniformly glabrous (surface without hair) to finely hairy (USFWS 1984, p. 59; Allan 1999, p. 82). Acmispon dendroideus var. traskiae has small yellow flowers that are bisexual and arranged in one to five flowered clusters on stalks that arise from axils between the stem and leaf of terminal shoots (Junak and Wilken 1998, p. 256). Pistils are initially yellow, turning orange then red as the fruit matures (USFWS 1984, p. 59).

2.3 Range and Distribution Acmispon dendroideus var. traskiae is endemic to SCI, located 64 miles (103 km) west of San Diego, California, and the southernmost of the California Channel Islands (Figure 2). The

SSA Report – San Clemente Island Lotus 15 March 2020

island is approximately 56 square mi (145 square km, 36,073 acres, or 14,598 hectares) (Junak and Wilken 1998, p. 2) and is long and narrow: 21 mi (34 km) long by 1.5 mi (2.4 km) wide at the north end, and 4 mi (6.4 km) wide at the south end (USFWS 1984, p. 5). The island consists of a relatively broad open plateau that slopes gently to the west. Conspicuous marine terraces line the western slope of the island while steep escarpments drop precipitously to the rocky coastline on the eastern side along the southern 75%. Many canyons, some of which are up to 500 feet (152 meters) deep, dissect the southern part of the island. Mount Thirst, the highest point on the island, rises to approximately 1,965 feet (599 meters) (US Navy 2013a, p. 1.4). Average monthly temperatures range from 58°F (14°C) to 66°F (19°C), with a monthly maximum temperature of 72°F (27°C) in August and a monthly minimum of 51°F (10°C) in December (US Navy 2013a, p. 3.11). Average monthly relative humidity varies from 54% to 86% depending on location and time of year, and the island experiences dramatic fluctuations in annual rainfall, averaging 6.6 inches (16.8 cm) (US Navy 2013a, pp. 3.11, 3.13). Precipitation is received mainly from November through April, with little from May through October. In addition to precipitation, fog drip during the typical dry season is a vital source of moisture to the SCI ecosystem (US Navy 2013a, pp. 3.9, 3.13).

SSA Report – San Clemente Island Lotus 16 March 2020

Figure 2. Location of San Clemente Island in the southern Channel Islands off the coast of California.

At listing, this taxon was mentioned to occur at Wilson Cove on the north end of the island (see Figure 4), but no other details were available (USFWS 1977, p. 40683). In 1984, the distribution of this taxon was considered to be restricted to 6 “populations” associated with rocky areas with the largest number of plants growing in the Wilson Cove area (USFWS 1984, p. 59). Additionally, there are only a few herbarium specimens of the taxon, making historical distribution and condition of the taxon difficult to assess. Thus, the historical range (based on herbarium records, CNDDB records, and the recovery plan) includes occurrences in the northern part of the island (Wilson Cove) down to the southern point (Pyramid Head). Since the final removal of all feral herbivores by 1992, the distribution of this taxon has steadily increased

SSA Report – San Clemente Island Lotus 17 March 2020

(USFWS 2012, p. 29110). By 1997, roughly 50% of documented occurrences of these plants were found in the vicinity of Wilson Cove and by 2004, 75% of the distribution of this taxon was found beyond this area and extended to the southern-most part of the island (USFWS 2007, pp. 4–5). The most recent survey data show the distribution of Acmispon dendroideus var. traskiae spans the entire length of the island from Wilson Cove to the southern tip east of Pyramid Cove (see Figure 4), a distance of approximately 19 mi (31 km) (Junak and Wilken 1998, p. 261; Junak 2006, Map A–C; Vanderplank et al. 2019, p. 27). The majority of locations tend to be clustered on north-facing slopes on the eastern side of the island (Vanderplank et al. 2019, p. 7). Acmispon dendroideus var. traskiae tends to occur in small groups of 10 to 50 individuals (Allan 1999, p. 84). Some historical locations are unknown to be extant currently due to a lack of repeat survey data and unknown survey extents (see Section 2.6.1); additionally, there are 15 watersheds that historically had Acmispon dendroideus var. traskiae in which the occupancy is currently unknown (see Section 2.6). However, the range has generally expanded over time, despite the potential loss of some of these areas (Figure 4).

2.4 Habitat Acmispon dendroideus var. traskiae establishes on north- and east-facing slopes and ridges at elevations ranging from 25 to 1,400 feet (7.6 to 463 m). and is found in canyon bottoms or along ridgelines (Junak 2006, p. 125). It appears to preferentially establish and grow somewhat colonially around rock outcrops and among large boulders situated in grassland areas and along the interface between grassland and maritime sage scrub (Allan 1999, p. 84; US Navy 2002, p. D–9); it also readily occupies disturbed sites and locations close to buildings, roads, and pipelines (US Navy 2013b, p. 3-201). It occurs on well- drained soils where adequate soil moisture is available to the plant (Junak and Wilken 1998, p. 256; US Navy 2002, p. D–9) and occurs mostly on clay to rocky soils (Vanderplank et. al 2019, p. 7). Acmispon dendroideus var. traskiae is generally associated with two habitat types on the island: canyon woodland supported on approximately 696 ac (282 ha), and maritime desert scrub along the northeastern escarpment supported on approximately 6,228 ac (2,520 ha) (US Navy 2002, pp. 3.57, 3.58). According to Junak and Wilken (1998, p. 256), A. d. var. traskiae is associated with numerous plant species, including, but not limited to: Artemisia californica, Avena fatua, Bromus spp., Calystegia macrostegia subsp. amplissima, Dichelostemma capitatum (wild hyacinth), Gnaphalium bicolor (bicolored everlasting), Hemizonia clementina (island tarplant), Opuntia spp. (prickly pear), Nassella pulchra (purple stipa), and Quercus tomentella (island live oak). The California Natural Diversity Database (2017) lists 61 associated species. Recently, the US Navy found that A. d. var. traskiae is associated with vegetation types that comprise approximately 83.5% of the island, although the taxon is not widespread throughout each type (Vanderplank et al. 2019, p. 11) (Table 1). A. d. var. traskiae was most often found in vegetation types dominated by Artemisia californica, Opuntia littoralis, and Rhus integrifolia (Vanderplank et al. 2019, p. 6).

SSA Report – San Clemente Island Lotus 18 March 2020

2.5 Life History Acmispon dendroideus var. traskiae is short-lived, with a reported lifespan of less than 5 years (USFWS 2008, p. 113); however, individuals near Wilson Cove have been observed to live longer than six years (Emily Howe, pers. comm. 2017 in Vanderplank et al. 2019, p. 6). Like other legumes, the roots of Acmispon are able to fix atmospheric nitrogen, making it available to plants in the form of NH3, enriching the soil and making them important post-fire colonizers (Sørensen and Sessitsch 2007 in Vanderplank et al. 2019, p. 4). Acmispon dendroideus var. traskiae flowers between February and August, peaking from March to May (Junak and Wilken 1998, p. 256; USFWS 2008, p. 113), with halictid bees (a family of small solitary bees that typically nest in the ground), bumblebees, and small beetles observed foraging on the flowers (Junak and Wilken 1998, p. 257; Allan 1999, pp. 64, 85). A sister taxon (A. glaber [syn. Lotus scoparius]) flowers in response to available moisture from fog and precipitation, primarily winter rainfall (Vanderplank and Ezcurra 2015, p.16), which may also be true of A. d. var. traskiae. The taxon is self-compatible, meaning it can self-pollinate (Allan 1999, pp. 85–86), but plants may also rely on insects for more effective pollination (Arroyo 1981, pp. 728–729). As they age, the flowers change color from yellow to orange to red (USFWS 2008, p. 113), which may increase pollination. Studies on sister taxon A. glaber, which also displays this trait, suggest that the orange/red flowers helped attract more pollinators to the plant overall, but individual pollinators select for yellow flowers and may visit fewer flowers per plant, which cumulatively works to reduce selfing and promote outcrossing (Jones and Cruzan 1999, p. 273). Fertilized ovaries develop into a slender, beak-like fruit 1 to 2 in (2.5 to 5 cm) long containing up to six seeds (Isely 1993, p. 619; Junak and Wilken 1998, p. 257; Allan 1999, p. 82). The fruits do not split open to release their seeds at maturity (Isely 1993, p. 619). Junak and Wilken (1998, p. 257) found that, on average, a single Acmispon dendroideus var. traskiae individual can produce approximately 36 to 64 flowering shoots, 118 to 144 flowers per shoot, and 4 to 6 seeds per fruit. This suggests that, under ideal conditions, an individual A. d. var. traskiae can produce a high volume of seeds (16,000 or more). Like most legumes, A. d. var. traskiae seeds require scarification or gradual seed coat degradation to germinate (Wall 2011, pers. comm. in USFWS 2012, p. 29095). Acmispon dendroideus var. traskiae is thought to have good long-term survival in the seed bank. Germination rates for seed stored for six years only dropped from 80% to 76%; one

SSA Report – San Clemente Island Lotus 19 March 2020

seed lot displayed 65% germination after more than 30 years in storage (Cheryl Birker, pers. comm. 2017 in Vanderplank et al. 2019, p. 6). The fire tolerance of Acmispon dendroideus var. traskiae is not well understood. Some studies have shown that the related mainland species, Acmispon glaber (deerweed), is fire tolerant and becomes more abundant in years after fire (Nilsen and Schlesinger 1981, p. 217; Westman and O’Leary 1986, pp. 184–185). Other studies indicate that intense or frequent burns (three times in 6 years) of A. glaber lead to establishment of fewer seedlings (Westman and O’Leary 1986, p. 185; Haidinger and Keeley 1993, p. 141). In SCI taxa, observations show that Acmispon argophyllus var. adsurgens (San Clemente Island bird’s-foot trefoil) germination is slowed or depressed after fire, but A. argophyllus var. argenteus (silver bird’s-foot trefoil) flourishes in burn areas (Allan 1999, pp. 90–91). Observations of A. d. var. traskiae before and several years following a fire in Canchalagua Canyon found that adult plants were usually killed by fire but were replaced with a similar number of seedlings after the fire (Navy 2002, p. D.10; Tierra Data Inc. 2005, p. 80). Based on A. d. var. traskiae’s growth characteristics and occurrence increases in areas affected by fire, and the fire adaptations of related taxa, A. d. var. traskiae may be resilient to at least occasional fire. Because it is short-lived and likely relies on its seed bank for recruitment, fire may benefit this taxon by opening up areas of bare ground for seedling germination. This is suggested by its ready occupation of disturbed areas close to buildings, roads, and pipelines (U. S. Department of the Navy, Southwest Division 2002; USFWS 2007, p. 9) and of areas where invasive species have been manually removed (O’Connor 2019, pers. comm.). However, frequent fires could exceed its tolerance of fire severity and frequency and exhaust the seed bank in repeatedly burned areas. Thus, how Acmispon dendroideus var. traskiae responds to fires of differing frequencies and severities is unknown. Fire and its potential impacts will be discussed in more detail in Section 4.4.

2.6 Population size and abundance Tracking the population size and local abundances of Acmispon dendroideus var. traskiae has been complicated by the difficulty of monitoring the taxon on a routine basis due to terrain, access restrictions implemented to minimize the risk of unexploded ordnance, and the level of effort required to survey a large area during a limited survey window. While a lot of occurrence data exists for A. d. var. traskiae, few comprehensive surveys have been attempted or conducted. Due to the constraints noted above, no survey has covered the entirety of the island. Survey areas have varied, and recorded observations have often been inconsistent and opportunistic, focused on capturing the distribution of the taxon, both during and outside of formal surveys. Survey data and opportunistic observation data collected by the US Navy consists of counts of individuals within “contiguous biologically relevant clusters that are unbroken within a line of sight and do not include any obvious barriers to dispersal, pollination, or recruitment” (Vanderplank et al. 2019, p. 8). Thus, survey data consist of a count of the number of individuals at each of these locations, represented as a single point in space. As resources and terrain limit the feasibility of an exhaustive survey of all 36,073 acres (14,598 hectares) on the island, the precise abundance of this taxon in any year or set of years is difficult to determine. Because we do not know the extent or intensity of any of these historical survey efforts, we cannot estimate the number of previously undetected individuals (Vanderplank et al. 2019, p. 10). To limit the possibility of double-counting when looking at

SSA Report – San Clemente Island Lotus 20 March 2020

individuals counted over time, we compared the results of individual surveys rather than combining surveys, as it is unknown if or to what extent survey extents overlapped. Watersheds have been suggested by the Navy as a means of delineating the population into segments in order to monitor trends in the number of Acmispon dendroideus var. traskiae individuals in these different segments in the future. Because every point on the island can be ascribed to a watershed, watersheds can serve as a unit for monitoring and management. Watersheds may also be meaningful biologically by influencing the dispersal of seeds; however, while it seems probable that new recruitment would be more likely to occur in an occupied watershed than an unoccupied one due to seed dispersal, this is speculation. Therefore, for consistency and in an attempt to standardize A. d. var. traskiae data, we will discuss the current population of A. d. var. traskiae, general trends over time, and future population estimates in terms of the documented number of individuals from various survey years and numbers of occupied watersheds. Comparing the results of individual surveys/studies of Acmispon dendroideus var. traskiae on SCI shows a general increase in the number of A. d. var. traskiae detected in each subsequent survey (Table 2). There is no information about the abundance of A. d. var. traskiae at the time of its listing in 1977. The 2002 Integrated Resources Management Plan (INRMP) states that only 9 occurrences and 1,340 individuals were known in 1980 (US Navy 2002, p. D- 10; USFWS 2007, p. 4). In the 1984 Recovery Plan (USFWS 1984, p. 59), six occurrences of A. d. var. traskiae were recognized, all generally associated with rocky areas. However, no other specific information regarding taxon location or numbers of individuals at those six sites was provided in the Recovery Plan, except the statement that ‘‘the largest number of plants grow in the vicinity of Wilson Cove’’ (USFWS 1984, p. 59). Acmispon dendroideus var. traskiae was thought to typically occur in small groups of 10 to 50 individuals (Allan 1999, p. 84), but much larger groups have been noted (750 by Junak and Wilken 1998, p. 256; over 1,000 during the 2011 and 2012 surveys by US Navy, in prep.). A study conducted in 1996 and 1997 documented 64 locations, comprising more than 3,000 individuals, approximately half of which were found in Wilson Cove (Junak and Wilken 1998, p. 261-266). Another study between 2003 and 2006 documented 69 locations and 6,570 individuals (Junak 2006 in USFWS 2008, p. 115). While we are considering them separate, these two studies were combined in a 2008 Biological Opinion, estimating a total of 9,674 individuals island wide (USFWS 2008, p. 116). Surveys conducted in 2011 and 2012 by the Navy yielded 136 locations and a total of 11,938 individuals (Vanderplank et al. 2019, p. 10). These numbers do not include as many as 34 historical populations that were previously recorded within Restricted Access and Impact areas which were inaccessible during the 2011–2012 surveys. These occurrences are understood to have recruited and expanded naturally, except for one documented exception which was partially re-planted at a site on Lotus Hill south of the Natural Resources office in 2009 after a disturbance (O’Connor 2019, pers. comm). We assessed the number of watersheds documented as occupied over time, and found that watersheds observed to be occupied in a given decade have increased, from 4 between 1980 and 1989, to 50 known from between 2010 and 2014 (Figure 3, Table 3). These watershed counts represent only those with documented records of Acmispon dendroideus var. traskiae in that decade and do not account for watersheds that may have remained populated but were not surveyed in subsequent decades.

SSA Report – San Clemente Island Lotus 21 March 2020

Table 2. Increasing population trend in Acmispon dendroideus var. traskiae. (USFWS 1984, p. 59; Junak and Wilken 1998, p. 261-266; Junak 2006 in USFWS 2008 p. 115; Vanderplank et al. 2019, p. 10). Survey extents are unknown. Taken from Vanderplank et al. 2019, p. 10.

Survey year Occurrences Individuals 1980 9 1,340 1984 6 Unknown 1996–1997* 64 3000+ 2003–2006* 69 6570 2011–2012 136 11,938 *These survey years were combined in USFWS 2008 to represent the population estimate.

Table 3. Number of watersheds known to be occupied by Acmispon dendroideus var. traskiae recorded by decade. These represent only surveyed watersheds where A. d. var. traskiae were found in that decade and do not account for watersheds that may have remained populated but were not surveyed in subsequent decades. Decade Occupied Watersheds 1980-1989 4 1990-1999 6 2000-2009 32 2010-2014 50

SSA Report – San Clemente Island Lotus 22 March 2020

Figure 3. Distribution of locations of Acmispon dendroideus var. traskiae per decade, with occupied watersheds in gray. Known locations occupied 4 watersheds between 1980 and 1989, 6 watersheds between 1990 and 1999, 32 watersheds between 2000 and 2009, and 50 watersheds between 2010 and 2014. Only records documented in each decade are shown, though they may have remained extant in subsequent decades when they were not surveyed or located.

Despite inconsistencies in the survey data, the data indicate that the number of individuals and the range of Acmispon dendroideus var. traskiae have increased over time (Table 2, Table 3, Figure 3), although the rate of increase and true population size in any year is unknown.

2.6.1 Current distribution Since no existing singular dataset or survey effort has been comprehensive enough to provide an estimate of the current distribution and abundance of Acmispon dendroideus var.

SSA Report – San Clemente Island Lotus 23 March 2020

traskiae across its range on SCI, to describe the current condition of the taxon for the purposes of this document, we first defined what will be considered the current distribution. While the 2011 and 2012 surveys represent the most extensive and recent surveys to date, these surveys alone likely do not represent the full distribution of A. d. var. traskiae on the island. Like the rest of the survey data, the full extent of the survey area from 2011 and 2012 is not known because areas within which plants were not found were not documented, but these surveys did not cover the entirety of the island due to the constraints described in Section 2.6 above. Available data consist only of presence data, with no known corresponding absence data. Additional survey data from 1977 through 2014 include additional locations not documented in 2011 and 2012, some of which are likely still extant. Therefore, we developed a rule set to conservatively define the current distribution and abundance of Acmispon dendroideus var. traskiae. To ensure we were basing estimates on the most recent and robust dataset, we used the 2011 and 2012 survey points and then added point locations based on our set of rules. First, we did not want to go back too far in time and assume a population was still present when it might not be. Second, experts concurred that locations of A. d. var. traskiae recorded as long as 15 years ago are likely to persist (Munson 2019 pers. comm.; O’Connor 2019 pers comm). Therefore, we applied the following criteria to the additional survey data to establish a current estimated distribution for Acmispon dendroideus var. traskiae. The estimate of current distribution was made by including some data collection made prior and subsequent to the 2011 and 2012 surveys, that met the following conditions: 1) Includes count of individuals at the point and year collected. 2) Was recorded within the past 15 years (2004–2018). 3) Does not occur within 50 m of a 2011 or 2012 survey point. 4) Does not occur within 50 m of a retained survey point from 2013 or 2014 if it is older than that point. 5) Does not occur within 100 m of a retained survey point from between 2005–2010 if it is older than that point. We took the following steps to get to this distribution: 1) We buffered the 2011 and 2012 data points by 50 m then deleted all points from other years that fell within these buffers. 2) We buffered the remaining 2014 data points by 50 m and deleted all remaining points that fell within these buffers. 3) We repeated step 2 with the 2013 data. 4) We then buffered the remaining 2010 data points by 100 m and deleted all remaining points that fell within the buffers. 5) We repeated this process with the 2009, then 2008, 2007, 2006, and 2005 data, in that order. This methodology makes the following assumptions. 1) Points recorded in the same year are considered separate, regardless of separation distance. 2) Recorded point locations may be inaccurate up to 50 m for newer points (2011–2014) and 100 m for older points (2010 and older), so to conservatively avoid double counting, we went with the more recent point if two points fell within these distances of each other.

