CALIFORNIA STATE UNIVERSITY, NORTHRIDGE THE ROLE OF LIFE-HISTORY TRAITS, TRADEOFFS, AND HABITAT IN THE RARITY OF SANTA MONICA MOUNTAINS DUDLEYA SPECIES (CRASSULACEAE) A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Biology by Ann Dorsey December 2009 The thesis of Ann Dorsey is approved: ______________________________________ Paula Schiffman, Ph.D. ________________ Date ______________________________________ Peter Edmunds, Ph.D. ________________ Date ______________________________________ Paul Wilson, Ph.D., Chair ________________ Date California State University, Northridge ii ACKNOWLEDGEMENTS This project would not have been possible without the help of many people. The most noteworthy are: my advisor, Paul Wilson, who helped and guided me throughout this project, my other thesis committee members, Paula Schiffman, Lawrence Talbot (who helped in the early stages and after the first draft), and Peter Edmunds (who helped in the later stages); the staff members at the CSUN Botanic Gardens, Brenda Kanno and Brian Houck, who gave me advice and a home for the study plants; all of the staff at Santa Monica Mountains National Recreation Area who helped with this project especially regarding showing me where to find the Dudleya; Stephen McCabe who shared his vast knowledge of Dudleya with me; and Steven Norris at CSU, Channel Islands, who made it possible to use the campus as one of my research locations. This project was also greatly assisted by grants from the Western National Parks Association and a Thesis Support Grant from the Office of Graduate Studies at CSUN. iii TABLE OF CONTENTS Signature page..................................................................................................................... ii Acknowledgements............................................................................................................ iii Abstract .............................................................................................................................. vi Introduction..........................................................................................................................1 Materials and Methods.......................................................................................................12 Study sites ..............................................................................................................12 Traits of rare versus common species....................................................................12 Habitat dependence: Coastal and inland gardens...................................................16 Data analysis ..........................................................................................................18 Results................................................................................................................................22 Traits of rare versus common species....................................................................22 Habitat dependence: Coastal and inland gardens...................................................30 Discussion ..........................................................................................................................38 Life-history tradeoffs .............................................................................................39 Environmental dependence of species’ prevalences..............................................42 Conclusion .............................................................................................................45 Literature cited ...................................................................................................................48 Appendix A: Interpretive background on Dudleya............................................................51 Appendix B: Nested ANOVAs and correlations ...............................................................55 Appendix C: The brick wall garden...................................................................................57 Appendix D: More information on coastal versus inland gardens ....................................62 Appendix E: Watering treatments......................................................................................65 iv Appendix F: Differences between confusable taxa ...........................................................77 Appendix G: Management implications and recommendations........................................83 v ABSTRACT The Role of Life-history Traits, Tradeoffs, and Habitat in the Rarity of Santa Monica Mountains Dudleya Species (Crassulaceae) by Ann Dorsey Master of Science in Biology In this study, life-history traits, tradeoffs in those traits, and habitat characteristics of rare and common species were compared in an attempt to explain differences in species distributions. The nine Dudleya species occurring in and around the Santa Monica Mountains were studied. Five are rare narrow endemics with small localized ranges, one is rare with an intermediate range, and three are common with broader ranges. Life-history traits were studied in wild populations and in plants grown from wild collected seeds. Habitat characteristics were recorded in two or three populations of each study species. Differences that could explain why the rare species have smaller distributions than the common species were found in aspects of growth, reproduction, and habitat specialization. In regard to life-history traits and tradeoffs, rare species grew to a smaller size and reproduced earlier than common species. The small body-size of the rare species was correlated with smaller reproductive outputs than the larger-bodied common species. The rare species also tended to have lower seedling survival. Reproductive output and survival affect population size, persistence, and dispersal, all of which affect vi species distributions. The habitat requirements of the study species differed in terms of co-occurring vegetation, geology, and microclimate, with the rare species being more restricted compared to the common species. To further understand how habitat plays a role in limiting species distributions, the nine species were grown in an inland garden and in a coastal garden. Plants of all species grown in the inland garden were smaller in size than those in the coastal garden. Moreover, the growth disparity between plants in the two gardens was greatest for the rare species. The rare species have a lower tolerance for hot dry conditions compared to the common species. In the Santa Monica Mountains region, the habitat conditions required by the rare species are not as prevalent as those of the common species. Differences in life-histories constrained by tradeoffs affect the prevalence of the species, as well as specialization on rare habitats. vii Introduction Evolution is a dance between the influences of genes and the environment sorted at different scales in time and space with different processes working at each scale. Willis and Whittaker (2002) and Whittaker et al. (2001) synthesized the different scales in time and space with the ecological processes that occur at those scales. At the global scale, over a span of 10 to 100 million years, evolution is shaped by processes such as sea level changes and plate tectonics. At the continental scale, over 1 to 10 million years, mountain building, aridification, and glacial cycles affect species richness and the distribution of major lineages across continents. At the regional scale, over tens of thousands of years, water-energy dynamics and climate influence the overlap of species ranges across large geographic areas within continents. On the scale of a landscape over hundreds of years, soils and topography are important to the turnover of species within a community and species richness between adjacent communities. Finally, at the local scale, individuals and groups of individuals living approximately at the same time vie against one another for resources, thereby making habitat structure, disturbance, and ecological interactions agents of selection. These delineations of processes into a given scale are not absolute. Processes at a higher scale can affect those at a lower scale and visa versa. Also, processes at different scales may have a combined effect. Levin (2000) proposed that as species evolve they pass through four stages: origination, expansion, differentiation, and decline to eventual extinction. This passage is unique for each species and is affected by its ability to disperse, interactions with other organisms, habitat conditions over time, ability to adjust to environmental change, genetic variation, and developmental flexibility. A species’ geographic range changes as 1 it passes through these stages. It is defined by interactions between the species’ traits, particularly their life-history, demographic, and dispersal characteristics, as well as resource requirements and environmental tolerances. Environmental conditions, especially the features that vary in space and time limiting population size and distribution, also have an impact. The particular environmental conditions that restrict a species’ distribution and abundance are unique to it and are expressed in the inability to survive and reproduce in habitats with different sets of conditions. When there are many conditions that limit a species, the places where it can live will be few and widely separated (Brown et al. 1996). The longevity of a species (how quickly it passes through the four stages) can be affected by its characteristics and range size. Characteristics that can reduce the chances of extinction are (1) life-history traits that allow for a rebound in population size following a harmful environmental change, (2) high dispersal and recruitment ability that allows the species to recolonize an area after local extinction, and (3) being a generalist as opposed to a specialist because generalists tend to be locally abundant and geographically widespread (McKinney 1997). Species with large geographic ranges and population sizes tend to survive for longer than species with small ones as a consequence of one or a combination of the following factors: environmental and demographic stochastity, population fragmentation, the Allee effect, inbreeding depression, or reduced genetic variation within populations (Gaston 1994, Kruckeberg and Rabinowitz 1985). Differences between species in characteristics that affect population persistence and range size can lead to some species becoming rare and others common. Specifically, 2 species distributions can be impacted by relationships between life-history traits and the availability of a species’ habitat. One evolutionary process that can affect species’ distributions is the expression of tradeoffs between life-history traits. In a review of life-history literature, Stearns (1992) discusses the role of tradeoffs in constraining evolution. Life-history traits include size at birth, growth pattern, age at maturity, size at maturity, number, size and sex ratio of offspring, age and size-specific reproductive investment, age and size-specific mortality schedules, and length of life. Tradeoffs occur when there are linkages between traits such that the simultaneous evolution of two or more traits is not independent. Nonindependence can ensue from physiological limitations such as energy resources being finite. For example, there might be a tradeoff between rapid growth when small versus the capacity to survive to an older age. Rapid growth at a small size is favored because reproduction at a larger size leads to a greater reproductive output, but putting energy into rapid growth instead of storage decreases the chances of surviving to reproduce later in life (Metcalf et al. 2006). The allocation can be depicted as trading off within an individual, but tradeoffs are manifested among individuals as they come to differ (Stearns 1992). Some individuals may put more energy into rapid growth when small, while others put energy into storage when small. Divergent populations that persists in different habitats may have one trait (rapid growth) selected for in one habitat type and the other (energy storage) in another habitat (Levin 2000). At the macroevolutionary level, over time, the selection of the different traits in the different populations could lead to the evolution of new species. Some species within a genus may express traits at one end of a tradeoff envelope (rapid growth but poor energy storage) and other species the other 3 (good energy storage but slow growth). The traits involved will be negatively correlated. Though both traits in a tradeoff may be adaptive, one may lead to greater fitness than the other. This, in turn, could lead to distributional differences between the species (Levin 2000). Beyond the life-history traits of the now divergent species, there will also be other modifications because the species will have adapted to the different habitats in which they live. Habitat conditions that determine which species are found at a site can be biotic or abiotic. Biotic examples include other plant species, herbivores, pollinators, and parasites (Kruckenberg and Rabinowitz 1985, Stebbins 1978). Abiotic conditions include edaphic conditions, temperature, rainfall, and day length (Levin 2000). A species will be able to live in some range of these conditions and not in others. An example of this for biotic conditions would be the presence of other plant species. They can limit the distribution of a given species either because of competitive exclusion (Ackerly 2003) or because of facilitative associations (Kruckenberg and Rabinowitz 1985). An abiotic example could occur in an environment characterized by gradients of conditions (e.g. soil moisture content). Species along this gradient will have traits that allow them to be most competitive along a part of this gradient. They will not be able to perform well across the entire range because of lack of physiological tolerances to some of those conditions, which limits where they can occur (Cavender-Bares et al. 2004). Species in the genus Dudleya exemplify these processes in which diversification has been seemingly constrained along tradeoff envelopes and by differences in habitat with concomitant effects on the geographic expanse of the resulting species. The species to be compared are nine terminal taxa of Dudleya that occur in and around the Santa 4 Monica