This yielded 249 locations within 58 watersheds totaling 21,251 individuals (Figure 4), which we estimate to be the current abundance and distribution of Acmispon dendroideus var. traskiae. While we acknowledge that these numbers may not be exact due to mortality or

SSA Report – San Clemente Island Lotus 24 March 2020

expansion since these counts were completed, these data represent the best available information on the taxon and should approximate reality. We assume that a small percentage of these point locations have been completely extirpated, and those that may have been lost could have been replaced by unknown locations. Further, count numbers should also be approximate, as we expect mortality to be made up for by expansion over time in the absence of additional threats. A fire severe enough to affect shrubs such as Acmispon dendroideus var. traskiae occurred in 2017 along the eastern escarpment in part of SHOBA after collection of this data; post-fire monitoring has indicated that many plants survived, but their numbers may have decreased. This will be further explored and accounted for in Section 5; however, while we expect the distribution is the same, a decrease in population numbers in this region is not portrayed in Figure 4. It is important to note that this distribution mostly overlaps with the 29 occurrences listed in the 2013 rule (Figure 15 in Appendix C). The current distribution, as we define it here, is missing a few points that are older than 2004 but includes some new points from 2013 and 2014 not included in the listing rule. The 29 occurrences listed in the 2013 rule included all known survey data, including historical data from 1979. Without repeated survey data in some of those locations, it is unknown whether individuals observed 40 years ago still persist. Compared to the historical distribution, there are 15 watersheds that were once occupied that are no longer considered occupied (Figure 4).

SSA Report – San Clemente Island Lotus 25 March 2020

Figure 4. Distribution of Acmispon dendroideus var. traskiae considered current as per methodology described in Section 4.1.

SECTION 3 – INDIVIDUAL AND POPULATION NEEDS

In this section we synthesize the information in the preceding sections to highlight the overall resource-needs of Acmispon dendroideus var. traskiae. Typically, we assess needs at the individual and population levels and then finally at the taxon level, and we consider that the effects are cumulative: survival of individual plants contributes to survival of the cohort of plants within the area in which it occurs which, in turn, contributes to the survival and persistence of the population and ultimately, the taxon. We consider that the locations of A. d. var. traskiae on SCI constitutes a single population defined as a group of interbreeding individuals that produce offspring (Primack 1995, p. 12-13), rather than a taxon comprised of multiple, but distinct, populations. There are few obvious natural divisions between locations across the range of the

SSA Report – San Clemente Island Lotus 26 March 2020

taxon (USFWS 2013, p. 45437) and this conclusion is supported by an assessment of population genetic diversity mentioned above. We do, however, assess resource-needs by location (spatially distinct localities) given variability in habitat conditions and military operations across SCI. Hence, we consider resource-needs at the group-level by considering the separate locations grouped by watershed. If the needs of some number of individuals in a population are being met, allowing for an adequate population size and a sufficient rate of growth, then that population will likely have sufficient resiliency. The number of resilient populations, their size, distribution, and their level of connectivity can be used as a measure of the taxon’s level of redundancy relative to potential catastrophes. Similarly, the breadth of genetic or environmental diversity within and among populations can be used as a measure of the taxon’s level of representation. Thus, for the taxon to sustain populations in the wild over time and be viable, the populations need to be able to withstand stochastic events (to have resiliency), and the taxon as a whole needs to be able to withstand catastrophic events (to have redundancy) and to adapt to changing environmental conditions (to have representation). For the purposes of this report, we define viability as the ability of Acmispon dendroideus var. traskiae to sustain itself in the wild over time. We describe the taxon’s needs at the individual, watershed (population segment), and taxon’s levels in terms of resiliency, redundancy, and representation.

3.1 Population Resiliency For Acmispon dendroideus var. traskiae to maintain viability, its population or some portion thereof must be resilient. Stochastic factors that have the potential to affect A. d. var. traskiae include drought, erosion, and fires. Other factors that influence the resiliency of the A. d. var. traskiae population include ecological integrity of the plant community, population size, and dispersal ability. Influencing those factors are elements of A. d. var. traskiae ecology that determine whether the population can grow to maximize habitat occupancy, thereby increasing resiliency of populations (Figure 5). These factors and habitat elements are discussed below. Assuming that these factors influence the number of individuals that can or will occupy available habitat on the island, we will use estimated population size (estimated number of individuals based on count data from our distribution considered current) as our measure to estimate resiliency.

SSA Report – San Clemente Island Lotus 27 March 2020

Figure 5. Habitat and population factors that influence the viability of Acmispon dendroideus var. traskiae throughout its range.

3.1.1 Individual Level At the individual level, habitat characteristics and requirements for individual plants include well-drained soils with adequate moisture, access to resources (including water, sunlight, and nutrients), and pollinators to successfully reproduce. Individual plants need these sufficient resources (good habitat, limited competition, etc.), and limited natural or anthropogenic disturbance (e.g., infrequent extreme rain or drought, limited foot traffic, etc.) for survival. Little information exists to inform our understanding of the fitness of individual plants. Seemingly few, if any, intrinsic factors limiting individual plant growth and development of Acmispon dendroideus var. traskiae (e.g., no evidence of genetic stochasticity having occurred). Extrinsic factors affecting individual fitness of plants could include dense vegetation that reduces or precludes successful germination, weather conditions that preclude germination until after seeds lose viability, and/or a lack of pollinator service that precludes successful reproduction. Habitat conditions that favor germination and recruitment may be enhanced by occasional fire, and fire may increase A. d. var. traskiae germination. However, post-fire monitoring has been conducted opportunistically following accidental fires and does not allow for an assessment of the response of A. d. var. traskiae to different fire severities over different time intervals.

3.1.2 Population Segment (Watershed) Level Within a watershed, habitat must be maintained to support a sufficiently large number of individuals needed for persistence, although that number is unknown. Habitat features must be sufficiently contiguous or absent dispersal barriers such that necessary levels of pollination and seed set required for population dispersal and gene-flow within each watershed are achieved to maintain sufficient levels of genetic diversity to preclude deleterious effects to the population from inbreeding depression and genetic drift (Ellstrand and Elam 1993, entire). However, the necessary levels of dispersal and gene flow are unknown for this taxon. No barriers to the movement of pollinators (for gene flow) or dispersal events appear to exist within watersheds.

SSA Report – San Clemente Island Lotus 28 March 2020

The influence of stochastic variation in demographic (recruitment and mortality) rates is much higher for small populations than large ones. Stochastic variation in demographic rates causes small populations to fluctuate randomly in size. In general, the smaller the population, the greater the probability that fluctuations will lead to extinction. There are also genetic concerns with small populations, including reduced availability of compatible mates, genetic drift, and inbreeding depression. Small populations have low resilience, leaving them particularly vulnerable to stochastic events. While much of the island is composed of the vegetation communities thought to provide potential habitat for Acmispon dendroideus var. traskiae, additional habitat limitations may exist that are currently unknown. The degree to which habitat is unfragmented and the suitability of available unoccupied habitat can help determine dispersal-success, which then influences gene flow, local adaptation, extinction risk, and the potential for organisms to move in response to a changing climate (Taylor et al. 1993, p. 572).

3.2 Taxon Redundancy and Representation 3.2.1 Taxon Level For the taxon to be viable, there must be adequate redundancy (suitable number and distribution of individuals) to allow the taxon to withstand catastrophic events. Redundancy improves with increasing numbers of occupied watersheds and with increasing numbers of individuals in each. Habitat for Acmispon dendroideus var. traskiae must be sufficiently contiguous or absent dispersal barriers such that necessary levels of population dispersal and gene-flow can occur among watersheds to maintain sufficient levels of genetic diversity (Ellstrand and Elam 1993, entire). This taxon will be most viable by occupying a range of watersheds in multiple habitats across the island. For instance, by occupying both the western and eastern sides of the island, which have differing moisture patterns, the taxon might be buffered from a localized catastrophic impact, such as a severe fire, or an island-wide catastrophic impact such as a prolonged drought, which could affect A. d. var. traskiae on the eastern and western sides of the island differently due to differences in soil, slope, insolation, fog, rainfall, or other variables. Proximity of other individuals in other occupied watersheds will allow gene flow among watersheds and individuals and improve the chances of dispersal to new locations or across watersheds, which would allow the taxon to persist and become reestablished after catastrophes. Again, while much of the island seems habitable to this taxon, additional habitat limitations may exist that are currently unknown. No barriers to the movement of pollinators (for gene flow) or dispersal appear to exist on SCI, although terrain may favor recruitment within rather than across watersheds. Also for taxon viability, there must be adequate representation (genetic and environmental diversity) to allow the taxon to adapt to changing environmental conditions. Representation improves with increased genetic diversity and/or diverse environmental conditions within and among populations. Genetic studies indicate that Acmispon dendroideus var. traskiae has moderate amounts of genetic diversity, with the majority of the variability occurring among, rather than within, specific locations on SCI. This may have occurred as plants that persisted in areas that escaped grazing pressure retained genetic diversity through outcrossing. In addition to genetic data, adequate representation for this taxon would be indicated by the population being distributed throughout multiple habitat types and across multiple elevations, indication that the taxon is adapted to these different environmental and habitat conditions.

SSA Report – San Clemente Island Lotus 29 March 2020

SECTION 4 – FACTORS INFLUENCING VIABILITY

The following discussion provides a summary of the factors that are affecting or could be affecting the current and future condition of Acmispon dendroideus var. traskiae throughout some or all of its range. The current habitat conditions for Acmispon dendroideus var. traskiae on SCI are the result of historical land use practices. SCI was used legally and illegally for sheep ranching, cattle ranching, goat grazing, and pig farming (USFWS 2012, p. 29-91, US Navy 2013a, p. 2-3). Goats and sheep were introduced early by the Europeans, and cattle, pigs, and mule deer were introduced in the 1950s and 1960s (US Navy 2013a, p. 3-185). These non-native herbivores greatly changed the vegetation of SCI and were cited in the final rule (USFWS 1977, p. 40863) for the listing of A. d. var. traskiae as the main cause of this taxon’s decline. Sheep were removed from the island in the 1930s, but pigs were not completely eradicated until around 1990 and feral goats were removed by 1992 (Keegan et al. 1994, p. 58; USFWS 2012, p. 29093). Although A. d. var. traskiae may not have been a primary target of the mammalian herbivores, overgrazing and browsing probably led to the direct loss of plants through trampling and rooting. These grazing, browsing, and rooting animals also altered the habitat by creating trail networks with bare, compacted soil. Overgrazing, erosion, and other impacts to the vegetation led to severe habitat degradation and loss of suitable habitat that likely curtailed the range of A. d. var. traskiae and other endemic plants on the island (USFWS 1997, p. 42697). The current distribution of Acmispon dendroideus var. traskiae is undoubtedly a product of topographical features that made certain areas inaccessible to goats and therefore provided refugia for A. d. var. traskiae to survive the intense browsing. Further, seeds may have survived in the soil seed bank in other areas of the island, and together, these refugia determined the pattern of A. d. var. traskiae’s expansion and recolonization of the island. At listing, nonnative herbivores were the primary threat to Acmispon dendroideus var. traskiae (USFWS 1977, p. 40682; Keegan et al. 1994, p. 58; USFWS 2007, pp. 1–22). As a result of their removal, habitat conditions improved and led to changes in the cover of native and nonnative plants on the island, further evidenced by the increases in A. d. var. traskiae and several other threatened and endangered taxa since the feral animals were removed (Uyeda et al. 2019, pp. 6, 22, 30). In the absence of the primary threat, additional threats to A. d. var. traskiae that have been identified include: (1) land use, (2) erosion, (3) nonnative plants, and (4) fire (Figure 6). Additionally, we looked at the potential threat due to hybridization and the emerging threat of climate change. We provide general information on each threat in this discussion; more specific information on the current status of A. d. var. traskiae in relation to these threats will be provided in Section 5.

SSA Report – San Clemente Island Lotus 30 March 2020

Figure 6. Factors that affect population resiliency in Acmispon dendroideus var. traskiae. This is not a complete compilation but represents the most important factors.

We assess threats to individuals of Acmispon dendroideus var. traskiae within each occupied watershed.

4.1 Land Use (direct effects) SCI is owned by the U.S. Department of the Navy (Navy) and, with its associated offshore range complex, the island is the primary maritime training area for the Pacific Fleet and Sea Air and Land Teams (SEALs) (USFWS 2012, p. 29078). The island also supports training by the U.S. Marine Corps, the U.S. Air Force, the U.S. Army, and other military organizations. As the western most training range in the eastern Pacific Basin, where training operations are performed prior to troop deployments, portions of the island receive intensive use by the military (US Navy 2008a, p. 2– 2). Various training activities occur within particular land use designations and training areas on the island. Military training activities within some of these training areas can involve the movement of vehicles and troops over the landscape and can include live munitions fire, incendiaries, demolitions, and bombardment (Table 4). The direct effects of military training and other land uses will be discussed here; indirect effects, such as erosion and fire, will be discussed in separate sections below. SCI supports 22 terrestrial Training Areas and Ranges (TARs), four Assault Vehicle Maneuver Areas (AVMAs), and the Infantry Operations Area (IOA). TARs are operating areas that support demolition, over-the-beach, and tactical ingress and egress training for Naval Special Warfare personnel (US Navy 2008a, p. 2–7). AVMAs are designated for off-road vehicle use, including tracked vehicles, and the IOA is designated for dispersed foot traffic by military units in support of a battalion-sized landing (US Navy 2008a, p. 2–37) (Figure 7). While the IOA is a broad designated area for foot traffic, use has been, and is anticipated to continue to be,

SSA Report – San Clemente Island Lotus 31 March 2020

concentrated around the AVMR (Artillery Vehicle Maneuver Road). Soldiers fan out from but move in concert with artillery vehicles, which are restricted to the AVMAs and AVMR; accordingly, foot traffic has occurred predominantly within 50 feet of the AVMR, within the IOA (Booker 2019, pers. comm.). Other major potential impacts (artillery firing points [AFPs] and bivouacking) within the IOA also occur near the road (USFWS 2008, pp. 42, 164). This buffer around the AVMR makes up less than 1% of the IOA (Table 4).

Figure 7. Locations of Acmispon dendroideus var. traskiae as considered current in relation to the training areas on SCI, including the Impact Areas, the Training Areas and Ranges (TARs), the Assault Vehicle Maneuver Areas (AVMAs), the Infantry Operations Area (IOA), and the Shore Bombardment Area (SHOBA), which occupies the southern third of the island. Current Restricted Access Areas (RAAs) are also shown, but these change as unexploded ordnance are removed.

SSA Report – San Clemente Island Lotus 32 March 2020

Table 4. Summary of training areas, their size, use, and the threats to Acmispon dendroideus var. traskiae within each. Size Training area (Acres) % of island Use Threat vehicular Soil erosion, trampling, devegetation AVMAs (3) 1,060.5 2.9% maneuvering (habitat removal) IOA 8,827.6 24.5% dispersed foot traffic Trampling, soil erosion Varies by TAR: Varies by TAR, but limited to TARs (22) 1,968.2 5.5% demolition, small trampling, localized ground arms, combat, etc. disturbance Devegetation (habitat removal), fires Impact Areas (2) 3,399.7 9.4% bombing, live fire (accounted for separately)

Additionally, six near-shore Special Warfare Training Areas (SWATs) have been designated on and around SCI (Figure 7). These large areas encompass land, water, and associated airspace. They are used as ingress and egress of small troops to specific TARs. Basic and advanced special operations training is conducted within these areas by Navy and Marine Corps units (US Navy 2013a, p. 2.10; US Navy 2008a, p. 2.7). Thus, impacts from training in these areas is infrequent and dispersed (Booker 2019, pers. comm.). The Shore Bombardment Area (SHOBA) is the largest terrestrial training area and supports a diversity of military training (including Anti-Surface Warfare, Amphibious Warfare, Naval Special Warfare, Bombing Exercises, and Combat Search and Rescue) (Figure 7). SHOBA occupies roughly the southern third of the island and is approximately 13,824 ac (5,594 ha) (US Navy 2008a: Tables 2–7; US Navy 2009, p. 2–4). Areas of intensive use within SHOBA include the two Impact Areas and three TARs, which lie within the Impact Areas. Impact Areas support naval gun firing, artillery firing, and air-to-ground bombing (US Navy 2008a, p. 2–7; US Navy 2013a, p. 2–8). Collectively, the Impact Areas and TARs within SHOBA encompass 3,400 acres [1,376 ha], which amounts to 24.6% of the area within SHOBA. Much of the remainder of SHOBA serves as a surface danger zone (buffer) around Impact Areas I and II, and 59% of SHOBA is not within the (IOA), Impact Areas, or a TAR and therefore not subject to any direct training activities. Some areas, particularly the escarpment along the eastern coast, have limited training value because precipitous terrain hinders ground access. The Impact Areas sustain live fire, which is a recurrent source of fires. Most fires are of low severity, which does not have a strong negative impact to shrubs. Fuel breaks are applied each year prior to fire season to help prevent spread of fire to areas outside of the Impact Areas for protection of natural resources. Fire will be discussed in greater detail in section 5.4. Because parts of SHOBA are used for bombardment, access to this area is restricted for nonmilitary personnel on days when bombing is occurring. Individuals conducting surveys or working on invasive species control projects are granted access to areas outside of the Impact Areas within SHOBA when military activities requiring exclusive use are not occurring. Because of the frequency of training, access to SHOBA can be restricted for periods of time. The IOA encompasses approximately 25% of the island, the Impact Areas encompass about 9.4% of the island, TARs, which in places overlap the IOA, Impact Areas, and AVMAs, cover 5.5 % of the island, and the AVMAs, which fall entirely within the IOA, encompass about

SSA Report – San Clemente Island Lotus 33 March 2020

3% of the island (US Navy 2008a, pp. 2.17, 2.45; US Navy 2008b, pp. 3.11–3.52) (Table 4, Figure 7). Altogether, 34.8% of the island’s area is located in one of these training areas, although training does not occur uniformly within each; much of the island is void of any infrastructure. In comparison to many other military installations, there is a very low visual presence of the military on SCI (McFarland 2019, pers. obs.). In 2008, the Southern California Range Complex Final Environmental Impact Statement/Overseas Environmental Impact Statement (EIS/OEIS) (US Navy 2008a) and the accompanying Biological Opinion: San Clemente Island Military Operations and Fire Management Plan (BO) (USFWS 2008) were finalized, and together, these documents allowed for increased training at SCI and addressed obligations for fire management and listed species management (US Navy 2008a, p. 2.1–2.52). To avoid underestimating impacts and to ensure adequate coverage under all applicable federal laws and regulations, including but not limited to the National Environmental Policy Act (NEPA), the ESA, and the Clean Water Act, the analyses considered a training tempo that was at the highest reasonably anticipated level. It is unlikely that the maximum operational tempo will be reached for all activities simultaneously because overseas deployments, availability of personnel and assets, planning and construction timelines, development of platforms and systems, and other factors can lower the tempo and/or delay implementation; however, it was necessary to analyze the potential impacts of such a tempo (O’Connor pers. comm., 2019). Training began to increase soon after issuance of the 2008 BO and Record of Decision (ROD) for the EIS, but increases in some types of training, particularly those that required acquisition of assets, development of platforms and systems, and/or planning and construction, have increased more gradually, and some have not reached the operational tempo in the documents. One example of the latter type of activity is the battalion-sized landing planned to occur within the Assault Vehicle Maneuver Corridor, which will be discussed further in Section 4.2. In contrast to the AVMAs, the TARs (all except TAR 19) were fully developed and utilized shortly after issuance of the ROD and have been in use since. TARs 10 and 17 were of particular concern due their location on the west side of the island within high-density San Clemente Bell’s sparrow habitat and the introduction of new ignition sources to the west shore (O’Connor 2019, pers. comm.). After approximately 11 years of use, no fires have occurred in either of these TARs, and focused monitoring has not detected impacts to Bell’s sparrow from military training in either location (Meiman et al. 2018, p. 39). Range schedulers are aware of the natural resources obligations within SHOBA, and at least 1 day a week is usually allowed for natural resources programs to conduct their activities. Weeks with reduced natural resource access, including infrequent events that exclude natural resources personnel from SHOBA for 10 to 20 days, are announced in advance and provide natural resources managers the opportunity to plan accordingly. Impact Areas I and II have been indefinitely closed ‘‘for any purpose, including monitoring and management of endangered and sensitive species and their habitat’’ for safety reasons (US Navy 2008a, p. 2–45). Access to additional areas on the island where unexploded ordnance has been found is often restricted for natural resources personnel, but these areas are reopened once they are assessed by unexploded ordnance specialists and are deemed safe for entry (Figure 7). When closed, these restricted access areas (RAAs) can be accessed if accompanied by a trained unexploded ordnance technician (Munson 2019, pers. comm.). Due to these various military training activities, land use has been considered a threat to Acmispon dendroideus var. traskiae. Training and other land use activities have multiple

SSA Report – San Clemente Island Lotus 34 March 2020

potential impacts to A. d. var. traskiae, including disturbances to soil and vegetation, spread of nonnative plant species, creation of road ruts and trails, and compaction of soils (USFWS 2008, pp. 83–87). Potential training impacts vary by area (Table 4). However, only approximately 1% of the current population (twenty-two individuals located in 4 watersheds) lies within one of these training areas (Figure 7, Table 5). All locations are within the IOA or a TAR, and no current or past documented locations fall inside the boundaries of the AVMAs or Impact Areas, which are subject to heavier impacts. No A. d. var. traskiae exist within 50 ft of the AVMR. Moreover, some areas within the IOA and TAR are not readily accessible to vehicles and troops due to terrain that could support A. d. var. traskiae.

Table 5. The numbers of locations and total individuals that occur within each of the training area types based on the distribution of Acmispon dendroideus var. traskiae considered current. % of Locations Individuals Watersheds* population TAR 1 14 1 0.07% AVMA 0 0 0 0% IOA 3 8 3 0.04% Impact Areas 0 0 0 0% *Individual watersheds overlap multiple training areas (TAR watershed is also one of the Impact Area watersheds, etc.)

Military training activities within training areas (primarily the IOA, TARs, and AVMAs) can entail the movement of vehicles and troops over the landscape and thus include the potential of trampling or crushing individuals or groups of plants (USFWS 2012, p. 29114). Any effects of foot traffic on a local occurrence of this taxon would be dispersed (because the troops are spread out), minor (trampled leaves or broken branches), infrequent (up to twice per year, generally less) and temporary (USFWS 2008, p. 99; Vanderplank et al. 2019, p. 12). Further, few documented locations of Acmispon dendroideus var. traskiae occur within these areas (Table 5). The AVMAs, IOA, and several TARs are located along the central plateau or spine of the island, where few plants have been documented to date (see Figure 4). Because of the taxon’s close proximity to Navy facilities, military activities have the potential to impact habitat near Wilson Cove (USFWS 2012, p. 29110). All construction, maintenance, and training activities in the Wilson Cove area go through a site approval request process. Through this process, the areas are assessed to see if the activities will potentially impact any listed or sensitive species, including Acmispon dendroideus var. traskiae. One location in this area occurs within a TAR where tactical training and movement were projected to occur, possibly causing habitat damage through troop traffic (USFWS 2008, pp. 119–120). Work done at Wilson Cove around 2008 impacted habitat occupied by A. d. var. traskiae. The Navy successfully restored the site by installing erosion control materials and outplanting 50 individuals of A. d. var. traskiae propagated from locally-collected seed. The habitat has rebounded, and more plants are present in the area than before the work was done (O’Connor 2019, pers. comm.; USFWS 2012, p. 29110). The southern portion of Acmispon dendroideus var. traskiae’s distribution extends through SHOBA (USFWS 2012, p. 29116). No discernible change in the vegetation is observed from inside versus outside of SHOBA (McFarland 2019, pers. obs.; Uyeda 2019, p. 22). Because of the elevated risk of fire ignition associated with live and inert munitions fire are targeted

SSA Report – San Clemente Island Lotus 35 March 2020

towards Impact Areas I and II within SHOBA, fires are often concentrated within SHOBA (USFWS 2008, p. 11–13). However, A. d. var. traskiae is concentrated in SHOBA on the eastern escarpment, where it is protected by terrain from land use impacts and fire (which will be further discussed below). Although the designation of RAAs could make it more difficult to schedule surveys needed to assess potential training-related habitat impacts in some watersheds in the future (Figure 7), surveys are not precluded within the RAAs, and other federally listed species, including the San Clemente bush mallow (Malacothamnus clementinus) and the San Clemente loggerhead shrike (Lanius ludovicianus mearnsi), are monitored regularly within them.

Management efforts The Navy has demonstrated its efforts to help conserve and manage listed species on the island through amelioration of habitat impacts by military activities through implementation of the 2008 BO (USFWS 2008) and INRMP (US Navy 2013a), including invasive species control island-wide, including near listed species, biosecurity protocols, public outreach to promote compliance, restoration of sites that support sensitive plants, habitat enhancement for sensitive and listed species, fuelbreak installation to minimize fire spread, and fire suppression inside and outside of SHOBA to protect threatened, endangered, and other priority species (US Navy 2013a, p. 3.45; Vanderplank et al. 2019, pp. 15, 18-19; Munson 2019, pers. comm.). The site approval process for construction, maintenance, and training activities evaluates the impacts to and thus provides protection for many plant species, not just state and federally listed, but other sensitive species and all California Native Plant Society (CNPS) taxa with rankings of 1B through 4 (except CNPS taxa that are widespread and abundant, e.g., Calystegia macrostegia ssp. amplissima). Changes to training have and will be subject to environmental review under applicable laws and regulations, including NEPA and ESA, and impacts to federally listed and sensitive species will be addressed (O’Connor 2019, pers. comm.).

Summary About 35% of SCI is located within a training area (Table 4), and less than 1% of the current population of Acmispon dendroideus var. traskiae occurs in these areas (Table 5). Since not all of the land within each training area is heavily impacted by training, land use likely threatens even less of the distribution of A. d. var. traskiae (Table 5). While dispersed foot traffic is possible throughout the entirety of the IOA, we expect this buffer around the AVMR to be where any major impacts may exist. We will discuss this impact further in Section 4.2, concerning roads. Military activities on SCI have been and will continue to be dynamic as they evolve to meet new requirements. The Navy has begun evaluating the need for changes to the actions analyzed in the 2008 EIS and associated BO. Any changes are expected to be incremental, as they have been in the past. Such changes are subject to environmental review under applicable laws and regulations, including NEPA and ESA, and impacts to federally listed and sensitive species will be addressed. Coupled with ongoing management of related threats (including wildland fire, soil erosion, invasive species) under the SCI Integrated Natural Resources Management Plan, it is highly unlikely that future changes in military training on SCI will impede or reverse advances in the recovery of Acmispon dendroideus var. traskiae.

SSA Report – San Clemente Island Lotus 36 March 2020

4.2 Erosion and Roads Erosion and associated soil loss caused by degradation of the vegetation due to the browsing of feral goats and rooting of feral pigs modified the island’s habitat significantly and resulted in increased erosion over much of the island, especially on steep slopes where denuded soils could be quickly washed away during storm events (Johnson 1980, p. 107; Tierra Data Inc. 2007, pp. 6–7; US Navy 2013a, pp. 3.32–3.33). Since the feral animals were removed, much of the vegetation has recovered, and natural erosion on the island has decreased significantly (US Navy 2013a, p. 3-33, Vanderplank et al. 2019, p. 15). Erosion problems currently are limited to localized areas, and because of topography and soil characteristics, there always will be the potential for localized erosion to occur at sites across the island. Periods of heavy rainfall can cause localized erosion, but these areas are difficult to predict. Accelerated soil erosion from anthropogenic causes tends to occur at the heads of canyons, ephemeral drainages, and in areas where groundwater drainage causes “piping,” the formation of underground water channels, which occur on clay soils. Military training activities could lead to erosion that could impact Acmispon dendroideus var. traskiae, but few individuals occur in these designated training areas (Table 5) (Tierra Data Inc. 2007, pp. 1– 45). Gullying and other processes could concentrate surface runoff to unnatural levels, leading to accelerated erosion in the canyons below where A. d. var. traskiae occurs (Tierra Data Inc. 2007, p. 6), but the realized erosion on the island has been rare and localized (Munson 2019, pers. comm.). If erosion were to occur or be initiated in a single watershed, the effects would be confined to that watershed. Military activities have the potential to adversely affect Acmispon dendroideus var. traskiae habitat where it occurs in proximity to Navy facilities. Acmispon dendroideus var. traskiae occurs in Wilson Cove, where erosion has been evident, caused by the construction of buildings and parking lots (USFWS 2008, p. 117). While no individuals have been documented to be affected by erosion in this area, areas of erosion persist in this area despite management efforts (SERG 2015, p. 40). As mentioned in Section 4.1, individuals affected by other human impacts have been successfully outplanted in this area. Roads can concentrate water flow, causing incised channels and erosion of slopes (Forman and Alexander 1998, pp. 216–217). Along the eastern escarpment, Acmispon dendroideus var. traskiae is found in steep canyons in proximity to Ridge Road, the primary road that traverses most of the island from northwest to southeast. Roadside occurrences of A. d. var. traskiae may experience runoff during storm events (US Navy 2008a, pp. G.4, G.8). Increased erosion near roads could potentially degrade habitat, especially along the steep canyons and ridges. On occasion after particularly heavy rainfall events, localized areas of high erosion stemming from roadways have been noted; however, regular road maintenance and repair of associated damage minimizes the potential for such problems to spread, and erosion impacts to A. d. var. traskiae from such events have never been observed and are unlikely. In the downlisting rule, it was proposed that A. d. var. traskiae that occur within 500 ft (152 m) of a paved or unpaved road could be subject to road effects that degrade the habitat quality (Forman and Alexander 1998, p. 217; USFWS 2013 p. 45419). However, based on expert opinion and observations on San Clemente Island since 2012, the likelihood of impacts to A. d. var. traskiae from runoff or other erosional processes stemming from roads is very low, and increased erosion associated with roads is not evident as far from the road network as analyzed in the proposed rule (O’Connor 2019, pers. comm.). The conditions of roads on San Clemente Island are evaluated and maintained, which minimizes the potential for habitat impacts adjacent to them, particularly

SSA Report – San Clemente Island Lotus 37 March 2020

as the distance from the roads increases. Erosion stemming from a roadway would have a high chance of being noticed early and managed accordingly. Still, as a precaution, the expert team, consisting of both USFWS and US Navy biologists, decided that a 100-ft (30-m) buffer around roads is a more appropriate distance over which negative impacts to habitat could be perceptible and should be evaluated. Likewise, the likelihood of crushing individual plants during military training involving foot traffic in maneuvers along roads tends to be more concentrated close to roads and decreases farther away (see Section 4.1) (O’Connor 2019, pers. comm.). Additionally, other indirect effects associated with roads, such as the introduction and spread invasive plants (see Section 4.3), are most likely alongside roads. It is important to note that much of the invasive species control on San Clemente Island is focused along roads for this reason, and new occurrences of invasive species generally are more likely to be noticed and eradicated if they occur adjacent to roads. The expert team applied the same distance (100 ft [30 m]) when considering these other potential mechanisms of habitat degradation associated with roads. Of the distribution considered current, 434 individuals in 6 watersheds are located within 30 m (100 ft) of a road (Table 6). Island-wide, this represents 2% of the total locations and 2% of the total individuals. Locations that could be affected by road impacts (including trampling, erosion, and increased invasive species) exist within 5 watersheds. Only one of these has 100% of their individuals located near a road, and all of the rest have fewer than 20% of the individuals or locations in areas considered in this assessment to be at risk of road impacts (Table 6).

Table 6. Locations and individuals by watershed located within 30 m (100 ft) of a road, and what percent of the total locations and total individuals that represents that could potentially be affected. Locations Individuals Percent Percent Watershed Near Near Total Total Locations Individuals ID Roads Roads Locations Individuals Affected Affected WS_1006 2 427 25 2,141 8% 20% WS_1058 1 4 22 624 5% 1% WS_1068 1 1 17 558 6% 0% WS_1116 1 1 9 176 11% 1% WS_1158 1 1 1 1 100% 100%

Management efforts The Navy monitors and evaluates soil erosion on SCI and uses multi-year data to assess priorities for remediation (SERG 2006, entire; SERG 2015, entire). Efforts are made to restore areas where erosion occurs, through revegetation efforts and the installation of erosion control materials (SERG 2016, p. 2). The Navy incorporates erosion control measures into all site feasibility studies and project design to minimize the potential to exacerbate existing erosion and avoid impacts to listed species The INRMP requires that all projects include erosion control work (US Navy 2013a, p. 3–33). These conservation actions include best management practices, choosing sites that are capable of sustaining disturbance with minimum soil erosion, and stabilizing disturbed sites (US Navy 2013a, pp. 3.33–3.37). Originally, the AVMAs were to allow for the most extensive off-road movement of tracked vehicles, and the area within them was anticipated to experience increased soil erosion due to reductions in vegetation cover (Vanderplank et al. 2019, p. 16). While no Acmispon dendroideus var. traskiae currently occur in the AVMAs or associated watersheds, erosion in

SSA Report – San Clemente Island Lotus 38 March 2020

these areas could affect future locations of the taxon. To address soil erosion within the AVMC, the US Navy included a conservation measure in the 2008 EIS to develop an erosion control plan for portions of the AVMC, including the AVMAs, Artillery Firing Points (AFPs), Artillery Maneuver Points (AMPs), and IOA, that would accomplish the following: (1) minimize soil erosion within these training areas and minimize offsite impacts; (2) prevent soil erosion from adversely affecting federally listed or proposed species or their habitats; and (3) prevent soil erosion from significantly impacting other sensitive resources, including sensitive plant and wildlife species and their habitats, jurisdictional wetlands and non-wetland waters, the Area of Special Biological Significance surrounding the island, and cultural resources. The plan includes specific guidelines for the development and application of best management practices (BMPs) to minimize impacts to sensitive resources, including A. d. var. traskiae and its habitat, site-specific erosion control recommendations, restrictions of vehicle maneuvering when soils are wet, operator education, vegetation management measures, methods to prevent gully development and restore existing gullies, and an adaptive management and monitoring plan to assess the effectiveness of and modify BMPs as needed (US Navy 2013b, p. 37-50, 111; Vanderplank et al. 2019, p. 16). The plan is prescriptive; measures have been or will be implemented prior to use of areas to which they apply. Following issuance of the BO and signature of the ROD, funding was secured, the Erosion Control Plan was developed, and the final plan received concurrence from the US Fish and Wildlife Service (O’Connor 2019, pers. comm.). Subsequent to finalization of the Erosion Control Plan in 2013, development of the AVMAs has involved working with military operators to determine more precisely how areas would be used based on findings and recommendations in the plan, and implementation has taken a phased approach. This effort has resulted in the delineation of unpaved roads to channel vehicle traffic through some portions of the AVMAs, which will substantially reduce the level of ground disturbance from those anticipated in the EIS. Design of these roads will be the focus of future planning efforts. Areas in which roads are being developed have not been used and a battalion sized landing has not been conducted yet, but platoon sized landings involving 14 AAVs have been conducted approximately quarterly in other portions of the AVMAs, the AFPs, the AMPs, and the IOA (O’Connor 2019, pers. comm.). While the Erosion Control Plan is unlikely to have direct conservation benefits to Acmispon dendroideus var. traskiae at this time, the effort will help reduce erosion and contribute to the overall island health.

Summary Despite existing levels of soil erosion on the island, the distribution of Acmispon dendroideus var. traskiae has increased since listing. Current erosion issues are localized, and erosion is generally decreasing on the island as the vegetation continues to recover. The Navy incorporates erosion control measures into all projects to minimize the potential to exacerbate existing erosion and avoid impacts to habitat and listed species. Although the erosional processes and potential related threats to A. d. var. traskiae must be considered at an island-wide scale, because erosion impacts are localized and managed, the loss of individuals due to erosion is unlikely.

4.3 Invasive plants Contemporaneous with and likely aided by feral grazing animals, a large number of invasive non-native plant species have become naturalized on SCI. At listing, the spread of nonnative plants was identified as a threat to the recovery of Acmispon dendroideus var. traskiae (USFWS 1977, p. 40682, 40684) and nonnative plants are considered an island-wide threat to the

SSA Report – San Clemente Island Lotus 39 March 2020

native vegetation (USFWS 2012, p. 29117). Nonnative plants can alter habitat structure and ecological processes such as fire regimes, nutrient cycling, hydrology, and energy budgets, and they can compete with native plants for water, space, light, and nutrients (USFWS 2012, p. 29117). Non-native annual grasses noted at the end of the grazing period are now widespread on SCI with the most-common being Avena barbata (slender wild oat), Bromus madritensis ssp. rubens (red brome), B. hordeaceus (soft brome), B. diandrus (ripgut brome) and Hordeum murinum (false barley) (Keeley and Brennan 2015, p. 4). The invasion of nonnative annual grasses on the island may have caused the greatest structural changes to Acmispon dendroideus var. traskiae habitat, especially in the coastal terraces and swales (USFWS 2007, p. 4-5). Annual grasses vary in abundance with rainfall, potentially changing the vegetation types from shrublands to grasslands and increasing the fuel load in wet years (Battlori et al. 2013, p. 1119). Although most of the invasive species likely were brought to the island while it was being ranched, invasions by previously undocumented non-native grasses continued to be found on SCI; e.g., the discovery of Schismus sp. (Mediterranean grass) and the fire-tolerant weeds Brachypodium distachyon (purple false brome) (USFWS 2007, p. 5), Ehrharta calycina, and E. longiflora (African veldt grasses) (US Navy 2013a, p. 3-90). A brief review of the occurrence data collected in 1996 and 1997 reveals that Acmispon dendroideus var. traskiae was associated with non-native annual grasses in 69 percent of its locations (45 of 65), but these grasses had approximately 40% cover and were therefore not dominant (Junak and Wilken 1998, p. 261; USFWS 2007, p. 6-7; Vanderplank et al. 2019, p. 12). Surveys conducted in 2011 and 2012 did not find A. d. var. traskiae in communities dominated by nonnative grasses (Vanderplank et al. 2019, p. 12). Populations of Acmispon dendroideus var. traskiae within 500 ft (152 m) of roads may be subject to effects that degrade the habitat quality along the road (Forman and Alexander 1998, p. 217). This disturbance along roadsides tends to create conditions (high disturbance, seed dispersal from vehicles, ample light and water) preferred by nonnative species (Forman and Alexander 1998, p. 210). Nonnatives, including Foeniculum vulgare (fennel) and Mesembryanthemum crystallinum (crystalline iceplant), have been found in the disturbed shoulders along China Point road in SHOBA (Braswell 2011, pers. obs.), but these nonnatives have not been considered a threat to locations of A. d. var. traskiae. Potential impacts of nonnative plants on Acmispon dendroideus var. traskiae include precluding germination (i.e., competitive exclusion), preventing pollination (e.g., A. d. var. traskiae plants are not obvious to pollinators due to tall stands of non-native grasses), and carrying fire in areas that would not otherwise burn. While there may be these or other unquantified indirect effects to the fitness of A. d. var. traskiae due to the invasive species already present on the island, they do not seem to be impeding the population growth of A. d. var. traskiae. Further, since the removal of feral grazers and browsers, the vegetation has been recovering and is no longer comprised of the early seral communities, the first to colonize disturbed areas (Stratton 2005, p. 216). The island has more intact habitats, reduced erosion, and a stronger suite of native competitor species, making the conditions less favorable to invasion. Some habitats that underwent considerable invasion historically, such as the central grasslands, are still heavily dominated by non-native species, but rocky soils, which support A. d. var. traskiae, are less susceptible to invasion by annual grasses (Allan 1999, in Vanderplank et al. 2019, p.12).