Mountains (located just west of Los Angeles, California: Dorsey 2007, Appendix A). Taxonomists have been inconsistent as to whether all these terminal taxa are to be treated uniformly at the rank of species or in some cases ranked as subspecies or even synonomized, but for the rest of this thesis they will all be called “species”. Although some of the characteristics of the rare species may be consequences of rarity and others causes (Kunin 1997, Gaston 1994), no attempt is made to differentiate between consequences and causes. Also, the terms “distribution” and “prevalence” will be used to describe the relative abundance of the species – how many and how large the populations are and how those populations are arranged geographically. A few key characters of the nine species are introduced in Table 1. Of the nine species, five are federally listed as threatened, and have distributions restricted even within the Santa Monica Mountains. One species, D. b. blochmaniae, is rare but has no legal status; it has a range size intermediate between the listed species and the common species. The remaining three species are common, occurring from as far north as Monterey County to as far south as Baja California. A preponderance of Dudleya species, over half of the 44 treated in The Jepson Manual, are said to be uncommon, rare, threatened, or endangered (Bartel in Hickman 1993). There are many factors that could explain this. A brief review of the evolution of Dudleya sets the context. Most Dudleya species live within a few miles of the Pacific Ocean from Baja California to southwestern Oregon, with a few having ranges that extend much farther east. This coastal terrain was formed by the movement of three tectonic plates. Going back 30 million years, scattered pieces of land that were part of the Farallon Plate came into contact with the North American Plate, and then about 27 5 million years ago this aggregate was pushed against the Pacific Plate. The resulting landmass consisted of a very complex topography made from different types of rocks, facing different directions resulting in many different types of habitats. The major lineages of Dudleya are believed to have originated here (starting about 5.5 million years ago), then diversified on the Pacific Plate, i.e., west of the San Andreas fault (Uhl 2004, 1994). Less than 3 million years ago, summer rains ended along the coast of California because of shifts in glacial climatic cycles and ocean circulation, and the rise of mountains along the coast. Only plants that had traits allowing them to tolerate summer drought persisted (Ackerly 2003). Dudleya species with their succulent leaves and CAM photosynthesis were able to live in local sites that were sheltered from the worst of the summer heat. The mountain ranges running along the coast of California created very different cismontane and transmontane climates (Stebbins 1978). The coastal predominance of Dudleya (Uhl 2004, 1994) could be because of this climatic difference. The varied topography of the coastal mountains and multitude of soil types derived from the many different geological formations allowed for the creation of pockets of unique environments (Stebbins 1978) contributing to the very local distribution of many Dudleya species (Uhl 1994). The prevalence of the conditions a species is adapted to can determine how common it is (Gaston 1994, Burgman 1989, Kruckeberg and Rabinowitz 1985). The rare Dudleya species included in this study have small geographic ranges, narrow habitat specificity, are locally abundant in the specific habitats where they occur, and are considered narrow endemics using the criteria of Rabinowitz (1981). The theme to be explored below is that Dudleya species’ distributions are constrained by life-history 6 tradeoffs at a macroevolutionary level and by habitat at a microevolutionary level. In particular, adapting to reproduce quickly and making the best of an unusual rock type may keep a species that has specialized from expanding beyond a few sites, in contrast to a species that has adapted to delay reproduction and grow to be larger and more stressresistant. This thesis, then, explores possible reasons for differences in the prevalences of the nine Dudleya that live in the Santa Monica Mountains and surrounding areas. Two topics are woven throughout the thesis. First, traits of rare versus common species were contrasted to establish in which traits, if any, there were differences. The traits were correlated against one another to determine which were involved in tradeoffs. The hypotheses tested were that rare species would be inferior in regard to some aspect of survivorship or reproduction, but would be better in life-history traits that were accentuated as opposed to those that were traded off. Also, aspects of habitat were contrasted. Data were collected on the microenvironments in which the nine species occur. These habitat characteristics were compared to ascertain, as hypothesized, if rare species occurred in more specialized habitats than common species. Second, the habitat dependence of variations in performance among species is presented. Specifically, the hypothesis tested was that rare species would be more impacted than common species by harsh inland conditions compared to conditions where there is a coastal influence. Differences in performance on the coast and inland were correlated with more conventional life-history traits. For instance, the growth of individuals at one location was compared with coastal and inland reproduction to determine if the disparity between sites was a consequence of a tradeoff. 7 Table 1. Rarity and distinctiveness of nine Dudleya taxa. Abbreviation Subgenus epithets D. Hasseanthus bloc blochmaniae subsp. blochmaniae Rarity Distributiona Rare but not San Luis Obispo Co. to listed northern Baja California along coast Petal charactersb White, ± yellowish green at the base, red on the keel, 2.5-3.5 mm wide Pale yellow sometimes red-lineolate on the keel, 2-3.5 mm wide Lemon yellow with a touch of green along the midrib, 2.5-4 mm wide Slightly orange yellow often marked with red, 2.5-3.5 mm wide Bright yellow, 2-2.5 mm wide Other distinguishing charactersb Rosette leaves vernal, resprouting from a corm D. Dudleya abramsii subsp. parva parv Threatened Ventura Co. D. Dudleya verityi veri Threatened Ventura Co. D. Dudleya cymosa subsp. marcescens marc Santa Monica mts D. Dudleya cymosa subsp. ovatifolia ovat USFWS threatened; CA rare Threatened D. Dudleya cymosa subsp. agourensis agou Threatened Santa Monica mts D. Dudleya lanceolata LANC Common D. Dudleya caespitosa CAES Common D. Dudleya pulverulenta subsp. pulverulenta PULV Common Santa Barbara Co, to Rosette leaves green or northern Baja glaucous California Orange-yellow to red, 2.5- Plants are clonal with up Monterey County to Los Angeles County, near the 5 mm wide to 150 rosettes coast San Luis Obispo Co. to Red & glaucus on keel, 2-4 Rosette leaves chalkycentral Baja California mm wide pulverulent on coastal ranges (ornithophilous) Santa Monica mts Bright yellow, occasionally glaucous along the midrib Yellow to red, 2.5-5 mm wide a Rosette leafless in summer Conejo volcanic breccia Plants are