SSA Report – San Clemente Island Lotus 40 March 2020

Management efforts The Navy has monitored and controlled the expansion of highly invasive non-native plant species on an ongoing basis since the 1990s (O’Connor 2019, pers. comm.), and primary target species have included Brassica tounefortii (Saharan mustard), B. nigra (black mustard), Foeniculum vulgare (fennel), Asphodelus fistulosus (aspohodel), Stipa milaceae (smileo grass), Ehrharta calycina (African veldt grass), Plantago coronopus (buckhorn plantain), Tragopogon porrifolius (salsify), and Caprobrotus edulis (iceplant); additional priority species may also be controlled as they are located (e.g., SERG 2016, p. 45-46). In general, the Navy treats over 100,000 individuals of these various species annually. Control of these invasive plants benefits the ecosystem on SCI by reducing their distribution and minimizing the potential that they will invade habitat occupied by listed and at-risk taxa. Because invasive species introductions are more likely to occur along roadsides and because roads function as corridors for the spread of invasive species propagules, much of the invasive species treatment on the island focuses on roadsides; however, other areas highly susceptible to invasive species introductions (such as graded areas, soil stockpiles, and mowed areas) also are focal areas for control. High-priority invasive plants are treated at locations across the island. This control strategy has minimized the need to treat invasive plant species within areas occupied by federally listed plants. While these targeted invasive species do not generally occur near Acmispon dendroideus var. traskiae, the iceplant is routinely pulled by hand in localized areas where they occur in A. d. var. traskiae habitat, and A. d. var. traskiae has been documented moving into the clearings, along with other native vegetation (Vanderplank et al. 2019, p. 13; O’Connor 2019, pers. comm.). While many conservation measures to limit the introduction and spread of nonnative plants are included in the INRMP (US Navy 2013a, pp. 3.289–3.290) and required in the 2008 BO (USFWS 2008, pp. 58–66) (Table 15 in Appendix B), the recently-completed Naval Auxiliary Landing Field San Clemente Island Biosecurity Plan (US Navy 2016, entire) will help more effectively control the arrival of potentially invasive propagules than similar plans on nonmilitary islands. The plan works to prevent and respond to new introductions of non-native species and bio-invasion vectors. Through implementation of this plan and the ongoing island- wide nonnative plant control program, potential impacts from nonnative plants are expected to be minimized (O’Connor 2019, pers. comm.; Munson 2019, pers. comm.).

Summary Non-native species are extensively distributed across SCI both as a result of post-grazing colonization of weedy species in highly disturbed habitat and accidental introduction of new weeds which may inevitably occur either naturally or inadvertently through human activities. However, all vegetation communities have been recovering on SCI since the final removal of feral grazers, and naturalized grasslands (the most fire-prone of non-native vegetation communities) constitute a small proportion of the island at this time. Non-native annual and perennial grasses, however, are widespread on the island and have been for many decades. No assessment to track weediness within occupied habitat areas has been done. While there is the potential that exotic annual grasses, which are widespread on the island, could affect fire- regimes, but it does not appear as if these grasses are expanding, and they have been present during the recorded fire history. This potential is further addressed in Section 4.4. The Navy makes significant efforts to control highly invasive non-native perennial grasses and non-native forbs to preclude their expansion into habitat areas and areas in which weed control would be difficult due to terrain and access challenges. Because data on the

SSA Report – San Clemente Island Lotus 41 March 2020

abundance and density of invasive plants and monitoring of change over time are lacking, we have no way to quantitatively assess the effect of non-native, invasive species on Acmispon dendroideus var. traskiae individual fitness; however, A. d. var. traskiae’s range has expanded despite the presence of nonnative plants on SCI.

4.4 Fire The history of fire on the island prior to 1979 is largely unknown, but while the island was used for ranching, fires were set intermittently to increase the cover of forbs and grasses (US Navy 2009, p. 3-2; US Navy 2013a, p. 3-47). After the island was purchased by the U. S. Department of the Navy in 1934, however, fire became a more common occurrence throughout much of the island. Fire history for most of the island has been documented since 1979. Since that time, over 50 percent of the island has experienced at least one wildfire with smaller areas on the island having burned up to ten times between 1979 and 2018 (US Navy 2013a, p. 3-47; US Navy, unpublished data). The number and extent of fires (acres burned) varies annually as does fire-severity (Figure 8). Most large fires are ignited in the Impact Areas, and thus, the majority of acreage that has burned has been concentrated in SHOBA (US Navy 2013a, p. 3-45). Most of these fires are classified as a severity of 4 or 5, considered lightly burned or scorched, which have little effect on shrubs (Table 7) (US Navy 2009, p. 4-52). For fires with associated severity data (2007 to present), 15.6% of the area burned has been of a severity class that has detrimental effects on shrubs, class 1 through 3, considered completely-burned to moderately-severe (Figure 11). The largest area that burned at these severities burned in 2017 (Figure 8). Typically, due to the patchy nature of fires, not all areas within a fire footprint are burned uniformly; therefore, not all plants in a burn polygon are necessarily burned or burned at the same severity (SERG 2012, p. 39).

Table 7. Fire severity classes and definitions, reproduced from the US Navy 2009 Fire Management Plan for SCI, with severity classes adapted from the National Park Service (1992). Effects on Effects on Effects on Fire severity class litter/duff herbs/grasses shrubs Effects on trees 1 Completely Burned Burned to ash Burned to ash Burned to ash, Burned to ash or few resprouts killed by fire 2 Heavily Burned Burned to ash Burned to ash Burned to ash, Killed by fire or some resprouts severely stressed 3 Moderately Burned Burned to ash Burned to ash Burned to Crown damage singed, some only to smaller resprouts trees 4 Lightly Burned Blackened, but Burned to ash, Singed/stressed, No effect on mature not evenly some resprouting many trees, may kill converted to ash resprout/recover seedlings/saplings 5 Scorched Blackened Singed/stressed, Not affected, No effect on trees many slight stress resprout/recover 6 Unburned* – – – – *Unburned inclusions within a fire should be marked as 6.

SSA Report – San Clemente Island Lotus 42 March 2020

Although fire was not considered a threat to habitat occupied by Acmispon dendroideus var. traskiae at the time of listing (USFWS 1977, p. 40682), intense or frequent fires can threaten individuals of A. d. var. traskiae. High fire frequency may be a potential threat that could limit the distribution of A. d. var. traskiae by overwhelming its tolerance threshold (Brooks et al. 2004, p. 683; Jacobson et al. 2004, p. 1). At higher than natural fire frequencies, fire has the potential to exceed a plant’s capacity to persist by depleting seed banks and reducing reproductive output (Zedler et al. 1983, pp. 811–815). A fire return-interval of three years or fewer has been shown to negatively impact woody shrubs on SCI (Keeley and Brennan 2015, p. 3). The historical fire return interval on SCI is unknown but a majority of the areas that have experienced more than one fire in the last 20 years (1999–2018) are located on the central plateau in SHOBA (Figure 10). Very few areas where Acmispon dendroideus var. traskiae occur have burned more than once in the last 20 years. Island-wide during this time, fires have burned a portion of 88 watersheds. Of the distribution considered current, 26 of 249 locations totaling 855 individuals in 10 occupied watersheds are located in an area that had a fire in the last 20 years (Figure 10). Thirteen occupied watersheds had 50% or more of their area burn in the last 20 years, and 5 watersheds had 50% or more of their area burn twice or more in the last 20 years (see Table 14 in Appendix A). Twelve locations, totaling 118 individuals, are located in an area that burned twice or more in the last 20 years. No A. d. var. traskiae burned more than 3 times in the last 20 years. Thus, fire return intervals have not exceeded the postulated ecological tolerance of A. d. var. traskiae in areas where the taxon occurs. While increased fire frequency could lead to localized changes in vegetation on SCI, fires are currently relatively infrequent across the majority of the island, even given the intensified military uses over the last decade (Figure 10). While fires are occasionally ignited by activities north of SHOBA, the risk of outbreak of frequent fire is higher in Impact Areas I and II and within SHOBA (USFWS 2008, p. 50; US Navy 2013, pp. 3.45–3.47; USFWS 2018, GIS data). The distribution of Acmispon dendroideus var. traskiae within SHOBA is mostly along the eastern escarpment, away from the Impact Areas and downslope, reducing the potential for frequent fire. However, in 2017, a large fire burned part of the eastern escarpment within SHOBA, where no other recorded fire has burned. After having seemingly gone out, the fire restarted the next day and response was therefore delayed, which has prompted a change to monitoring fires that are thought to be out (O’Connor 2019, pers. comm.). The fire burned 1,522 acres, almost all (98%) of which were of moderate to high severity (3, 2, or 1 severity class) (Figure 11).

SSA Report – San Clemente Island Lotus 43 March 2020

Figure 8. Total acres that have burned annually in wildfires and acres that have a moderate to high severity (severity classes 1, 2, or 3).

The outline of this 2017 high severity fire encompasses encompasses 634 individuals of Acmispon dendroideus var. traskiae in 9 watersheds of the distribution considered current; vegetation plot monitoring in 2019 has indicated that some of the plants have persisted or are resprouting here (SERG 2019, unpublished data). The response of A. d. var. traskiae following the 2017 higher severity fire on the eastern escarpment is difficult to interpret; surveyors traversed the burn and attempted to relocate 9 previously-known locations totaling 345 individuals that burned at severity 2 or 3 in 2017. Because locations are an approximate point representation of the center of a group of plants, we were unable to match up specific locations. However, surveyors located 106 individuals of A. d. var. traskiae at 7 locations. It is unknown if or how many other individuals were missed, as the terrain and vegetation limits visibility and can impede detection. Tracks show that surveyors were within 10 m of original documented locations, but given the intrinsic GPS error and the nature of a location definition, they may have been further away in some cases. The majority of the plants were noted to be large plants that existed during the burn; surveyors estimated <5% were new recruits, but specific data on plant age were not collected. Surveyors located an additional 261 individuals at 19 locations outside the burn polygons that were previously unknown. These data indicate that many individuals that burned, up to 75%, were lost in the fire. However, the data also indicate that some, although seemingly few, new recruits have become established after the fire burned. The data also indicate that there are many individuals on the landscape in 2019 that have not experienced fire that were previously unknown, indicating that the current population is likely larger than currently estimated. The Navy has significantly expanded the number of locations where live fire and demolition training can take place (USFWS 2008, pp. 21– 37). In addition to demolitions, certain proposed munitions exercises involve the use of incendiary devices, such as illumination rounds, white phosphorous, and tracer rounds, which pose a high risk of fire ignition. However, the number of acres that burn annually varies greatly, and the biggest fire years in the last 15 years (2012 and 2017) have burned less than half the acreage of the biggest fire years between the time of listing and now (Figure 9). Fires topping 8,000 acres burned in 1985 and 1994, before fuel

SSA Report – San Clemente Island Lotus 44 March 2020

breaks were routinely installed and other fire management was carried out (O’Connor 2019, pers. comm.). Although fire ignition points are concentrated in the military training areas, fires that escape these areas could potentially spread to other areas of the island, but due to vegetation and topography, these fires are usually confined to the same small areas (Munson 2019, pers. comm.). However, fires that escape from training areas are not likely to disturb the entire distribution of Acmispon dendroideus var. traskiae at one time because this taxon is widely distributed and associated with steep canyon areas where fires are less likely to impact the plant (USFWS 2012, p. 29121). One difficulty in assessing the interaction between fire and distribution of Acmispon dendroideus var. traskiae is that little survey data exists in the burn polygons in the years prior to and immediately following fires. Apart from the more-recent fire data (i.e., from 2012–2017), we do not know whether A. d. var. traskiae occurred in a burned area at the time of the fire. While much of the range of A. d. var. traskiae has burned since 1979, much of the survey effort has occurred after 1996. Using the historical datasets, we found a total of 96 of 249 locations in an area that had a documented fire since 1979. However, the vast majority of these locations (90) either were in places where the last fire had burned more than 10 years before the survey was recorded or the fire happened after the point was recorded. We did find six locations in four watersheds totaling 146 individuals are in areas where a fire burned 10 or fewer years before (Table 8). While this provides some indication that individuals can survive fires, it is still unknown whether the rarity of locations in recently burned areas is because A. d. var. traskiae is unlikely to survive a fire, will only reestablish itself from the seed bank in an area after a certain number of years post fire, or are a product of the survey efforts. The lower abundance of plants in fire-prone regions could be due to fire (fire precludes the establishment of plants or affects their survival), or A. d. var. traskiae may just naturally occur less frequently in habitats where fires are prone to occur (the western slope and the central plateau). . Table 8. Location points and individuals counted at points where a fire had burned within the past 10 years. Watershed Count Year of Years ID Individuals Year Last Fire Since Fire WS_1082 6 2004 1994 10 WS_1116 1 2013 2009 4 WS_1116 106 2008 2000 8 WS_1116 14 2010 2000 10 WS_1137 18 2009 1999 10 WS_1158 1 2013 2010 3

SSA Report – San Clemente Island Lotus 45 March 2020

Figure 9. Acres burned annually for years where fires were estimated since listing.

SSA Report – San Clemente Island Lotus 46 March 2020

Figure 10. Locations of Acmispon dendroideus var. traskiae (ACDET) points considered current in relation to areas where fires have burned in the last 20 years (1999–2018, after the initiation of fire management), including number of fires in that time.

SSA Report – San Clemente Island Lotus 47 March 2020

Figure 11. Locations of Acmispon dendroideus var. traskiae (ACDET) points considered current in relation to areas where fires where severity data is known have burned (2007–2018). Severity categories 1, 2, and 3 have the potential to burn shrubs where they will not resprout; severity categories 4 and 5 have little to no effect on shrubs.

Management efforts The San Clemente Island Wildland Fire Management Plan (FMP) stipulates monitoring of live fuel-moisture and establishes a threshold below which training requirements are altered to reduce ignition risk (US Navy 2009, p. 4.15–4.16). These live fuel moisture levels, combined with wind speed, define a fire danger rating which at various levels indicate specific munitions that are allowable and precautions that must be in place (standby firefighting engine, crew, and other resources, helicopter on fire alert, etc.) (US Navy 2009, p. 4.19). The FMP stipulates that Acmispon dendroideus var. traskiae is a management focus plant, such that individuals are given special consideration and protection from fires (US Navy 2009, p. 4.10).

SSA Report – San Clemente Island Lotus 48 March 2020

The Navy’s fire management program maintains ground and aerial suppression assets to fight fires in all areas outside of Restricted Access Areas and Impact Areas (US Navy 2013a, p. 3.45) (see Figure 7). While most fires burn themselves out in a short amount of time, fires are monitored closely after ignition (Munson 2019, pers. comm.). If a threat is perceived to lives, structures, or sensitive species, the fire is fought unless there is a threat of unexploded ordnance or another a safety risk, such as high winds. Since the 2017 fire on the eastern escarpment, monitoring for complete extinguishment has increased, and increased monitoring requirements will be included in all future versions of the FMP (Munson 2019, pers. comm.). The US Navy has constructed fuelbreaks around the Impact Areas to manage the spread of fire out of the Impact Areas (USFWS 2012, p. 29118). However, these fuelbreaks rarely have helped contain a fire as fires have infrequently approached them, and those that did were only sometimes contained by the fuelbreak. Thus, fuelbreak locations and installation methods have changed over time, and for the 2019 fire season, fuelbreaks were installed only along the existing roadways (Munson 2019, pers. comm.). As roads already serve as good fuelbreaks, increasing the width of this vegetation gap through application of a fire retardant along the existing roadways creates a more effective fire management tool. These fuelbreaks were designed with the protection of the sensitive species and resources on the eastern escarpment, which is protected by Ridge Road, in mind (Munson 2019, pers. comm.). Maintenance of these fuelbreaks reduces the likelihood and frequency of fires spreading to sensitive areas and habitats, such as those occupied by Acmispon dendroideus var. traskiae. Fuelbreaks on SCI are created using herbicides and strip burning and are maintained using herbicides and fire retardant (Phos-Chek D75F) (USFWS 2008, pp. 97–98). The Navy avoids application of Phos-Chek within 300 ft (91.4 m) of mapped listed species locations to the extent that has been allowable with previous fuelbreak installation (USFWS 2008, pp. 97–98). The Navy conducts preseason briefings for firefighting personnel on the guidelines for fire suppression and limitations associated with the use of Phos-Chek (USFWS 2008, pp. 97–98). To minimize the potential for effects to listed species, the Navy considers the documented locations of listed species on the island as fuelbreak lines are developed (USFWS 2012, p. 29119). The Navy also conducts annual reviews of fire management and fire occurrences that allow for adaptive management and aim to minimize the frequency and spread of fires that could result in loss of individuals of Acmispon dendroideus var. traskiae (USFWS 2012, p. 29121).

Summary Fire poses a threat to individuals of Acmispon dendroideus var. traskiae, as fires have the potential to burn most places on the island, but land use, vegetation, and historical patterns indicate that fires are most likely to burn in the same areas they have historically. Fires have been generally localized, infrequent, and of low severity, and most have burned in regions were A. d. var. traskiae is not documented. While the 2017 fire was in an unprecedented location and of an unprecedented size and severity, efforts are in place to keep this type of incident from happening again. Severe fires can kill plants, but A. d. var. traskiae may benefit from periodic fire, and individuals have been noted to survive and resprout after fire; however, an increase in fire frequency or high severity fires could be detrimental to A. d. var. traskiae. Although the Navy continues to work to minimize any adverse impacts of fire on A. d. var. traskiae through adaptive management and implementation of its Wildland Fire Management Plan, the Navy has managed fire sufficiently to allow the taxon to increase significantly in distribution and abundance.

SSA Report – San Clemente Island Lotus 49 March 2020

4.5 Climate Change Since the listing of Acmispon dendroideus var. traskiae, the potential impact of ongoing, accelerated climate change has become a recognized threat to the flora and fauna of the United States (IPCC 2007a, pp. 1–52; PRBO 2011, pp. 1–68). Climate change is likely to result in warmer and drier conditions, with high overall declines in mean seasonal precipitation but with high variability from year to year (IPCC 2007, pp. 1–18; Cayan et al. 2012, p. ii; Kalansky et al. 2018, p. 10). SCI is located in a Mediterranean climatic regime with a significant maritime influence. Current models suggest that southern California will likely be adversely affected by global climate change through prolonged seasonal droughts and rainfall coming at unusual periods and different amounts (Pierce 2004, pp. 1–33, Cayan et al. 2005, pp. 3–7, CEPA 2006, p. 33; Jennings et al. 2018, p. iii; Kalansky et al. 2018, p. 10). Climate change models indicate a 4 to 9 degrees Fahrenheit (2 to 4 degrees Celsius) increase in average temperature for the San Diego Area of southern California by the end of the century (Jennings et al. 2018, p. 9), with inland changes higher than the coast (Cayan et al. 2012, p. 7). By 2070, a 10 to 37 percent decrease in annual precipitation is predicted (PRBO 2011, p. 40; Jennings et al. 2018, p. iii), though other models predict little to no change in annual precipitation (Field et al. 1999, pp. 8–9; Cayan et al. 2008, p. S26). SCI typically receives less rainfall than neighboring mainland areas (Tierra Data Inc. 2005, p. 4). However, predictions of short and long-term climatic conditions for the Channel Islands remain uncertain. It is unknown at this time if climate change in California will result in a warmer trend with localized drying, higher precipitation events, and/or more frequent El Niño or La Niña events (Pierce 2004, p. 31). Low-level temperature inversions are common along the California coast and Channel Islands, and these inversions form low cloud cover (fog), otherwise known as the marine layer, which has a strong influence on coastal ecosystems and SCI (US Navy 2013a, pp. 3.13, 3.26). Although the island has a short rainy season, the presence of fog during the summer months helps to reduce drought stress for many plant species through shading and fog drip, and many species are restricted to this fog belt (Halvorson et al. 1988, p. 111; Fischer et al. 2009, p. 783). Thus, fog could help buffer species from extinction brought on by climatic change, as evidenced by the elevated levels of endemism along the coast of Baja California and on the Channel Islands (Vanderplank 2014, p. 5). Climate on the Channel Islands continues to support paleoendemic plants, such as Lyonothamnus, which once was widespread in the southwest of North America and is thought to have been extirpated on the mainland as conditions became warmer and drier (Bushakra et al. 1999, pp. 473-475). However, coastal fog has been decreasing in southern California in recent decades, possibly due to urbanization (which would not affect SCI) or climate change (Williams et al. 2015, p. 1527; Johnstone and Dawson 2010, p. 4537; LaDochy and Witiw 2012, p. 1157), and costal cloud cover and fog are poorly addressed in climate change models (Qu et al. 2014, p. 2603-2605). Warming projections in California, particularly the possibility that the interior will experience greater warming than the coast (Cayan et al. 2012, p. 7), suggest that the fate of coastal fog is uncertain (Field et al. 1999, pp. 21–22; Lebassi-Habtezion et al. 2011, p. 8-11). Iacobellis et al. (2010, p. 129), however, showed an increasing trend in the strength of low-level temperature inversions, which suggests that the marine layer is likely to persist and may even increase. Recent work examining projected changes in solar radiation and cloud albedo show projected increases in cloud albedo during the dry season (July–Sept) and decreases during the wet season (Nov–Dec, Mar–Apr) (Clemesha 2020, entire). The summer projections mean an

SSA Report – San Clemente Island Lotus 50 March 2020

increase in fog and low clouds the decreases in the winter likely reflect a decrease in a combination of precipitation and fog (Clemesha 2020, pers. comm.; Clemesha 2020, entire). Such a scenario would moderate the effects of climate change on the Channel Islands and would be expected to reduce its potential threat to island plants, including Acmispon dendroideus var. traskiae. Dry season low clouds and fog are particularly important to plant growth, survival and population dynamics in arid systems through both a reduction in evapotranspiration demand and potentially water deposition (Corbin et al. 2005, p. 511, Johnstone and Dawson 2010, p. 4533, Oladi et al. 2017, p. 94). Predicting impacts to Acmispon dendroideus var. traskiae due to climate change are further complicated by the timing of increased or decreased rainfall; wetter conditions in the winter and early spring can lead to more growth early in the season which can provide more fuel for fire later. However, wetter summers and falls can prevent the fuel from drying out enough to burn (Lawson 2019, pers. comm.). Changes in temperature or rainfall patterns also has the potential to affect biotic interactions, such as decoupling the timing of plant phenology versus insect activity. Therefore, making predictions about future fire patterns as affected by climate change is difficult. We focus on a 20 to 30-year window, in which we do not expect major impacts to A. d. var. traskiae from these long-term effects of climate change. However, in this short-term 20 to 30-year window, climate change may result in more frequent or severe fires, heavy periods of rainfall that could lead to major erosion events (see Section 4.2), or periods of drought (Kalansky et al. 2018, p. 10).

Summary The impacts of predicted future climate change to Acmispon dendroideus var. traskiae remain unclear. While we recognize that climate change is an important issue with potential effects to listed species and their habitats, information is not available to make accurate predictions regarding its effects to A. d. var. traskiae (USFWS 2012, p. 29121). However, given the timeframe presented in climate change studies, a major impact on A. d. var. traskiae from climate change is unlikely to occur in the next 20 to 30 years.

4.6 Hybridization As discussed above in Section 2.1, Acmispon dendroideus var. traskiae is known to hybridize with Acmispon argophyllus var. argenteus. In 1990, Liston et al. (p. 240) confirmed hybridization between co-occurring populations of A. d. var. traskiae and A. argophyllus var. argenteus in Wilson Cove. At that time, they detected only four hybrid individuals out of 38 individuals tested and failed to detect hybridization in another area of co- occurrence at the southern end of the island. Although hybrid individuals seem to be restricted to Wilson Cove (Liston 1990, p. 240; Allan 1999, p. 91), other unconfirmed hybrids (no genetic testing done) have been observed elsewhere on the island (Howe 2009b, pers. comm.; Braswell 2011, pers. obs. in USFWS 2013, p. 45426). Recent genetic work (McGlauglin et al. 2018, p. 754) has shown moderate levels of genetic diversity in A. dendroideus var. traskiae, with gene flow between neighbor populations and little threat from hybridization with other Acmispon species (Wallace et al. 2017, p. 743). Because hybridization is infrequently documented and is not expected to occur at higher frequencies than occurred historically, we do not have evidence that hybridization is a threat to A. d. var. traskiae.

SSA Report – San Clemente Island Lotus 51 March 2020

4.7 Summary of Factors Influencing Viability The largest threat to Acmispon dendroideus var. traskiae and its habitat, grazing and rooting by feral herbivores, was removed by 1992 (Keegan et al. 1994, p. 60), and the island and its native habitats have recovered considerably. The habitat for A. d. var. traskiae is still threatened by destruction and modification associated with land use, erosion, the spread of nonnatives, fire, and fire management (USFWS 2012, p. 29119). To help ameliorate these threats, the Navy implements a fire plan (US Navy 2009) to address fire-management. The Navy addresses erosion and targeted removal of invasive species, in general, through the INRMP, addresses training-related erosion through the Erosion Control Plan, and addresses further introduction of invasive species through implementation of the biosecurity plan (US Navy 2013a, entire; US Navy 2013b, entire; USFWS 2008, pp. 1–237; US Navy 2016, entire). In our most-recent review of the status of Acmispon dendroideus var. traskiae, we considered that land-use both in terms of ground-disturbance and wildfire remained threats to recovery (USFWS 2012, p. 29124). As noted above, most of the population of A. d. var. traskiae occurs outside designated training areas and hence, direct impacts to the population are minimal. No naturally occurring erosion nor that induced by human activities has been documented to threaten A. d. var. traskiae to date, and the one site in Wilson Cove where individuals were affected by human activity was successfully restored through outplanting. Through implementation of the island-wide conservation measures applied to project design, road- maintenance and repair, and the Erosion Control Plan to address training-related erosion, the threat of erosion is no longer considered significant across the range of A. d. var. traskiae. While the full impact of invasive species on Acmispon dendroideus var. traskiae is unknown, the effects are likely minimal or localized, given the expansion of A. d. var. traskiae on the island despite the presence of invasive species. Natural resources managers have been successful at decreasing the prevalence of particularly destructive nonnatives; while specific efforts may or may not directly impact A. d. var. traskiae, the significant efforts that the Navy makes to prevent the expansion of more invasive species (listed above) into habitat areas constitutes a significant benefit to the island ecosystem overall. Future impacts from fire remain uncertain. Fires are typically small, of low severity, and infrequent, and given they are most often ignited due to training, their typical locations are somewhat predictable. However, the fire that burned the eastern escarpment of SHOBA in 2017 was severe and individuals of Acmispon dendroideus var. traskiae were lost, though post fire monitoring showed many plants survived the fire and new recruits are also colonizing the area. Fire management includes the installation of fuel breaks and fire-suppression and is intended to reduce the spread of fire from designated training areas and to reduce ignition probability. The 2009 fire plan helps inform strategic decisions for training using live fire or incendiary devices, and the Navy reduces activities liable to cause fire during fire-season. Climate change may influence Acmispon dendroideus var. traskiae by affecting germination or persistence of adult plants if drought or increasing temperatures result in significant changes in vegetation communities on SCI. Warm temperatures appear to better- facilitate seed germination than do cold temperatures; however, reductions in annual precipitation may nullify any potential benefit of a warming climate, but these effects are speculative. Therefore, the magnitude of this rangewide threat and how it may affect this taxon is unknown at this time, but significant impacts to A. d. var. traskiae from climate change are unlikely to occur in the next 20 to 30 years.

SSA Report – San Clemente Island Lotus 52 March 2020

Potential changes to military training and testing on SCI are being considered, and changes that would affect Acmispon dendroideus var. traskiae are currently unknown. For instance, future training may include expansion of or additions to terrestrial designated training areas, introduction of new training methods, equipment, and activities that would affect fire frequency, fire severity, or soil erosion. While military training on SCI has been and will continue to be dynamic as it evolves to meet new requirements, changes are expected to be incremental, as they have been in the past. Such changes are subject to environmental review under applicable laws and regulations, including NEPA and ESA, and impacts to federally listed and sensitive species will be addressed. Therefore, we consider the main threats to Acmispon dendroideus var. traskiae to be 1) training impacts to individuals located within the Impact Areas, AVMAs or TARs, 2) impacts from training or erosion to individuals within 100 ft of a road or 50 ft of the AVMR (to account for troop movements as well as erosion potential), and 3) impacts from fire to individuals that lie within areas most likely to burn. No individuals currently exist within the Impact Areas, AVMAs, or within 50 ft of the AVMR. While other threats to A. d. var. traskiae exist to various degrees throughout its range (invasive grasses and other species, potential of trampling within the SWATs, IOA, or elsewhere, erosion events, etc.), we consider these threats to be minor enough that they will not have major impacts to the taxon. Looking at each of our threats by watershed (percent of locations or individuals near a road or in the TARs, and percent of the area that burned once or >1 time in the past 20 years), we found that 45 watersheds (78%), totaling 18,640 individuals (90%) had either zero or low total threats. Watersheds with low total threats were defined as those where potential impacts from roads, training areas, or fire frequency (assessed as fire acreage that burned in last 15 years) could threaten 50% or more of the number of locations, individuals, or for the case of fire, where fires in the last 20 years had burned less than 50% percent of the area within each watershed (Table 14 in Appendix A). We assume that if fires follow a similar fire pattern in the future, then watersheds that burned before have a higher risk of burning again. There are 31 occupied watersheds where no fires have burned in the past 20 years, and 14 more where fires in the last 20 years burned less than 50% of the watershed’s area (Table 14 in Appendix A). Therefore, we found that 50% of watersheds and 66% of individuals are located in areas where no quantifiable threats exist. Only 22% of watersheds and 10% of individuals are associated with a threat that could potentially adversely impact 50% or more of the locations, individuals, or area within the watershed (Table 9; Figure 12).

Table 9. Numbers and percentages of watersheds and individuals assessed to have varying levels of threats: none, low (threats that could potentially affect <50% of the locations, individuals, or area within the watershed), or medium (threats that could potentially affect ≥50% of the locations, individuals, or area within the watershed). Threats identified include locations or individuals near the AVMR or a road or in the TARs, and percent of the watershed area that burned once or >1 time in the past 20 years (1999– 2018). Watersheds Individuals Threats Watersheds Individuals (%) (%) none 29 13,620 50% 66% low 16 5,020 28% 24% moderate 13 2,103 22% 10%

SSA Report – San Clemente Island Lotus 53 March 2020

Figure 12. Representation of locations of watersheds where no threats exist to Acmispon dendroideus var. traskiae (ACDET), a low level of threats exist to the watershed (threats could potentially affect <50% of the locations, individuals, or area within the watershed), or a moderate level of threats exist to the watershed (threats could potentially affect ≥50% of the locations, individuals, or area within the watershed). Threats identified include locations or individuals within 100 ft of a road or the AVMR, in the TARs, and percent of the watershed area that burned once or >1 time in the past 20 years.

SSA Report – San Clemente Island Lotus 54 March 2020

SECTION 5 –CURRENT CONDITION

In this chapter, we describe the current condition of Acmispon dendroideus var. traskiae using our definition of the current distribution (Section 2.6). We describe the current condition using the 3Rs based on the taxon’s current distribution, population size, and trends.

5.1 Populations and Management Units Resiliency is typically measured at the population level. While we consider that Acmispon dendroideus var. traskiae represents a single population, as it is widespread on the island and there is no natural division in its range (USFWS 2013, p. 45436), for monitoring and tracking the population in the future, we noted that a delineation of the population into watershed units would be useful, and such a delineation could further help to quantify threats across the range. Watersheds have been suggested for use in delineation for monitoring purposes by the Navy (Vanderplank et al. 2019, pp. 6–7), as every point on the island can be easily assigned to a watershed and watershed boundaries on SCI are not expected to change significantly during the 20 to 30-year time frame of this analysis. We divided the taxon’s range into watershed units to assess resilience. These units are not meant to represent “populations” in a biological sense; rather, these units were designed to subdivide the taxon’s range in a way that facilitates assessing and reporting the variation in current and future resilience across the range. In this document, we assessed the taxon’s ability to withstand stochastic events in each watershed, and how these occupied watersheds contribute to the viability of the entire island population (the taxon). Note that this way of delineating management units within which to measure resiliency does not follow the methods used in the 2013 downlisting rule (USFWS 2013, p. 45407) and it is therefore not directly comparable. However, 58 watersheds that are represented here correspond to the 29 extant occurrences from the downlisting rule in Appendix C for cross-referencing purposes (Table 16).

5.2 Methods for Estimating Current Condition To assess the resiliency of Acmispon dendroideus var. traskiae, we assessed the overall condition of the population by evaluating occupancy, locations, and individuals within each watershed and the population trends which indicate the ability of A. d. var. traskiae to withstand and recover from stochastic events. Using our assumptions in our analysis of the current distribution of the taxon (Section 2.6), we defined the taxon’s current resiliency based on the locations at which the taxon has occurred within the past 15 years as this is considered a time-frame over which persistence, even absent verifying survey information, can likely be assumed. We based this assumption on the knowledge that no destructive impacts to occupied habitat have occurred across the majority of the range of this taxon. Resiliency was considered higher for plants within watersheds supporting a greater number of individuals over time; however, if all of the individuals within a watershed were in just one location, we assumed that they are less resilient than a watershed with the same number of individuals that are spread out across multiple locations, as plants will be more likely to persist through stochastic events if one localized event is unable to affect all the plants in the entire watershed. Since the majority of the survey data considered current was collected, there have been two fire seasons on SCI with fires that were ranked as moderate to high severity and therefore had the potential to kill adult plants. While some post-fire monitoring has shown that at least

SSA Report – San Clemente Island Lotus 55 March 2020

some Acmispon dendroideus var. traskiae individuals have persisted and others likely have become established in the burned areas since the fire, it is unknown what percent of the total plants were lost. While the actual percentage of existing individuals that survived is unknown, from the 2019 survey data, we used the 2019 survey data to estimate that at least 20% survived, while others may have been missed by surveyors. Therefore, we adjusted numbers of individuals that fall into these burn polygons downward to account for the unknown response to these severe fires that occurred after the count data were collected.

Within individual watersheds To evaluate the resiliency of Acmispon dendroideus var. traskiae within individual watersheds, we will assess the following:

Number of individuals: the minimum conservative estimate of the number of locations and current population within each watershed based on the methodology outlined in Section 2.6. Watersheds occupied at just one location were noted.

Severe fire: the number of locations and individuals potentially adversely affected by the 2012 and 2017 fires that were assessed as fire severity class 1 (completely burned) through severity class 3 (moderately burned). These fire seasons were the only ones where the fires were of this severity, which has the ability to kill some or all shrubs. Plant surveys in 2019 confirmed that Acmispon dendroideus var. traskiae still occur within these fire polygons; some existing plants were either unburned or are respouting, while others may have established themselves since the fire. To account for the unknown effects of this fire, we determined the number of individuals within each watershed’s population that fell into the burn polygons and therefore may have been affected. We then modeled the effect at the highest expected percent lost as a worst-case scenario and removed 80% of those individuals.

Island-wide To evaluate the resiliency of the overall island population, we will assess the following:

Trends (counts over time): the trajectory of the population based on the best scientific data through counts of locations and populations over time.

Number of occupied watersheds: the number of watersheds that are currently occupied.

Number of individuals: the minimum conservative estimate of the current population, using the methods described in Section 2.6. Due to the time that has elapsed since the extensive surveys, we are unable to confirm that all of these individuals and groups still persist; however, using knowledge of the plants’ biology, expert opinion, and personal observations, we have reason to believe that the majority of these still exist or have been replaced over time. While some of these locations may have disappeared, especially the ones with fewer individuals, it is equally likely that other locations have been established and individual counts have grown at many locations, as well.

SSA Report – San Clemente Island Lotus 56 March 2020

5.3 Current Condition Results Within individual watersheds Number of individuals: Eight of the 58 occupied watersheds (14%) have a local population of under 10 individuals. Twenty-five watersheds (43%) have fewer than 50 individuals. Just under half, 25 watersheds (43%), have 100 or more individuals, and 9 watersheds (14%) have over 500 individuals (Table 14 in Appendix A). However, 54 watersheds (93%) have Acmispon dendroideus var. traskiae present at fewer than 10 locations, and 17 (29%) have just one documented location within the watershed.

Severe fire: Nine of the 58 occupied watersheds had locations and individuals that may have been affected by moderately to high severity fires that burned during the 2012 and 2017 fire seasons. Fewer than 10% of the individuals were potentially affected in 3 watersheds, fewer than 50% of the individuals were potentially affected in 6 watersheds, over 90% of individuals were potentially affected in 3 watersheds, and all of the individuals in 1 watershed were potentially affected (Table 10). Removing 80% of these potentially affected individuals yielded new population estimates for these watersheds that were greatly reduced for many and extirpated the population in one watershed where only one individual previously occurred (Table 10).