clonal with 25- Volcanic rock 100 rosettes outcrops Rosette leaves wither in the summer Sheer volcanic surfaces Rosette leaves green, not glaucous, reddish on the underside Rosette leaves glaucous Sedimentary conglomerate Distribution from Moran 1951 , Nakai 1983 , Nakai 1987, for a more detailed description of habitat and range see Dorsey 2007. Floral and other distinguishing characters from Moran 1951, Nakai 1983 , Nakai 1987, Aigner 2004, and pers. obs. c USFWS 1999, Bartel in Hickman 1993. b 8 Associated geologyc Rocky clay or serpentine soil Pleistocene dissected gravels Various rocky slopes Various rocky sites Rocks, mineral soil Materials & Methods Study sites Seeds were collected and habitat observations were made in the Santa Monica Mountains and surrounding areas in two counties. Sites in Los Angeles County were: Malibu Creek State Park (34.103o N, 118.734o W), Cornell Corners (34.110o N, 118.775o W), Semler Open Space (34.146o N, 118.806o W), and Seminole Hot Springs (34.107o N, 118.791o W). Sites in Ventura County were: Joel McCrea Wildlife Preserve (34.236o N, 118.859o W), Wildwood Regional Park (34.219o N, 118.919o W), Rancho Potrero Open Space (34.167o N, 118.946o W), Thornhill Broome Beach (34.084o N, 119.036o W), Leo Carrillo State Park (34.046o N, 118.928o W), Conejo Mountain (34.188o N, 118.984o W), and California State University, Channel Islands (34.162o N, 119.043o W). Traits of rare versus common species Field observations – Twelve sites were studied, each with 1 to 3 species of Dudleya. At the time of fruit collection, data on the parent plant’s number of inflorescences, tallest inflorescence height, number of fruits, longest leaf length, and number of leaves were recorded. Information was noted about the habitat the Dudleya were growing in: any vegetation the plant was growing in close proximity to, the distance to the nearest neighbor of the same species, slope, aspect, and if any shading was provided by trees, cliffs, etc. Seed collection – Fruit ripeness was monitored from June through August 2005. Fruits were collected from mid-July through the end of August when they looked dry but before the follicles had opened. Three randomly chosen fruits each were collected from 12 30 randomly chosen individuals, from 2 or 3 populations of each species (three for D. b. blochmaniae, D. lanceolata and D. p. pulverulenta, and two for D. a. parva, D. verityi, D. c. marcescens, D. c. ovatifolia, D. c. agourensis and D. caespitosa). Fruits were cut from the inflorescences and placed individually in labeled envelopes. Collection deviations – Many of the D. p. pulverulenta had been partially eaten by deer or rabbits so only plants that had intact inflorescences and leaves were used for fruit and data collection. Only one fruit per plant was collected from D. c. marcescens and two fruits from D. c. ovatifolia because of their protected status. Leaf number, and nearest neighbor distances may have been underestimated for D.verityi and D. caespitosa because individual plants are capable of forming multiple rosettes. Only the main rosette was studied. Also, rosettes of the same individual may have been considered as belonging to a neighbor. At the time of seed collection leaves were not present for D. b. blochmaniae and were not present on all D. a. parva plants. Therefore, leaf-related data is missing for these two species. For D. p. pulverulenta and D. lanceolata, in some cases, the individuals were too sparsely spaced to include five neighbors. In addition, it was not always possible to distinguish D. b. blochmaniae plants from their neighbors, because at the time of collection only the inflorescences were present. It was assumed each inflorescence was a separate plant, but later observations revealed there could be multiple inflorescences. Nearest neighbor distances, therefore, were underestimated for this species. Although precautions were taken to ensure the purity of the species at the time of data collection, some of the D. c. marcescens and D. c. ovatifolia individuals’ data and seeds collected may actually be from Dudleya hybrids. 13 Seed preparation – Fruits were stored in envelopes at room temperature and allowed to dry fully for at least one month before the seeds were counted. Follicles were held over a piece of white paper inside a box lid. Seeds were shaken loose as much as possible, then follicles were broken open over the paper to release the remaining seeds. Non-seed material was removed. Seeds were picked up by gently placing an index finger over them. Seeds on the finger tip were counted and placed into a labeled envelope with the corners folded up to prevent the loss of seeds. Seeds from the fruits of the same individual were placed in the same envelope until there were 30 seeds. This process was repeated for a second envelope. Any seeds remaining were placed in a third labeled envelope. All seeds of all collected fruits were counted. Seeds were stored in a refrigerator. Seed projected area – Seeds from a randomly chosen subset of individuals and fruits of each species were measured using the ocular micrometer of a dissection microscope. Seeds were placed in a small box lid on a piece of white paper and were placed under the microscope for observation. The largest (presumably healthiest) five seeds were chosen for measurement in order to include only viable seeds. Chosen seeds were picked up on a piece of transparent tape and taped onto an index card for measurement under the microscope. Length and width of each seed were recorded. Once all of the seeds were measured, ocular units were converted to millimeters. Projected area was estimated as the surface area of an ellipse (SA = πr1r2). Seed sowing and seedling care – On 20 December 2005 sixty seeds of each individual for all nine species were surface sown in 4” plastic pots (30 seeds per pot, 2 pots per mother plant). Not all mother plants of all species had 60 seeds, in which case all 14 the available seeds were sown, 1/2 in one pot and 1/2 in the other. A few individuals produced no seeds (one individual each of D.verityi, D. c. ovatifolia, D. lanceolata, and D. caespitosa). The potting mixture consisted of 2 parts sunshine potting mix, 2 parts vermiculite, and 1 part perlite. Vermiculite was also sprinkled over the top of the filled pots before the seeds were sown. Pots were labeled as to species and mother. After the seeds were sown, they were watered with low water pressure to avoid the spread of seeds from one pot to the next. Pots were kept in flats in a greenhouse. The flats were rotated 180o and along the bench weekly. The potting mix was gently watered as necessary to keep the mixture moist. Once the seedlings had emerged and were established, the mixture was allowed to dry somewhat between waterings to prevent fungus growth, the proliferation of fungus gnats, and to harden the plants. Seedlings were fertilized the first time 13 January and were fertilized weekly until 30 March 2006. The first fertilization was with 1/4-strength Miracle Grow (15-30-15). The rest of the fertilizations were with Grow More (20-20-20) at alternating strengths 1/4 or 1/3 until 17 March, at which time the strength was increased to 1/2. The final