Table 10. Occupied watersheds that may have lost Acmispon dendroideus var. traskiae during the 2012 and 2017 fire seasons where fires burned at a severity that can kill shrubs. Percentages are given for the numbers of individuals that could have been affected in each severity class, as well as the total percent of individuals that may have experienced negative effects of the fire, and the resulting adjusted estimate of the total individuals. Total Severity Total % Individuals in potentially Adjusted Watershed Watershed 1 2 3 affected Individuals WS_1116 176 18% 18% 150 WS_1119 152 33% 1% 34% 110

WS_1123 30 100% 100% 6 WS_1135 302 <1% <1% 301 WS_1137 331 91% 91% 91 WS_1139 87 5% 5% 84 WS_1158 1 100% 100% 0 WS_1159 467 43% 1% 1% 45% 297 WS_1160 417 <1% <1% 415

Due to the patchy nature of fires and because these methods do not account for new individuals that have established since the fire, populations in these watersheds may be higher.

Island wide Number of occupied watersheds:

SSA Report – San Clemente Island Lotus 57 March 2020

There are 216 watersheds delineated on SCI, and our current distribution of Acmispon dendroideus var. traskiae encompasses 58 of these. After accounting for the impacts of the 2017 fire, 57 watersheds remain occupied.

Number of individuals: Our assessment of the current distribution of Acmispon dendroideus var. traskiae retained 249 locations representing 21,251 individuals. The total population drops to 20,997 individuals after accounting for the effects of the 2017 fire. Retained numbers of locations and individuals for each year are presented in Table 11. Counting only the most recent data years, 2011 through 2014, a total of 175 locations and 13,293 individuals are still represented.

Table 11. Total locations and individuals considered current, broken down into survey points retained by year. Our methodology estimates approximately 21,251 individuals at 249 locations. Year Locations Individuals 2004 19 3,672 2005 5 202 2006 6 130 2007 4 34 2008 2 107 2009 6 35 2010 32 3,778 2011 113 10,980 2012 23 958 2013 27 923 2014 12 432

Trends (counts over time): As reported in Section 2.6, counts of individuals have increased over time with each survey effort, from approximately 1,340 at the time of listing to 11,938 in 2011–2012 (Table 2). We cannot be sure to what extent the growth is attributed to differences in survey effort or survey extent, but we are certain that the population has grown significantly since listing. Survey data indicates that both the locations and numbers of individuals have increased over time (Table 2, Table 3, Figure 3).

5.4 Current Population Resiliency Trends indicate that the population of Acmispon dendroideus var. traskiae on SCI has been increasing over time and withstanding current stochastic effects such as drought cycles, impacts from fire, erosion, or military training and land use, and the presence of nonnative invasive species. Therefore, to quantify resiliency, used the number of individuals within each watershed. We first assessed the resiliency of each watershed, and then scaled up to the entire population on the island. We binned our assessed resiliency scores by watershed based on number of individuals; these breakpoints are based loosely on expert opinion but are mainly for visualization and have no verifiable biological meaning. Resiliency scores are as follows: • Very high— populations with >500 individuals.

SSA Report – San Clemente Island Lotus 58 March 2020

• High— populations with 100-499 individuals. • Moderate— populations with 10-99 individuals. • Low— populations with <10 individuals.

However, for any watershed where all Acmispon dendroideus var. traskiae occurred in just one location, we lowered the resiliency score by one level. Having all the individuals in just one location means that a single localized impact could extirpate that watershed’s population, regardless of how many individuals are present. Resiliency scores for individual watersheds can be found in Table 14 in Appendix A. After adjusting for the effects of the 2017 fire, one watershed was considered no longer occupied. Of the 57 occupied watersheds, 9 (15.8%) are considered very highly resilient, 13 (22.8%) are considered highly resilient, 17 (29.8%) are considered moderately resilient, and 18 (31.6%) are considered to have low resiliency (Table 12). When we scale this up to the entire population, we find that 22 watersheds accounting for over 92% of the entire population and those watersheds are high to very highly resilient. Another 17 watersheds account for over 5% of the population that is moderately resilient. The majority of the watersheds (35 watersheds, 61%) contain fewer than 100 known individuals. Given the population trends and trajectory, the overall population has adequate resiliency to sustain itself in the presence of the current level of stochastic pressures and threats and the current level of conservation and management efforts (Figure 13).

Table 12. The number of watersheds that fall into each of our resiliency categories, the numbers of individuals the watersheds in each category accounts for, and the percent of the total island wide population represented. Percent Watersheds Individuals Total Very High 9 15,942 76.9% High 13 3,330 16.1% Moderate 17 1,078 5.2% Low 18 393 1.9%

SSA Report – San Clemente Island Lotus 59 March 2020

Figure 13. Current resiliency of Acmispon dendroideus var. traskiae (ACDET) (based on estimated number and distribution of individuals) by watershed.

5.6 Current Representation Genetic data and the distribution on SCI suggest that there is just one population of Acmispon dendroideus var. traskiae. Genetic work indicates that this population has moderate levels of genetic diversity, with variability among groups of plants. Therefore, even though there is just one representative unit, the taxon appears to have adequate genetic variation. It occupies habitat at varying elevations and slopes, both east-facing and west-facing, and throughout the island (Figure 4) and occupies different habitat types across the island, indicating that it has environmental plasticity and adaptability. While we do not know the historical distribution of this taxon, it is unlikely that it has lost much of its historical range or habitat diversity. Thus, we expect the species should be able to withstand potential short-term changes in environmental conditions or catastrophic pulse events, such as trampling, major erosion events, or severe fires.

SSA Report – San Clemente Island Lotus 60 March 2020

5.6 Current Redundancy There are 57 watersheds currently estimated to be occupied by Acmispon dendroideus var. traskiae, whereas only 4 were known to be occupied between 1980 and 1989 (Table 3). The number of individuals has grown substantially since listing, and the majority of the current individuals are located in areas that infrequently experience fire and have no major threats (Figure 10, Table 9). Given the population size and distribution, and considering the likely potential catastrophic events, we envision that only an unusually severe event could foreseeably threaten the taxon’s viability. A catastrophic fire, a major erosion event (such as caused by periods of heavy rainfall), or a severe drought are the most plausible potential impacts; an outbreak of an invasive, predatory, or pathogenic species is also possible but highly speculative. We expect that even a large, severe fire would be unlikely to affect a significant portion of the island; however, the population is especially dense along the eastern escarpment. A large, severe fire that burned a substantial portion of this or other high-density areas could have significant effects to the population size. While only an extreme, prolonged drought could wipe out the taxon entirely, the effects of multiple, severe drought years, coupled with other stressors, could have substantial impacts to species viability. A severe drought, for instance, has the ability to impact the vegetation island-wide, although drought impacts would have both an elevational gradient and an east/west difference due to prevailing wind direction, insolation and evapotranspiration, and the presence and persistence of fog. Given A. d. var. traskiae’s wide distribution, we’d expect at least some individuals would be able withstand the drought, by collecting adequate fog moisture or tapping water reserves in the soil. However, depending on the length and severity of drought, impacts to the taxon could be substantial. Over 60% of the watersheds have fewer than 100 individuals, putting them at higher risk of stochastic threats; the loss of some of these watersheds could decrease the taxon’s range. Like all endemics, A. d. var. traskiae has a small range and is confined to SCI and would be unable to disperse elsewhere. While the species is numerous, impacts to areas where the species is particularly numerous (such as the eastern escarpment) could decrease the population size substantially.

SECTION 6 – FUTURE CONDITIONS AND VIABILITY

We have considered what Acmispon dendroideus var. traskiae needs for viability and the current condition of those needs (Sections 3 and 5), and we reviewed the factors that are driving the current and future conditions of the taxon (Section 4). We now consider what the taxon’s future condition is likely to be. We apply our future forecasts to the concepts of resiliency, representation, and redundancy to describe the future viability of A. d. var. traskiae. Using our analysis of the factors influencing viability in Section 4, we reevaluate the current threats to Acmispon dendroideus var. traskiae. In 2008, SCI completed environmental analyses (2008 EIS/BO) to support new and increased training activities on the island, the effects of which were not well known. Since the 2013 rule, many of the potential effects of that training expansion have not been realized, and through an Erosion Control Plan (US Navy 2013b, entire), fire management plan (US Navy 2009, entire), and the INRMP (US Navy 2013a, entire), the potential for those effects are reduced.

SSA Report – San Clemente Island Lotus 61 March 2020

6.1 Introduction Since the removal of feral browsers and grazers from SCI by 1992, few threats exist to the viability of Acmispon dendroideus var. traskiae on the island, and there are many ongoing management efforts designed to minimize these threats. The taxon currently occupies a broad distribution on the island and exists mostly in locations that occur outside the training areas, are inaccessible to vehicular and foot traffic, have burned infrequently, and have not experienced erosion events. Throughout much of its distribution, the taxon exists in areas where there are no threats to its habitat. It is unknown whether A. d. var. traskiae generally occurs infrequently within the training areas and other highly used or fire-prone areas because the plants are less able to become established there, or whether it is because they have just not yet dispersed into those areas. The factors that appear to have the most potential to impact taxon viability in the future are land use, fire, and climate change, including potential compounded effects. However, we are unable to address the full impacts of climate change because how climate change will affect SCI in the long term remains unclear, and most importantly, the persistence and timing of the fog layer, which provides moisture and a refuge from the full impacts of warming, is unknown. However, based on our review of the literature and models on climate change, we assume that climate change will not have major effects on Acmispon dendroideus var. traskiae in the next 20- 30 years. We do account for the possibility of more frequent or severe fires, which could be a short-term impact of climate change, a product of increased military training, or both. Therefore, we consider the future of A. d. var. traskiae in terms of its threats and conservation efforts over the next 20 to 30 years. Similarly, we cannot predict the future of fire on the island, but we do know what the fire pattern has been in the past. Fire seasons generally have consisted of low-severity fires; however, in 2017, one fire was of an unprecedented severity for its size and occurred over a large area in which fire had not been documented previously. Whether this fire was an outlier or whether severe fires will become more common in the future is unknown, though fire management was modified following the fire to minimize the likelihood of such a burn in the future. If training increases, or if fires become more frequent for other reasons (climate change and weather/precipitation patterns, invasive grasses, etc.), we would not expect the current fire pattern to change, as we would not expect the locations of live fire use to change, but increased use of these areas might increase the frequency of fires. While it is possible that future changes or impacts might alter the footprint of where fires generally burn, this type of threat is not expected and thus is not easily predictable; thus, the sort of future threats that could alter the fire footprint is addressed in Sections 6.5 and 6.6, under Representation and Redundancy. Land use could have potential impacts to Acmispon dendroideus var. traskiae due to military training and other uses, specifically where individuals exist within boundaries of a TAR or near the AVMR or a road (where individuals could experience other road impacts such as erosion). While the plants known to exist in these areas and near roads have persisted currently, future increases in the amount or location of training (including increased foot traffic and maneuvering along roads) could affect the individuals within these locations. Threats associated with these areas, while more likely within the training areas than near roads, were accounted for. While impacts from fire, training, and roads are somewhat likely to occur and the location of these impacts can be predicted, severe erosion events are less likely and it is harder to predict their location. Therefore, we considered severe erosion events as plausible catastrophic impacts and will also account for them in our assessment of Representation and Redundancy.

SSA Report – San Clemente Island Lotus 62 March 2020

Therefore, to assess the future viability of Acmispon dendroideus var. traskiae, we considered several future scenarios that encompass the uncertainty associated with fire and military training, as well as uncertainty in the levels of recruitment over time.

6.2 Methods To assess future resiliency of each watershed, we will address the following metrics:

Growth and recruitment: Abundance in existing watersheds: Based on survey data and trends, the population of Acmispon dendroideus var. traskiae appears to be increasing and expanding. However, the rate and patterns of recruitment are unknown. Existing data do not indicate at what rates groups of plants expand over time, where expansion is occurring, or areas that may be decreasing. While the population has grown in the past several decades, it is unknown whether this level of growth could be maintained into the future due to unknown habitat limitations, threats, or other factors. Therefore, we conservatively modeled recruitment at two levels: • Moderate recruitment: additional 25% individuals added to the watershed. • Low recruitment: additional 5% individuals added to the watershed

These estimates are based solely on expert opinion of a conservative estimate of expected future growth. The overall known population has increased each decade since listing, and comparatively, these population growth estimates should be biased low. Further, in our future scenarios, we modeled recruitment before we model threats; therefore, all new recruits were subject to the threat analysis. In reality, some new recruitment would occur after some impact occurred. Therefore, this further kept our estimates biased low. Some watersheds may recruit at closer to the higher level and some may recruit at closer to the lower level.

New watersheds: Over time, Acmispon dendroideus var. traskiae has both increased its numbers within watersheds and has become established or sprouted from the seed bank within new watersheds. After the removal of feral browsers and grazers when the island began to recover, A. d. var. traskiae was able to become established in watersheds where occupation had not been documented previously. Between 1999 and 2009, 26 new watersheds were observed to be occupied by A. d. var. traskiae. From 2009 to 2014, 18 new watersheds were discovered to be occupied. Thus, 44 new watersheds were discovered to be occupied in less than two decades. While we expect that new watersheds will become occupied (or recently extirpated ones will become repopulated), we also expect that this growth will slow down, as nearby unoccupied watersheds become occupied. Currently there are 77 unoccupied watersheds that border an occupied one; while the habitat types and soil types in these watersheds do not appear to limit recruitment, we cannot predict how likely they are to become occupied. We do not know what the historical distribution of A. d. var. traskiae was, or how much recruitment is from the seed bank. While not impossible, we cannot expect the taxon to continue expanding into new watersheds at the current rate, but it is also unlikely that expansion will cease altogether. Therefore, for the next 20 to 30 years, we model recruitment into new watersheds at two levels: • Moderate recruitment: 10 new watersheds at the end of 20 to 30 years. We will assume these will recruit <10 individuals in 20-30 years (low resiliency).

SSA Report – San Clemente Island Lotus 63 March 2020

• Low recruitment: 5 new watersheds at the end of 20-30 years. We will assume all of these will recruit <10 individuals in 20-30 years (low resiliency).

Fire frequency and severity: Frequent, recurrent fire represents a potential threat to Acmispon dendroideus var. traskiae. While we cannot predict the future fire patterns, we can use past fire data to project potential impacts from fire in the future. Because fires tend to occur in the same areas, we will use the fire footprints from the past 15 years to model where future fires will burn. We will model the impacts of fire both 1) as if the 2017 fire season was an anomaly, and 2) as if the 2017 fire season is indicative of a change in the fire severity pattern and fires will increase in frequency or severity in the future due to a training increase, climate impacts, or both. Using the percentage of each occupied watershed that has burned once and more than once within the last 20 years, we will calculate the total number of individuals that could be affected by fire in that watershed. For instance, if 50% of the watershed has burned in the last 20 years, we assume that 50% of the individuals of A. d. var. traskiae could be affected by fire in the future. While we know that A. d. var. traskiae can withstand fires and can colonize areas that have burned in the past, evidenced by their presence within the areas that have burned in the last 20 years, the actual fire tolerance of A. d. var. traskiae is unknown. Severe fire has potentially reduced the number of individuals in an area by 75%. Therefore, we modeled a range of fire effects assuming that fires will neither spare 100% nor eradicate 100% of the individuals in any burn polygon. To capture this uncertainty, we modeled future fire impacts as:

• Status quo fire severity: We assume that 25% of individuals that could be affected by more than 1 fire (area that burned >1 time) perish. We assume plants persist in areas that burned just once. • Increased fire severity: Assume that 50% of individuals that could be affected by more than 1 fire (area that burned >1 time) perish. Further, we assume that 25% of individuals that could be affected by 1 fire (area that burned 1 time) perish. • Extreme fire severity: Assume that 75% of individuals that could be affected by more than 1 fire (area that burned >1 time) perish. Further, we assume that 50% of individuals that could be affected by 1 fire (area that burned 1 time) perish.

The status quo scenario attempts to model the current rarity of high-severity fires, yet still bias the effect high. These models assume that for some percent area of each watershed that experienced a fire between 1999 and 2018, those watersheds would lose 25–50% of a corresponding percent of their individuals under the increased and extreme fire severity scenarios. Likewise, for some percent area of each watershed that experienced more than one fire between 1999 and 2018, those watersheds would lose 50–75% of a corresponding percent of their individuals under the increased and extreme fire severity scenarios, and 25% of a corresponding percent of their individuals under the status quo scenario. While this does not explicitly model fires that could potentially break out in new areas and burn previously unburned watersheds, it does attempt to account for them by assuming some mortality by all fires. Severe fires in previously unburned areas are accounted for in our discussion of population redundancy and representation. Further, these scenarios do not specifically account for new recruitment from the seed bank after the fire, since we modeled recruitment before applying the threats, and therefore, in these scenarios, individuals killed by fire are not replaced in the population. Thus,

SSA Report – San Clemente Island Lotus 64 March 2020

even if severe fires always remove 80–95% of the individuals on the landscape, we assume that the fires will be patchy, severity will differ from area to area, and at least some of the individuals will be replaced over time. By using 75% as our maximum removed, our estimates of persistence given future fires should not be biased high.

Land Use/Training Training and land use in the TARs and along the roads represent a potential threat to Acmispon dendroideus var. traskiae. While we cannot predict exactly how land use will change over time, changes in training and land use are common. Using the percent of individuals that occur either within a training area or near a road, we calculated the total number of individuals that could be affected by increased training in that watershed. While we know that Acmispon dendroideus var. traskiae can withstand some training impacts, evidenced by their presence within TARs, the actual realized impacts of increased training on Acmispon dendroideus var. traskiae is unknown. Therefore, to capture this uncertainty, we modeled future training impacts as: • Training stays the same: no impact • Training increase: Assume that 50% of all locations and individuals within the TARs or within 100 ft of roads perish. • Extreme training increase: Assume that all locations and individuals within the TARs or within 100 ft of roads or the AVMR perish.

By modeling these training increases, we attempted to project plausible although unlikely scenarios in which 50% to 100% of all individuals in the TARs and those near roads perish over time. Any changes to training would have to be substantial to affect the taxon in this way; this assumes that all areas within each training area would be utilized and severely degraded. Changes in training on the island may result someday in new training footprints, but the area within them utilized and degraded is unlikely to be as large as the entirety of the current footprints; therefore, we have tried to bias these impacts high. The possibility of training footprints changing dramatically will be addressed in Section 6.6, future redundancy.

6.3 Models and Scenarios We will model future effects on resiliency over 20 to 30 years, the maximum projection we feel we can project out to before the effects of climate change cause too much uncertainty.

• Scenario 1: Status quo. This assumes same fire severity, and current training impacts (current plants persist under current training). • Scenario 2: Moderate threat increase. This scenario assumes increased fire severity, and increased training impacts. • Scenario 3: High threat increase. This scenario assumes extreme fire severity, and extreme training impacts.

We applied each of these scenarios to the number of individuals considered current in each watershed, but using the adjusted numbers for the individuals that are located inside the severe 2017 fire polygons. We first adjusted the numbers to reflect population growth and recruitment; we then applied the threats and the associated declines in numbers to the total new number of individuals.