fertilization was 25 April after all plants had been transplanted and placed outside in a shade house. Germination and seedling survival – Numbers of sprouted seedlings were counted from the first week after sowing, 27 December 2005, weekly until 1 February 2006 when the seedlings became too crowded to accurately count their numbers. Seedlings were transplanted once most or all of the individuals in a pot had 2 to 3 post-cotyledonous leaves. Transplantation started 20 February 2006 and continued until 21 April 2006. Most of the transplantation occurred in April. Final counts of seedlings were taken at the time 15 of transplantation. Seedlings were transplanted individually into labeled plastic 4” pots. Once transplanted the seedlings were watered. Eleven of the biggest, healthiest looking seedlings (siblings) from each mother plant were transplanted into pots. The seedlings transplanted into pots were randomly assigned to 1 of 6 treatment groups or to be extras. One set was grown between bricks. One set was grown in a coastal garden, and another in an inland garden. The remaining three sets were assigned to three watering-treatments (Appendix E), which in the present context are only relevant in that they were used to measure each species’ proclivity to bolt. Plants were allowed to acclimate for approximately one month before the first experimental treatments commenced. Brick wall garden – A set of plants were grown with the roots sandwiched between vertically stacked bricks. This mimicked the way Dudleya often grow in nature, inside cracks in rock outcrops. It also facilitated the measurement of roots and shoots of simultaneously established plants. During planting, drip irrigation lines were installed. Watering was more generous than it would have been in nature but was still seasonal. Plants were harvested after two summers. Roots were dried separately from shoots. Details are given in Appendix C. Habitat dependence: Coastal and inland gardens Site descriptions – Two gardens were planted in different near-natural environments, inland and coastal, within the naturally occurring range of the common species. The coastal garden site was in Zuma Canyon (34.106o N, 118.819o W), a site that in late spring frequently receives maritime fog and cloud cover. The inland garden was inland of the first ridges in the hills overlooking the San Fernando Valley to the north 16 (34.118o N, 118.585o W). Plants at the coastal location were planted out 21 June 2006. Plants at the inland location were planted out 7 June 2006 and (only a few) on 9 June 2006. Three plots per garden location, roughly 6.5’ × 3.5’, were dug to a depth of 4 – 6”. A sheet of 6’ × 3’ hardware cloth was placed on the bottom of the plot. Plants of all the species were placed in random order in the dug out area on top of the hardware cloth. The potting mixture in the pots was retained in the garden so that the plants were close to the same level as the surrounding ground. Individuals were placed next to each other in rows of four. As each row was filled, the next row was put in place until a 6’ × 3’ area was filled. Soil from the location was added to fill in gaps between plants as necessary. Coastal plots contained 83 to 85 individuals each. Inland plots contained 80 to 88 individuals each. Exclosures (6’ × 3’ × 1.5’) constructed of 2 × 4s and hardware cloth were placed over the plots to prevent herbivory. HOBOs were placed at each garden site at the time the plants were planted out to record temperature, dew point, and humidity in the coming year. The coastal garden was shaded by Platanus racemosa and Juglans californica trees. The inland garden was shaded by Juglans californica trees. Data collection – Plants were watered once they were planted out and weekly thereafter for 4 weeks to allow them to become established in the gardens. After that, they received only natural precipitation. Observations of the general condition of the plants were noted every week from the time of planting to 20 August 2006 after which time observations were made every two weeks. One month after planting, 4 plants in the inland garden were lost to small mammals that had broken through the hardware cloth, and some coastal garden plants lost outer leaves to pill bugs. This damage did not appear to interfere with the plants’ growth. Flowers were cut off the inflorescences as necessary 17 to limit the amount of energy allocated to reproduction (Jongejans et al. 2006), control for energy allocation differences between reproductive and non-reproductive individuals, and to prevent the accidental introduction of Dudleya species to the sites. Every two months, from June 2006 to June 2007, data were collected on the number of leaves, longest leaf length, and number of inflorescences of each individual. Final size and reproductive data were collected 12 November 2007 for the coastal garden and 10 November 2007 for the inland garden. Coastal garden plants were removed 15-16 December 2007. Inland garden plants were removed 24-26 November 2007. At the time of removal, plant roots were cut just below the rosette, any dry inflorescence parts were removed, ink prints of the crosssectional area of the root crown below the leaves were made, and the rosettes were placed into labeled paper bags. The bags were dried at 37.8o C for several months until weights stopped decreasing. Data analysis Habitat and life-history data – Mixed-model nested ANOVAs were done comparing rare and common species, and testing for significant differences among species within rarity categories. The “rare” category included six species (except for leaf data in which case it included four species) and the “common” category included three species. These were done for nearest same-species neighbor distances, slope, degrees from north, log longest leaf length, square root leaf number, log inflorescence height, cube root fruit number, seeds per fruit raised to the negative one-sixth power, seed projected area, arcsine percent germination, and percent seedling survival. 18 Germination and seedling survival – Germination percentages were estimated by using the highest number of seedlings counted per individual divided by the total number of seeds sown. Seedling survival percentages were estimated by multiplying the probability of germinating by the probability of surviving from the germination stage to the transplantation stage. Reproductive output and percent bolting – Reproductive output was estimated by multiplying the number of fruits produced on a given individual by the average number of seeds per fruit on that individual. Species averages were then calculated. Percentage of reproductive individuals for each species during the 1st, 2nd and 3rd springs after germination were calculated by dividing the number of individuals with inflorescences in each species in the three watering treatment groups by the total number of individuals of that species in these groups. Brick wall – Two variables were considered from the plants grown in the brick wall. Total biomass dry weight was the sum of rosette biomass dry weight and root biomass dry weight. Root-to-shoot ratio was the quotient of these two numbers, and used as a measure