SSA Report – San Clemente Island Lotus 65 March 2020

We expect that in all of these scenarios, resulting population sizes are biased low. By applying the threats second, we ensure that all new individuals are subject to the modeled threats. In reality, over a 20- to 30-year period, fires or training increases may eliminate individuals but it is likely that individuals would return and become reestablished in future years. Further, increased training to the point that it impacts 50–100% of the individuals in the training areas and fires that impact such high percentages of individuals in all of the past fire polygons are unlikely. And finally, the current population size on SCI is likely already underestimated. Surveys have not been comprehensive across all potential habitat, as an island-wide systematic survey would be both unfeasible due to terrain, time, and effort required, and plants are further difficult to locate among the grasses and other plants. That plants often go undetected is evidenced by the 2019 surveys that located 261 previously unknown individuals on the eastern escarpment in areas that have not burned (and therefore are probably not all newly established).

6.4 Future Resiliency We again binned our assessed resiliency scores by watershed based on number of individuals, where resiliency is as follows: • Very high— watersheds with >500 individuals. • High— watersheds with 100-500 individuals. • Moderate— watersheds with 10-99 individuals. • Low— watersheds with <10 individuals.

Again, for any watershed where all Acmispon dendroideus var. traskiae occurred in just one location, we lowered the resiliency score by one level. We did not assume growth affected the number of locations. Population changes within individual watersheds under each scenario can be found in Table 14 in Appendix A. Our methods predict that, in the next 20 to 30 years, the island wide population is likely to increase (Table 13). The number of occupied watersheds increase by 5 to 10 in all of our scenarios, as no additional watersheds are extirpated even with extreme training and fire impacts. The numbers of watersheds considered highly or very highly resilient remain similar in all of our scenarios. Likewise, the population grows in Scenario 1, and the current population estimate falls into the range for the population size in Scenarios 2 and 3 (Table 13; Figure 14). Thus, we expect the island population will maintain its current levels of resiliency and is therefore likely to continue to withstand normal stochastic impacts under all future scenarios.

Table 13. The number of watersheds considered of very high, high, moderate and low resiliency and the total estimated population as considered current and in each of our three future scenarios. Watershed numbers in parentheses represent the total watersheds assuming recruitment into new watersheds, with a range from low (5 new watersheds) to high (10 new watersheds) recruitment. Watershed Resiliency Very High High Moderate Low Total Individuals Current 9 13 17 18 57 20,743 Scenario 1 10 13 16 18 (23-28) 57 (62–67) 21,595–25,708 Scenario 2 10 11 18 18 (23-28) 57 (62–67) 20,627–24,556 Scenario 3 10 9 18 20 (25-30) 57 (62–67) 19,706–23,460

SSA Report – San Clemente Island Lotus 66 March 2020

Figure 14. Resiliency estimates by watershed (based on number of individuals) currently as well as under each of our three scenarios. Extant watershed counts do not account for recruitment into new watersheds.

6.5 Future Representation In all scenarios, the number of watersheds occupied by Acmispon dendroideus var. traskiae and the number of individuals is likely to remain the same or increase over the next 20 to 30 years. Even ignoring new watersheds, no additional watersheds are expected to become extirpated even Scenario 3, and the existing watersheds cover a broad distribution on SCI.

SSA Report – San Clemente Island Lotus 67 March 2020

Currently, this taxon appears to have adequate genetic diversity and it inhabits diverse areas and habitats on the island. Under all of the future scenarios, we expect this taxon would retain current levels of representation.

6.6 Future Redundancy If Acmispon dendroideus var. traskiae is able to recruit into new watersheds as we expect, then the number of watersheds occupied by A. d. var. traskiae is likely to increase over time in all of our scenarios. The number of individuals is also expected to increase under Scenario 1 (no additional impacts) and may either increase or decrease by at most ~1,000 individuals over the next 20 to 30 years in the other two scenarios (where new impacts occur), indicating redundancy is likely to increase, also. Thus, we do not expect redundancy to change much from current, and we expect that the taxon will retain its redundancy and be able to sustain most major catastrophic events, such as unprecedented fires, major erosion events (such as caused by periods of heavy rainfall), or an outbreak of an invasive, predatory, or pathogenic species. Even the most severe of these events would be unlikely to affect the entire island-wide population. Even the largest fires that have ever burned are small in comparison to the range of A. d. var. traskiae. Severe erosion events resulting from heavy rainfall, a potential effect of climate change, could remove individuals, but this would occur within localized areas. Continued management efforts on the island would make the possibility of an extreme fire, predatory or pathogenic invasion, or major erosion event unlikely. Only an unprecedented, unusually severe or catastrophic could threaten the viability of the taxon. For instance, the effects of multiple, severe drought years, coupled with other stressors, could have substantial impacts to taxon viability. A severe drought has the ability to impact the vegetation island-wide; still, we’d expect at least some individuals would be able withstand even severe drought, perhaps by collecting adequate fog moisture or tapping water reserves in the soil. However, depending on the length and severity of drought, impacts to the taxon could be substantial. A change in the fire pattern could put more individuals at risk of fire, especially in the unlikely event that fires become frequent on the eastern escarpment. Like all endemics, A. d. var. traskiae has a small range and is confined to SCI and would be unable to disperse elsewhere. While the species is numerous, impacts to areas where the species is particularly numerous (such as the eastern escarpment) could decrease the population size, and thus redundancy, substantially.

6.7 Limitations and Uncertainties In any species status assessment, the process of projecting a population into the future requires making strategic simplifications of reality, accounting for multiple uncertainties, and making informed assumptions when necessary. Our assessment addressed some of the key uncertainties and yielded useful predictions for characterizing the future status of Acmispon dendroideus var. traskiae, and through the use of predictive constructs and multiple scenarios, we captured a range of plausible conditions in the future. However, there are still limitations to these predictions; we outline these uncertainties and assumptions of the analyses below. Our assessment of the current population used data collected between 2004 and 2014 (Table 11), following a rule set designed to prevent overcounting. However, the rule set assumes that numbers of individuals have not changed since they were counted, and that any declines in count numbers are likely to have been replaced elsewhere in the distribution. The reality of the size of the population and the number of currently occupied watersheds is unknown, although

SSA Report – San Clemente Island Lotus 68 March 2020

there is evidence that the population may in fact be much larger than our estimate (previously unknown individuals have been located since the data were collected). Our assessment of the impacts of the 2017 fire that burned a portion of the eastern escarpment also relied on assumptions. Absent systematic counts from both before and after the fire, following the same methods and search extent, we were unable to ascertain the full impacts to the individuals in those areas from fire. Survey data from 2019 showed that individuals did still persist in those areas, but the population numbers appeared reduced. We were unable to quantify the amount by which they were reduced; thus, we used the available data to make a conservative estimate and not overestimate the number of surviving individuals. However, again, the true impact of this fire is unknown. Our future scenarios also made several assumptions; first, we assumed that individuals within each watershed will be able to successfully recruit more individuals into those watersheds. Anecdotal evidence exists that recruitment in some groups of plants may be higher than in others; however, in absence of quantifiable data, we assumed recruitment and persistence would be consistent across the island, with number of new recruits calculated as a percentage of the existing number of individuals in each watershed; however, we attempted to use a conservative estimate of this level of recruitment so as to not overestimate growth over the projected timeframe. We also made assumptions regarding how training, fire, and impacts from proximity to roads would affect the numbers of individuals in each watershed where these threats may occur; again, no quantifiable data exists to accurately predict these impacts. Our estimates of these impacts were made in an attempt to not underestimate these threats but instead create worst-case scenarios. Further, we assumed that training, fire, and impacts from proximity to roads would impact the same areas as they have historically; our models do not account for a change to training area footprints or changes to where ignition sources are located or where fires are likely to burn. While we do not anticipate these sorts of changes, they cannot be ruled out. We also assumed that the Navy will continue to manage habitats on the island into the future, continuing their efforts to manage fire, invasive species, and erosion. If the Navy were to cease being good land stewards, our conclusions would likely be invalidated. The final major uncertainty regarding the future of SCI is the impacts of climate change in the long term and fire and drought impacts in the shorter term. While we do not expect climate change to have major impacts to the vegetation on SCI in the next 20 to 30 years, data may change as climate science evolves and new climate models come out. In the short term, drought cycles, which have been a part of the historical climate on SCI, may intensify. The full impacts of rainfall patterns and the future of the fog layer are unknown.

6.8 Conclusions Despite historical and current land uses, historical drought, historical and current fire patterns, and other existing threats, Acmispon dendroideus var. traskiae has substantially increased both its distribution and population numbers on SCI since listing. Currently, we expect that A. d. var. traskiae has adequate resiliency, redundancy, and representation to withstand most likely stochastic impacts, environmental changes, and the reasonably plausible, potential catastrophic events on SCI. Projecting the population into the future, given even impacts from increased fire frequency, fire severity, and training, we find that a substantial proportion of the population occurs outside areas where these threats are projected to occur. Thus, outside of a

SSA Report – San Clemente Island Lotus 69 March 2020

catastrophic or unprecedented impact, we expect the population will retain somewhat similar levels of resiliency, representation, and redundancy as current.

SSA Report – San Clemente Island Lotus 70 March 2020

REFERENCES CITED

Allan, G. J. 1999. Molecular systematic and biogeographic studies of the temperate herbaceous papilionoid tribes Loteae and Coronilleae (Fabaceae: Faboideae). PhD Dissertation, Claremont Graduate University, California. Arroyo, M.T. 1981. Breeding systems and pollination biology in Leguminosae. Pp. 723–769 In: R.M. Polhill and P.H. Raven [eds.]. Advances in legume systematics. Royal Botanic Garden, Kew, United Kingdom. Brouillet, L. 2008. The taxonomy of North American loti (Fabaceae: Loteae): new names in Acmispon and Hosackia. Journal of the Botanical Research Institute of Texas 2: 387– 394. Luc Brouillet 2012, Acmispon dendroideus var. traskiae, in Jepson Flora Project (eds.) Jepson eFlora, http://ucjeps.berkeley.edu/eflora/eflora_display.php?tid=91804, accessed on July 02, 2019. Bushakra, J. M., S. A. Hodges, J. B. Cooper and D. D. Kaska. 1999. The extent of clonality and genetic diversity in the Santa Cruz Island ironwood, Lyonothamnus floribundus. Molec. Ecol. 8: 471–475. Cayan, D., M. Tyree, D. Pierce and T. Das. 2012. Climate change and sea level rise scenarios for California vulnerability and adaptation assessment. California Energy Commission Publication CEC-500-2012-008. California Department of Fish and Game Natural Diversity Database [CNDDB]. 2011. Element Occurrence Reports for Acmispon dendroideus var. traskiae. Unpublished cumulative data current to January 30, 2011. Available on the internet at: http://www.dfg.ca.gov/biogeodata/cnddb/rarefind.asp. Accessed February, 2011. 14 pp. California Native Plant Society [CNPS]. 2011. Inventory of Rare and Endangered Plants (online edition). California Native Plant Society. Sacramento, California. Available on the internet at: http://www.rareplants.cnps.org/detail/1061.html. Accessed May 2011. 3 pp. Clemesha, R.S. 2020. Progress Report on Cooperative Agreement W9126G-17-2-0029 for period ending December 31, 2019. Prepared for US Army Corps of Engineers. 7 pp Corbin, J. D., M. A. Thomsen, T. E. Dawson, and C. M. D’Antonio. 2005. Summer water use by California coastal prairie grasses: Fog, drought, and community composition. Oecologia 145:511–521. Ellstrand, N. C. and D. R. Elam. 1993. Population genetic consequences of small population size: Implications for plant conservation. Annual Review of Ecology and Systematics 24:217-242. Field, C.B., G.C. Daily, F.W. Davis, S. Gaines, P.A. Matson, J. Melack, and N.L. Miller. 1999. Confronting climate change in California. Ecological impacts on the Golden State. A report of the Union of Concerned Scientists, Cambridge, Massachusetts, and the Ecological Society of America, Washington, DC. Fischer, D. T., C. J. Still and A. P. Williams. 2009. Significance of summer fog and overcast for drought stress and ecological functioning of coastal California endemic plant species. J. Biogeogr. 36: 783–799. Halvorson, W.L., D. B. Fenn, and W. R. Allardice. 1988. Soils and vegetation of Santa Barbara Island, Channel Islands National Park, California, USA. Environmental Management 12: 109-118.

SSA Report – San Clemente Island Lotus 71 March 2020

Iacobellis, S., D. Cayan, J. Norris, and M. Kanamitsu. 2010. Impact of Climate Change on the Frequency and Intensity of Low-Level Temperature Inversions in California. Scripps Institution of Oceanography, University of California San Diego. Isely, D. 1978. New varieties and combinations in Lotus, Baptisia, Thermopsis and Sophora (Leguminosae). Brittonia 30: 466–472. ______. 1993. Lotus. In: J.C. Hickman [ed.]. The Jepson manual. University of California Press. Berkeley, California. 616 pp. ______. 1998. Native and naturalized Leguminosae (Fabaceae) of the United States (exclusive of Alaska and Hawaii). ML Bean Science Museum Press, Brigham Young University, Provo, Utah. Jennings, Megan K., Dan Cayan, Julie Kalansky, Amber D. Pairis, Dawn M. Lawson, Alexandra D. Syphard, Udara Abeysekera, Rachel E.S. Clemesha, Alexander Gershunov, Kristen Guirguis, John M. Randall, Eric D. Stein, and Sula Vanderplank. (San Diego State University). 2018. San Diego County ecosystems: ecological impacts of climate change on a biodiversity hotspot. California’s Fourth Climate Change Assessment, California Energy Commission. Publication number: CCCA4-EXT-2018-010 Johnstone, J. A. and T. E. Dawson. 2010. Climatic context and ecological implications of summer fog decline in the coast redwood region. Proc. Natl. Acad. Sci. U.S.A. 107: 4533–4538. Jones, C. E. and M. B. Cruzan. 1999. Floral morphological changes and reproductive success in deer weed (Lotus scoparius, Fabaceae). Amer. J. Bot. 86: 273–277. Junak, S. 2006 (updated 2010). Sensitive plant status survey Naval Auxiliary Landing Field, San Clemente Island, California. Draft Final Report. Prepared for Department of the Navy Southwest Division, Naval Facilities Engineering Command, and Natural Resources Office, Commander, Navy Region Southwest San Diego, California. Letter of Agreement No. N68711-02-LT-00036. Santa Barbara Botanic Garden Technical Report. December. Junak, S. A. and D. H. Wilken. 1998. Sensitive Plant Status Survey, Naval Auxiliary Landing Field, San Clemente Island, California, Final Report. Santa Barbara Botanic Garden Technical Report No. 1 prepared for Department of the Navy, Southwest Division, San Diego, California. Kalansky, Julie, Dan Cayan, Kate Barba, Laura Walsh, Kimberly Brouwer, Dani Boudreau. (University of California, San Diego). 2018. San Diego Summary Report. California’s Fourth Climate Change Assessment. Publication number: SUM-CCCA4-2018-009. Keegan, D.R., B.E. Coblentz, and C.S. Winchell. 1994. Feral goat eradication on San Clemente Island, California. Wildlife Society Bulletin 22: 56-61. LaDochy, S. and M. Witiw. 2012. The continued reduction in dense fog in the southern California region: possible causes. Pure and Applied Geophysics 169: 1157–1163. Lebassi-Habtezion, B., J. González and R. Bornstein. 2011. Modeled large‐scale warming impacts on summer California coastal‐cooling trends. Journal of Geophysical Research: Atmospheres, 116 (D20). https://doi.org/10.1029/2011JD015759. Liston, A., L. H. Rieseberg and O. Mistretta. 1990. Ribosomal DNA evidence for hybridization between island endemic species of Lotus. Biochem. Syst. & Ecol. 18: 239–244. McGlaughlin, M.E., R. Lynn, K. Helenurm, and L.E. Wallace. 2018. Does Channel Island Acmispon (Fabaceae) form cohesive evolutionary groups? Western North American Naturalist: Vol. 78: 739-757.

SSA Report – San Clemente Island Lotus 72 March 2020

Munz, P.A. 1974. A Flora of Southern California. University of California Press, Berkeley and Los Angeles, California. Oladi, R., M. Emaminasab, and D. Eckstein. 2017. The dendroecological potential of shrubs in north Iranian semi-deserts. Dendrochronologia 44:94–102. Qu, X., A. Hall, S. A. Klein and P. M. Caldwell. 2014. On the spread of changes in marine low cloud cover in climate model simulations of the 21st century. Climate Dynamics 42: 2603–2626. Raven, P. H. and D. I. Axelrod. 1978. Origin and relationships of the California flora. Univ. Calif. Publ. Bot., vol 72. 134 p. Soil Ecology and Restoration Group (SERG). 2006. Soil erosion control and habitat restoration plan on NALF San Clemente Island. Report N68711-04-LT-A0061, 12 June 2006. 46 pp. ______. 2012. San Clemente Island Native Habitat Restoration Program, 2012 Annual Report. Seed Collection, Propagation, and Outplanting in Support of San Clemente Island Endangered Species Programs. San Diego State University. 31 pp. ______. 2015. Erosion control re-survey on NALF San Clemente Island. Final Report. 65 pp. ______. 2016. San Clemente Island Native Habitat Restoration Program, 2016 Annual Report. Seed Collection and Propagation; Site Selection, Out-planting and Maintenance; and Erosion Control at NALF San Clemente Island, Naval Base Coronado, California. 72 pp. Sokoloff, D. 2000. New combinations in Acmispon (Leguminosae, Loteae). Ann. Bot. Fennici 37: 125–131. Sørensen, J. and A. Sessitsch. 2007. Plant-associated bacteria—lifestyle and molecular interactions, pp. 211–236. In J. D. van Elsas, J. K. Jansson and J. T. Trevors [eds.], Modern soil microbiology, 2nd ed. CRC Press, Taylor and Francis Group, Boca Raton, FL. Taylor, P.D. 1993. Connectivity is a vital element of landscape structure. OIKOS 68: 571-573. Tierra Data Inc. 2005. San Clemente Island vegetation condition and trend analysis 1992–2003. Report prepared for the Southwest Division, U.S. Naval Facilities Engineering Command, San Diego, California. Pp. 1 to 96 and Appendices. ______. 2007. Project increase in sheet and rill erosion due to military operations proposed on San Clemente Island. Report prepared for the Southwest Division, U.S. Naval Facilities Engineering Command, San Diego, California. Pp. 1 to 36 and Appendices. U. S. Department of the Navy [US Navy]. 2002. Integrated Natural Resources Management Plan (INRMP) Naval Auxiliary Landing Field San Clemente Island, California. Prepared by Tierra Data Systems. 784 pp. ______. 2008a. Southern California Range Complex Environmental Impact Statement, Overseas Environmental Impact Statement: Volumes 1 and 2, Final, December 2008. Naval Facilities Engineering Command Southwest, San Diego, California. Pp. 1 to 10-201 and Appendices. ______. 2008b. Programmatic Terrestrial Biological Assessment, San Clemente Island, Final May 2008. Prepared by Commander, U.S. Pacific Fleet, San Diego, California. Pp. 1 to 8-12 and Appendices. ______. 2009. Wildland Fire Management Plan, San Clemente Island, Final June 2009. Prepared by Tierra Data Inc., Escondido, California. Pp. 1-1 to 6-6 and Appendices. ______. 2013a. Integrated Natural Resources Management Plan (INRMP) Naval Auxiliary Landing Field San Clemente Island, California. Prepared by Tierra Data Inc. 784 pp.