of allocation. Evidence of tradeoffs – Correlations among species were calculated. In some cases, they did not include D. b. blochmaniae because many of the plants of that species were not in evidence above ground for much of the year. Individual Pearson correlations were done comparing (1) the percent of each species reproductive over three springs (summed) with dry weights of plants grown in the brick wall garden (not including D. b. blochmaniae), (2) the root-to-shoot ratios and the total biomass dry weights from the plants grown in the brick wall garden (excluding D. b. blochmaniae), (3) the seed 19 projected area and the number of seeds per fruit for all study species, (4) the size and reproductive output of all the study species except D. b. blochmaniae from plants growing in the wild, and (5) seed projected area and percent of individuals that survived to be seedlings for all species. Differences between coastal and inland gardens – A variable indexing size was calculated as log ([longest leaf length × number of leaves] + 1). This “leaf size” variable was subjected to a repeated measures ANOVA in which site, rarity, and species-withinrarity were crossed with census date, “site” being the one coastal versus the one inland garden location. For the variables measured at the end of the experiment, namely root crown cross-sectional areas and rosette dry weights, split-plot ANOVAs were done involving site, rarity, and species within rarity. The above analyses did not include D. b. blochmaniae. The root crown cross-sectional areas of two plants (D.verityi #15 in the inland garden and D. caespitosa #25 in the coastal garden) were winsorized because they greatly exceeded the other values for that species. Pearson correlations were used to compare a life-history variable and a coastal minus inland difference to test for the presences of a tradeoff between these traits. For leaf size, the date with the maximum differences between the coastal and inland gardens was used. This difference was correlated against a measure of size from different plants, the total biomass dry weight of plants grown in the brick wall garden. A Fisher’s exact test was used to see if the percent of individuals that were reproductive for each species in the coastal garden differed from those in the inland garden. The difference in percent reproductive in the coastal and inland gardens was correlated against the total biomass dry weight of the plants grown in the brick wall garden, not including D. b. blochmaniae. 20 Statistical caveats – In many analyses, the factor “rarity” is included. It consists of three common species (D. lanceolata, D. caespitosa, D. p. pulverulenta) versus five or six rare species (sometimes excluding D. b. blochmaniae). With such small sample sizes, null hypotheses should be accepted cautiously, and even non-significant differences noted as possibly representing a difference. Slightly more powerful are correlations in which eight or nine species are used as data points. The eight or nine were treated as independent even though these species are related at various phylogenetic levels. Differences between the species or between rare and common groups could be because of evolutionary adaptations required for survival in the unique microhabitats or because of phylogenetic conservatism of the traits involved (Westoby et al. 1995). It is not clear which of these causes may have influenced the study species. Another caveat is that there was pseudoreplication in the analysis of the gardens (Hurlbert 1984). There were only two gardens and the analyses were performed as if individual plants were replicates. Given the difficulty of having multiple gardens of each type and the observable and measurable differences between the gardens, the significance tests seem useful to check for variance greater than the individual variance within species in a garden. 21 Results Traits of rare versus common species Life-history data – Appendix B and Dorsey (2007) give some growth and lifehistory information for each of the nine species. Briefly, longest leaf length ranged from 1.4 to 11.7 cm. The leaves of common species were significantly longer than those of rare species (P = 0.009, F = 17.199, df 1, 5). The number of leaves ranged from 5 to 44. Dudleya cymosa ovatifolia and D. c. marcescens, both rare species, had the fewest number of leaves and D. p. pulverulenta, a common species, had the greatest number of leaves, but differences between rare and common species were not significant (P = 0.149, F = 2.910, df 1, 5). Inflorescences varied between 11 and 88 cm tall, with common species having taller inflorescences than rare species (P < 0.001, F = 46.838, df 1, 7). The number of fruits per individual ranged from 10 to 157. Rare species produced significantly fewer fruits per individual than did common species (P = 0.008, F = 13.401, df 1, 7). The numbers of seeds per fruit, ranged from 17 to 551, and were greater for common species than for rare species (P = 0.031, F = 7.280, df 1, 7). Seed projected area ranged from 0.185 to 0.487 mm2, with rare species generally producing smaller seeds than common species (P = 0.009, F = 12.700, df 1, 7). Rare species tended to have a greater number of individuals reproduce each spring than common species (Figure 1). In the first and third springs more individuals of each rare species reproduced than of each common species. In the second spring, most of the rare species had a greater number of individuals reproduce than the common species. The exceptions were the summer deciduous D. a. parva (97%) and D. b. blochmaniae (96%), which were equal to or 22 slightly less than D. lanceolata (97%), the common species with the greatest number of reproductive individuals. 100 1st spring 90 2nd spring % reproductive 80 3rd spring 70 60 50 40 30 20 10 0 bloc n = 82 n = 82 n = 79 parv veri marc ovat 78 78 73 90 90 88 84 82 79 75 74 74 agou LANC CAES PULV 90 90 88 87 87 87 87 87 87 78 1st spring 77 2nd spring 77 3rd spring Figure 1. Percent of individuals of each species reproductive in each of three springs (2006-2008). Data from plants in the three water treatment groups. The number of individuals with inflorescences was divided by the total number of individuals for each species. Survival – The survival of individuals of all of the study species to three different stages of life were tracked (Figure 2). Survival to the germination stage ranged from 4380%. Of all of the stages, this was the one that showed the greatest decline in numbers of surviving individuals. Survival to the seedling stage, age 2 to 4 months (just before they were transplanted) ranged from 30-75%. After plants were transplanted very few died. Survival to the end of the study (age ~ 2 years) ranged from 30-75%. Although there were significant differences among species there were not between rare and common species when comparing percent germination (P = 0.013, F = 10.875 and P = 0.252, F = 1.558 respectively, df 1, 7) and percent survival to the transplantation stage (P = 0.002, F = 21.218 and P = 0.302, F = 1.239 respectively, df 1, 7). 23 100 80 % survivorship LANC parv CAES 60 bloc agou veri PULV ovat 40 marc 20 Seeds Germinated Transplanted End Figure 2. Comparing species in terms of survival to different life stages. Seeds: number of seeds sown. Germinated: maximum number of seedlings divided by number of seeds sown. Transplanted: probability of seeds germinating multiplied by probability of survival from germination to transplantation. End: probability of survival to transplantation multiplied by probability of survival from transplantation to the end of the study (all groups). Life-history tradeoffs – Correlations were calculated between life-history traits in order to determine which were involved in tradeoffs. The brick wall biomass dry weights were strongly negatively correlated with the sum of the percent of reproductive individuals over three springs in the water treatments (r = -0.846, P < 0.01, n = 8: Figure 3A). Rare species were smaller and had more individuals reproduce than the larger common species. Some traits were negatively related but not significantly. The correlation of root-to-shoot ratios plotted against total biomass dry weight (roots plus shoots) was negative and non-significant (r = - 0.438, P > 0.20, n = 8: Figure 3B). 24 B -0.2 3 parv parv LANC -0.3 ovat 2.5 Reproduction over 3 springs veri marc agou -0.4 agou marc 2 log(root:shoot) A LANC 1.5 1 CAES CAES veri -0.5 -0.6 ovat -0.7 -0.8 0.5 -0.9 PULV r = -0.846, P < 0.01 0 -0.75 -0.25 PULV r = -0.438, P > 0.2 0.25 0.75 log(dry weight, g) -1 -0.75 1.25 C -0.3 -0.5 -0.25 0 0.25 0.5 log(root+shoot) 0.75 PULV 4.5 -1/6 -0.4 LANC -0.45 -(Seeds/fruit) log (reproductive output in wiild) -0.35 -0.5 ovat agou -0.55 marc -0.6 veri CAES parv bloc r = -0.269, P > 0.20 -0.7 0.15 0.2 0.25 0.3 0.35 0.4 0.45 LANC parv 70 CAES 65 bloc agou 55 50 PULV veri ovat 40 35 30 marc r = 0.792,P < 0.02 25 0.16 0.2 LANC veri 3 marc ovat r = 0.860, P < 0.01 E 80 60 agou parv Seed projected area, mm2 75 CAES 4 3.5 -0.65 % Seedling survival 1.25 D 5 PULV 45 1 0.24 0.28 0.32 0.36 Seed projected area, mm2 0.4 0.44 25 2.5 -0.5 0.5 1.5 log(size in wild) 2.5 Figure 3. Signs of life-history tradeoffs and other relationships. A. Biomass dry weights of each species are from plants grown in the brick wall garden. The percent of individuals of each species reproductive each of three springs were summed. Data were from plants in the three water treatment groups in 2006-2008. B. Root-to-shoot ratios are biomass dry weights of roots divided by that of rosettes. Root + shoot are biomass dry weights of roots and rosettes added together. Data from plants grown in the brick wall garden. C. Projected area of seeds was calculated using the length and width of the seeds and the formula for the area of an ellipse. Number of seeds per fruit generally from 3 fruits of 30 plants growing in the wild in 2 or 3 populations were counted. D. Size was measured as the number of leaves times the longest leaf length. Reproductive output is the number of fruits times the number of seeds per fruit. Data were collected from 30 plants of each species in 2 or 3 wild populations. E. Projected area of seeds was calculated using the length and width of the seeds and the formula for the surface area of an ellipse. Seedling survival was measured as the product of the probability of seeds germinating and of seedlings living to be transplanted. 26 Total biomass was involved in a life-history tradeoff with rare species being lighter and common species heavier, but most of the species had similar root-to-shoot ratios ranging from 0.28 to 0.55 (22 – 35% root). Dudleya a. parva, a species that dies back to a caudex in the summer, had a disproportionately large root mass of 0.84 g (46% root). Dudleya p. pulverulenta had a disproportionately large rosette mass of 0.13 g (11% root) and is one of the large-bodied species with the greatest delay in reproduction. (There are no data for D. b. blochmaniae for which root-to-shoot data are not applicable because they die back to corms: Appendix C, Figure C-3). There was a non-significant negative correlation between seeds per fruit versus seed size (r = -0.269, P > 0.20, n = 9: Figure 3C). Rare species tended to have fewer seeds per fruit, the exceptions being D. c. agourensis as compared to D. caespitosa, and rare species generally produced smaller seeds than common species, the exception being D. p. pulverulenta, which produced the smallest seeds. Other pairs of characteristics were not involved in tradeoffs and included size in the wild (number of leaves multiplied by the length of the longest leaf) and reproductive output (number of fruits multiplied by seeds per fruit) which were strongly positively correlated (r = 0.860, P < 0.01, n = 7). Smaller sized rare species had smaller reproductive outputs than larger sized common species (Figure 3D). Finally, seed projected area and percent seedling survival were positively correlated (r = 0.792, P < 0.02, n = 9). Species with smaller seeds tended to have fewer individuals survive to the seedling stage (Figure 3E). Similarly, seed projected area and percent germination were positively correlated (r = 0.712, P < 0.05, n = 9). Species with smaller seeds tended to have fewer individuals germinate. Correlations are treated in a more thorough fashion in Appendix B. 27 Average distance to near neighbors 1600 A 1200 800 400 0 200 Degrees from north B 150 100 50 0 c rv ri rc at ou NC ES UL blo pa ve ma ov ag LA CA P V Figure 4. Comparison of species for two microhabitat variables. A. Average distances (cm) from nearest same-species neighbors for each species. The distances of the five nearest same-species neighbors were measured for 30 individuals of nine species in 2 or 3 populations. The distances were averaged per individual and then by species. B. The aspects of the substrate for 30 individuals of nine species in 2 or 3 populations. The data were then converted into degrees from north. For both variables, rare species were not significantly different from common species, but there were significant differences among species. Habitat data – The average distance to the five nearest neighbors (a measure of population density) ranged from 6.3 to 501.4 cm, with the common species being the sparsest and the rare species being the densest (Figure 4A). There was significant variability among species (P < 0.001, F = 37.397, df 7, 261), but not between rare and common species (P = 0.317, F = 1.160, df 1, 7). Slope of the substrate a given individual was growing on ranged from flat (0o) to vertical (90o). There was no significant difference between rare and common species (P = 0.807, F = 0.065, df 1, 7), but there 28 was significant variability among species (P = 0.003, F = 20.187, df 7, 261). Aspect of the substrate a given individual was growing on ranged from 0 (directly north) to 180o from north (directly south). Most individuals of both rare and common species grew on substrates with aspects within 90o of north, but common species had more individuals facing further away from north (Figure 4B). Of the forty individuals that faced greater than 90o from north, only five were rare. They were D. c. ovatifolia individuals that faced 180o from north and are a special case in that they were shaded by trees and an opposing rock face and so received little direct sun. Again there