SSA Report – San Clemente Island Lotus 73 March 2020

______. 2013b. Erosion Control Plan for San Clemente Island. Prepared by Science Applications International Corporation, Carpinteria, CA. 132 pp. ______. 2016. Naval Auxiliary Landing Field San Clemente Island Biosecurity Plan FINAL. November 2016. Prepared by ManTech SRS Technologies, Inc. U.S. Fish and Wildlife Service [USFWS]. 1977. Determination that seven California Channel Island animals and plants are either endangered species or threatened species [endangered: San Clemente loggerhead shrike, Lanius ludovicianus mearnsi; San Clemente broom, Lotus scoparius ssp. traskiae; San Clemente bushmallow, Malacothamnus clementinus; San Clemente Island larkspur, Delphinium kinkiense; San Clemente Island Indian paintbrush, Castilleja grisea. Threatened: island night lizard, Klauberina riversiana; San Clemente sage sparrow, Amphispiza belli clementae]. Fed. Reg. 42: 40682–40685. ______. 1984. Recovery Plan for the Endangered and Threatened Species of the California Channel Islands. U.S. Fish and Wildlife Service, Portland, Oregon. 165 pp. ______. 2007. Acmispon dendroideus var. traskiae (San Clemente Island lotus) 5-year review: Summary and evaluation. U.S. Fish and Wildlife Service, Carlsbad, California. 22 pp. ______. 2008. Biological Opinion for San Clemente Island Military Operations and Fire Management Plan, Los Angeles County, California (Service File FWS–LA–09B0027– 09F0040). [November, 2008]. ______. 2012. 12-month finding on a petition to downlist three San Clemente Island plant species; proposed rule to reclassify two San Clemente Island plant species; taxonomic correction; proposed rule. Fed. Reg. 77: 29077–29128. ______. 2013. Reclassification of Acmispon dendroideus var. traskiae (=Lotus d. subsp. traskiae) and Castilleja grisea as threatened throughout their ranges. Final Rule. Fed. Reg. 78: 45405–45439. ______. 2016. USFWS species status assessment framework: an integrated analytical framework for conservation. Version 3.4.8, August 2016. Uyeda, K., K. Warkentin, D.A. Stow, J.F. O’Leary, T. Zink, J. Lambert, L. Coulter, R. Snavely, A. Loerch, A. Schurlock, and G. Schag. 2019. Vegetation mapping at NALF San Clemente Island, Naval Base Coronado, California. Final Report. Agreement No.: W9126G-15-2-0034. 38 pages. Vanderplank, S.E. and E. Ezcurra. 2015. How Marine Influence Controls Plant Phenological Dynamics in Mediterranean Mexico. Journal of Plant Ecology. 9(4), pp.410-420. Rtv066. Vanderplank, S., K. O’Connor, B. Munson, and D. Lawson. 2019. A Conservation Assessment for Acmispon dendroideus var. traskiae (San Clemente Island Lotus, Fabaceae). Rancho Santa Ana Botanic Garden Occasional Publications, Number 16, vi + 35 pages. Wallace, L. E., G. L. Wheeler, M. E. McGlaughlin, G. Bresowar AND K. Helenurm. 2017. Phylogeography and genetic structure of endemic Acmispon argophyllus and Acmispon dendroideus (Fabaceae) across the California Channel Islands. Amer. J. Bot. 104: 743– 756. Williams, A. P., R. E. Schwartz, S. Iacobellis, R. Seager, B. I. Cook, C. J. Still, G. Husak and J. Michaelsen. 2015. Urbanization causes increased cloud base height and decreased fog in coastal Southern California. Geophysical Research Letters 42: 1527–1536. doi:10.1002/2015GL063266.

SSA Report – San Clemente Island Lotus 74 March 2020

Wolf. S., B. Hartl, C. Carroll, M.C. Neel, and D.N. Greenwald. 2015. Beyond PVA: why recovery under the Endangered Species Act is more than population viability. BioScience 65:200-207. Yatsko, A. 2000. Of Marine Terraces and Sand Dunes: The Landscape of San Clemente Island. Pacific Coast Archaeological Society Quarterly 36(1): 25-30. Zedler, P.H., C.R. Gautier and G.S. McMaster. 1983. Vegetation change in response to extreme events: The effect of a short interval between fires in California chaparral and coastal scrub. Ecology 64:809–818.

Personal Communications Booker, Melissa (San Clemente Island Natural Resources Manager & Wildlife Biologist). 2019. Documentation of take aways from series of phone calls, document edits, etc. between January and November 2019. Clemesha, Rachel (Scripps Institution of Oceanography). 2020. Email to Dawn Lawson, Re: Progress Report Fog and Coastal Low Cloud Analysis for San Clemente Island. Monday January 27, 2020. Lawson, Dawn (Adjunct Faculty, Biology Department, San Diego State University). 2019. Telephone conversation with T.M. McFarland (Texas A&M NRI), April 2019 and email to Kim O’Connor (US Navy) November 25, 2019. O’Connor, Kim (Conservation Program Manager, US Pacific Fleet). 2019. Documentation of take aways from series of phone calls, document edits, etc. between January and November 2019. McFarland, Tiffany (Senior Research Associate at Natural Resources Institute, Texas A&M University). 2019. Personal observations from trip to San Clemente Island, April 2019. Munson, Bryan (Botany Program Manager, Naval Base Coronado). 2019. Documentation of take aways from series of phone calls, document edits, etc. between January and November 2019.

SSA Report – San Clemente Island Lotus 75 March 2020

APPENDIX A

Table 14. Occupied watersheds, including the current number of locations and individuals present (adjusted for the 2017 fire season), the percent of locations and individuals near roads, the percent of each watershed that burned in the last 20 years and more than once in that timespan, and the projected individuals (within a range given low versus high growth) that will occur in that watershed in 20-30 years under each of three scenarios. Threats are represented as low (yellow) and moderate (pink). Numbers of individuals under current and future scenarios are represented as low (pink), moderate (yellow), high (light green), very high (dark green) and extirpated (red), depending on population size and number of locations (see Section 5.4 and Section 6.4). Area Area with with >1 Loc.s Ind.s Loc.s Ind.s fire in fire in Near Near in in last 20 last 20 Total Total Adjusted Roads Roads TAR TAR years years Watershed Loc.s Ind.s Ind.s (%) (%) (%) (%) (%) (%) Scen 1 L Scen 1 H Scen 2 L Scen 2 H Scen 3 L Scen 3 H WS_1006 25 2,141 2,141 8% 20% 4% 0% 0% 2,248 2,676 2,023 2,409 1,798 2,141 WS_1007 5 532 532 1% 559 665 557 663 556 662 WS_1029 1 11 11 12 14 12 14 12 14 WS_1040 2 80 80 84 100 84 100 84 100 WS_1043 8 3,241 3,241 3,403 4,051 3,403 4,051 3,403 4,051 WS_1048 2 28 28 29 35 29 35 29 35 WS_1049 1 90 90 95 113 95 113 95 113 WS_1051 1 180 180 189 225 189 225 189 225 WS_1052 3 1,504 1,504 1,579 1,880 1,579 1,880 1,579 1,880 WS_1053 8 484 484 508 605 508 605 508 605 WS_1057 6 3,402 3,402 3,572 4,253 3,572 4,253 3,572 4,253 WS_1058 22 624 624 5% 1% 655 780 652 776 649 772 WS_1059 2 19 19 20 24 20 24 20 24 WS_1060 9 2,175 2,175 2,284 2,719 2,284 2,719 2,284 2,719 WS_1061 3 26 26 27 33 27 33 27 33 WS_1062 6 1,765 1,765 1,853 2,206 1,853 2,206 1,853 2,206 WS_1064 3 98 98 103 123 103 123 103 123

SSA Report – San Clemente Island Lotus 76 March 2020

Area Area with with >1 Loc.s Ind.s Loc.s Ind.s fire in fire in Near Near in in last 20 last 20 Total Total Adjusted Roads Roads TAR TAR years years Watershed Loc.s Ind.s Ind.s (%) (%) (%) (%) (%) (%) Scen 1 L Scen 1 H Scen 2 L Scen 2 H Scen 3 L Scen 3 H WS_1065 2 6 6 6 8 6 8 6 8 WS_1066 14 178 178 187 223 187 223 187 223 WS_1068 17 558 558 6% 0% 586 698 586 698 586 698 WS_1069 1 20 20 21 25 21 25 21 25 WS_1070 3 40 40 42 50 42 50 42 50 WS_1071 4 48 48 1% 50 60 50 60 50 60 WS_1072 2 22 22 23 28 23 28 23 28 WS_1073 1 58 58 61 73 61 73 61 73 WS_1075 1 15 15 16 19 16 19 16 19 WS_1077 6 106 106 2% 111 133 111 132 110 131 WS_1078 6 38 38 40 48 40 48 40 48 WS_1080 2 131 131 25% 138 164 129 153 120 143 WS_1082 6 30 30 45% 32 38 28 33 24 29 WS_1086 1 60 60 63 75 63 75 63 75 WS_1088 2 3 3 2% 3 4 3 4 3 4 WS_1093 1 3 3 3 4 3 4 3 4 WS_1098 3 432 432 58% 454 540 388 461 322 383 WS_1099 2 150 150 7% 158 188 155 184 152 181 WS_1103 2 4 4 4 5 4 5 4 5 WS_1104 1 4 4 2% 4 5 4 5 4 5 WS_1105 2 7 7 7 9 7 9 7 9 WS_1110 2 150 150 20% 158 188 149 178 141 168 WS_1114 1 200 200 73% 59% 185 220 109 130 40 47 WS_1116 9 176 150 11% 1% 76% 38% 145 173 96 115 51 61 WS_1119 4 152 110 57% 34% 108 129 79 94 53 63 WS_1120 3 31 31 1% 33 39 32 39 32 39 SSA Report – San Clemente Island Lotus 77 March 2020

Area Area with with >1 Loc.s Ind.s Loc.s Ind.s fire in fire in Near Near in in last 20 last 20 Total Total Adjusted Roads Roads TAR TAR years years Watershed Loc.s Ind.s Ind.s (%) (%) (%) (%) (%) (%) Scen 1 L Scen 1 H Scen 2 L Scen 2 H Scen 3 L Scen 3 H WS_1123 1 30 6 56% 49% 6 7 4 5 2 3 WS_1126 1 10 10 52% 38% 10 12 7 8 5 6 WS_1131 1 50 50 61% 32% 49 58 36 43 24 28 WS_1135 3 302 301 58% 12% 309 368 252 300 198 235 WS_1137 4 331 91 59% 47% 87 103 59 71 34 41 WS_1139 6 87 84 43% 22% 84 100 69 82 55 65 WS_1141 7 425 425 79% 63% 390 464 218 260 60 72 WS_1149 1 30 30 70% 57% 28 33 17 20 7 8 WS_1156 1 1 1 1 1 1 1 1 1 WS_1158 1 1 0 100% 100% 82% 70% 0 0 0 0 0 0 WS_1159 8 467 297 82% 61% 274 326 152 182 40 48 WS_1160 5 417 415 42% 19% 419 499 348 415 282 336 WS_1161 2 13 13 46% 34% 13 15 10 12 7 8 WS_1164 1 15 15 16 19 16 19 16 19 WS_1202 2 50 50 53 63 53 63 53 63 21,25 Totals 249 1 20,743 21,595 25,708 20,627 24,556 19,706 23,460

SSA Report – San Clemente Island Lotus 78 March 2020

APPENDIX B

Table 15. Conservation measures for terrestrial plants on San Clemente Island (SCI) as relevant to Acmispon dendroideus var. traskiae, were taken from the Biological Opinion (BO; USFWS 2008) and Table 3-48 of the Integrated Natural Resources Management Plan (INRMP; US Navy 2013). Taken from Vanderplank et al. 2019, p. 14.

Source Measure Requirements INRMP AVMC-M-7 Require the following measures to reduce the potential for transport and BO of invasive plants to the island. Prior to coming to SCI, military and non-military personnel will be asked to conduct a brief check for visible plant material, dirt or mud on equipment and shoes. Any visible plant material, dirt or mud should be removed before leaving for SCI. Tactical ground vehicles will be washed of visible plant material, dirt and mud prior to embarkation for SCI. Additional washing is not required for amphibious vehicles after 15 minutes of self-propelled travel through salt water prior to coming ashore on SCI. INRMP G-M-1. Continue invasive species control on an island-wide scale, with and BO emphasis on the AVMC, IOA, TARs and other operations insertion areas such as West Cove, Wilson Cove and the airfield. A pretreatment survey to identify areas needing treatment, one treatment cycle and a retreatment cycle (when necessary) will be planned each year to minimize the distribution of invasive species. Where feasible, the Navy will include future construction sites in a treatment and retreatment cycle prior to construction. INRMP G-M-9. Conduct monitoring and control activities for invasive non-native and BO plant species outside of the Impact Area boundaries. Navy installations will prevent the introduction of invasive species and provide for their control per EO 13112. The Navy will identify actions that affect the introduction of invasive species, prevent their introduction, respond rapidly to their control, monitor populations, restore affected native species and their habitat, conduct research and develop technologies to prevent further introductions, and promote public education of the issue. BO A goal will be reducing the percent cover of invasive plants from 2008 the 1992-1993 baseline of 41% on terrace faces and 53% on terrace flats.

INRMP FMP-M-10. Conduct prescribed fire experiments to evaluate their effectiveness and BO in controlling non-native annual plants. INRMP FMP-M-11. Establish post-fire recovery plots to monitor recovery and identify and BO new infestations of non-native invasive plants associated with both wildfire and prescribed fire.

SSA Report – San Clemente Island Lotus 79 March 2020

Source Measure Requirements INRMP FMP-M-12. Evaluate burn areas and prioritize them, as appropriate, for and BO inclusion in the weed eradication program. INRMP To prevent the transfer of invasive species from the mainland to SCI, soil and fill brought to the island are treated with herbicide before importation (INRMP 2012). INRMP Further prevention for the transfer of invasive species to the island is established through the Do Not Plant list maintained by the Naval Facilities Engineering Command, Southwest Botanist and Landscape Architect (INRMP 2012).

INRMP The NRO participates in a Channel Islands biosecurity working group which meets quarterly to discuss and develop measures to prevent non-native species from invading Channel Islands ecosystems, and to share resources and knowledge of potential threats to the islands (INRMP 2012).

SSA Report – San Clemente Island Lotus 80 March 2020

APPENDIX C

Comparison to 2013 downlisting rule (USFWS 2013).

Since the 2013 downlisting rule (USFWS 2013), new data collection on the distribution of the plants has been minimal; this SSA includes only a few data points from 2013 and 2014 that were not included in the distribution considered in 2013. However, despite the lack of new survey data, this SSA will differ from and build upon the 2013 downlisting rule by both reassessing the level of threats perceived in 2013, as well as assessing the data in a different way. This SSA differs from the 2013 downlisting rule in the following ways: • Focuses on data points that were collected within the last 15 years, while the downlisting rule considered all historical points in the analyzed distribution. • Includes 39 additional datapoints representing 1,355 individuals collected in 2013 and 2014. • Moves away from a focus on “occurrences” as defined in the downlisting rule (and further discussed in Section 2.6), and instead focuses on watersheds and the estimated number of individuals in each. The downlisting rule did not consider the number of individuals. • Evaluates the current threats to Acmispon dendroideus var. traskiae; in the 2013 rule, San Clemente Island had recently adopted the Military Operations and Fire Management Plan (MOFMP) which allowed an increased amount and intensity of training activities on the island (US Navy 2008a, pp. 2–1 to 2–52), the effects of which were not well known. Since the 2013 rule, many of the potential effects of that training expansion have not been realized despite the training increase. • Evaluates the current threats to Acmispon dendroideus var. traskiae based on the implementation of several new management plans. An Erosion Control Plan, a Biosecurity plan, and a new INRMP have been implemented, all aimed to further protect the ecosystems and natural vegetation on San Clemente Island.

Survey data and opportunistic observation data collected by the US Navy (through 2012) consists of point locations of “contiguous biologically relevant clusters that were unbroken within a line of sight and did not include any obvious barriers to dispersal, pollination, or recruitment,” represented as point locations, with an associated count of the number of individuals at that location, which were also termed “occurrences” (Vanderplank et al. 2019, entire). In the 2013 dowlisting rule, these grouping of plants (hereafter “locations”) were mapped and combined with other locations that fell within 0.25 mi (402 m) of one another with any corresponding California Natural Diversity Database (CNDDB) polygons at the time of the proposed rule. These combined groups of locations were referred to as “occurrences,” and the rule defined 29 occurrences of Acmispon dendroideus var. traskiae (USFWS 2013, p. 45407- 45409) using this methodology. Therefore, these occurrences are each composed of one or more groups of plants spread across sometimes large spatial scales (Figure 15). The distribution or numbers of individuals within these occurrences (represented as polygons) was not specified. A cross-reference of overlapping occurrence numbers, canyon names, and watershed IDs are provided in Table 16.

SSA Report – San Clemente Island Lotus 81 March 2020

Figure 15. Approximate boundaries of the 29 occurrences used in the 2013 downlisting rule; polygons represent the bounding geometry (minimum convex polygons) around the the point locations and element occurrences used to define each occurrence, and canyon names (used to reference each occurrence) are provided.

SSA Report – San Clemente Island Lotus 82 March 2020

Table 16. Names referencing the 29 occurrences used in the 2013 downlisting rule, including the corresponding element occurrences contained in each, and the overlapping Watershed IDs used as units of resiliency in this SSA.

Element Canyon Occurrence Watershed IDs 1110, 1116, 1119, Eagle 1, 21 1120, 1123, 1124, 1126 1126, 1128, 1129, Bryce 1 1131, 1135, 1137 1137, 1139, 1141, North Mosquito Cove 1 1142 1146, 1149, 1159, Canchalagua 4, 23 1160, 1161, 1168, 1171 1096, 1097, 1099, Thirst 20 1100, 1101, 1102, 1103, 1104 Cave 22, 42, 43 1114, 1156, 1164 Horse 41 1106 Pyramid Head 5 1202 1072, 1083, 1086, 17, 18, 19, SHOBA Boundary north 1087, 1088, 1089, 33 1091 Twin Dams 32 1071, 1072 1050, 1051, 1052, 1053, 1056, 1057, Horton 13 1060, 1061, 1062, 1064, 1066 Tota 13 1050 Lemon Tank 16, 25 1043, 1048, 1049 Larkspur 24 1040, 1042 Chamish 3 1031, 1033 Box 40 1092 Norton 36, 38, 39 1098 Upper Middle Ranch 10 1082 Lower Middle Ranch 37 1082

SSA Report – San Clemente Island Lotus 83 March 2020

Element Canyon Occurrence Watershed IDs Waymuck 34 1080, 1082 Warren 35, 12, 20 1077, 1078, 1105 Middle Wallrock 29, 31 1068, 1079, 1085 Upper Wallrock 30 1068 1055, 1058, 1059, Seal Cove Terraces 14, 27, 28 1069, 1070, 1073, 1075 Eel Cove 26 1059 Middle Island Plateau 7 1046 Wilson Cove 11 1006, 1007 North Wilson Cove 9 1003, 1005 North Island Terraces 15 1029

SSA Report – San Clemente Island Lotus 84 March 2020