were no significant differences between rare and common species (P = 0.417, F = 0.744, df 1, 7), although there were among species (P = 0.001, F = 28.048, df 7, 261: Appendix B). Habitat observations – Dudleya b. blochmaniae grew in shallow soil deposits on rocks or rock outcrops and in patches of rocky soils surrounded by grasses. Co-occurring vegetation included grasses, if sparsely spaced, and Dudleya verity, when growing on rock outcrops, but the two species do not grow intermixed. Selaginella, mosses, and lichens were also found on the rocks. If any shading was present, it was from rock outcrops. Dudleya a. parva occurred on rocks (mostly) and in soil. The rocks often had Selaginella growing on them. Co-occurring vegetation included grasses, if sparsely spaced, and Selaginella. There was no shading other than by rock outcrops. Dudleya vertiyi was found on rock outcrops with lichens, mosses, and Selaginella. Co-occurring plants, growing peripherally, were sometimes other Dudleya especially D. b. blochmaniae. The rock outcrops they were growing on provided the only source of shade. Dudleya c. marsescens occurred on sheer rock faces. Co-occurring vegetation was often mosses. Plants grew mostly in areas shaded by trees and with water nearby. Dudleya c. 29 ovatifolia grew on conglomerate rock formations mostly, but was found on other rock types. There was little co-occurring vegetation other than mosses and lichens. Plants grew in the shade of trees or canyons mostly with water nearby. Dudleya c. agourensis occurred on rock outcrops or in patches of rocky soils surrounded by grasses. Plants were found especially with Selaginella but mosses and lichens also co-occurred. The only shading was provided by the rocks they grew on. Dudleya lanceolata grew in rocky soils or soil deposits on rocks of various sorts. If growing on rock outcrops, they were sometimes found with D. p. pulverulenta and in small numbers peripherally with D. c. ovatifolia and D. c. marsescens. If growing on rocky soils they co-occurred with other vegetation, predominantly Eriogonum fasciculatum and sometimes D. p. pulverulenta. Some shading could be provided by rock outcrops or the co-occurring vegetation. Dudleya caespitosa grew near the Pacific Coast on rocks or rock outcrops with soil deposits. Co-occurring vegetation included Eriogonum fasciculatum and Yucca whipplei. Dudleya p. pulverulenta also occurred in the same area. No shading, other than by rock outcrops, was provided. Dudleya p. pulverulenta occurred mostly on rock outcrops but was found in rocky soils also. When growing on rock outcrops in the less exposed more north facing areas, other Dudleya species were present in the general area. Co-occurring vegetation, when growing on more exposed rocky soils, was most often Opuntia littoralis or O. oricola, often Eriogonum fasciculatum, and seldom Yucca whipplei. This species may have received some shading from rock outcrops or nearby vegetation, but tolerated full exposure and south facing slopes. 30 Habitat dependence: Coastal and inland gardens Environmental data – From April through August, the coastal garden had cooler temperatures and higher humidity than the inland garden (Figure 5). From September through November, the temperature and percent humidity of both gardens were similar. Plant size – In general, plants grew to be larger at the coastal garden than the inland garden (Figure 6). The size changed through the seasons, increasing during the growing season and decreasing when the plants showed signs of physical dormancy. All species were smaller in the inland garden than in the coastal garden, but the disparities in size were greater for the rare species than for the common species. This disparity was greatest in late summer and fall when the weather was hot and dry. Using the species other than D. b. blochmaniae, there was a significant difference in size, log([leaf number × longest leaf length] + 1), between rare and common species in the two gardens over time (date × rarity × site P = 0.038, F = 2.386, df 7, 42), between gardens over time (P < 0.001, F = 15.683, df 7, 42), between rare and common species over time (P = 0.011, F = 3.045, df 7, 42 ), over time (P < 0.001, F = 32.142, df 7, 42), between rare and common species in the two gardens (P = 0.015, F = 11.335, df 1, 6), and between the gardens (P < 0.001, F = 106.834, df 1, 6), but only marginally between the rare and common species (P = 0.074, F = 4.686, df 1, 6: repeated measures ANOVA given in Appendix D). The rosette biomasses were greater for plants grown in the coastal garden than for those grown in the inland garden (Figure 7A). Inland root crown cross-sectional areas were smaller than coastal ones for the rare species (Appendix D). For common species coastal root crown areas were smaller, except in the case of D. p. pulverulenta. Differences in root crown sizes between the gardens were not significantly different for any of the 31 Temperature C (average 12-3 pm) A 45 coastal inland 40 35 30 25 20 15 10 % Humidity (average 12-3 pm) B 100 90 80 70 60 50 40 30 20 10 0 0 2 J 4 6 8 10 12 14 16 week in garden J A S 18 20 22 41 43 45 A O 47 49 51 M 53 55 57 59 61 63 65 week in garden J J A S 67 69 71 73 O N Figure 5. Climate of a coastal and an inland garden over time. A HOBO was placed in each garden and readings were taken every 15 minutes. Data in the graphs were taken from June through November 2006 and from May through December 2007. The readings from 12:00 pm to 3:00 pm were averaged. A. Temperature. B. Percent humidity 32 3 coastal A bloc B parv C veri D marc E ovat F agou G LANC H CAES 2.5 inland 2 1.5 1 0.5 log(linear size +1) 0 3 2.5 2 1.5 1 0.5 0 3 I PULV 2.5 2 1.5 1 0.5 0 J J A S O N DJ F M A M J J A S O N 0 20 40 60 J J A S O N DJ F M A M J J A S O N 80 0 20 40 60 J J A S O N DJ F M A M J J A S O N 80 0 20 40 60 80 week in garden Figure 6. Leaf size over time for coastal and inland garden plants by species. Generally 30 sibling plants from each of the study species were planted in a coastal and an inland garden. Every two months from June 2006 to June 2007 and in November 2007 number of leaves, longest leaf length, and other data were collected. Leaf linear size is the product of longest leaf length and leaf number. The gap between the coastal and inland lines represents the difference in size between the plants growing in the two gardens. Note: D. b. blochmaniae and D. a. parva are summer-deciduous species so leaf data were available only part of the year or for only a few individuals. 33 A 3.5 Sqrt(rosette weights + 0.5, g) coast inland 3 2.5 2 1.5 1 0.5 parv n=30 n=24 P=1 % Reproductive B veri 25 23 .086 marc 26 21 ovat 26 19 agou 30 29 1 LANC CAES PULV 29 30 27 coastal 29 30 25 inland 1 .215 .447 veri 25 23 marc 26 23 ovat agou LANC CAES 26 29 30 30 21 29 30 29 .730 .601 1 100 80 60 40 coastal 20 inland 0 bloc n=18 n=21 parv 30 26 PULV 27 coastal 25 inland Figure 7. Comparisons of Dudleya species grown at coastal and inland gardens. Sibling plants of each species were grown in a coastal and inland garden from June of 2006 to November of 2007. A. Rosettes were collected and dried. Standard error bars are shown. B. Percents of reproductive individuals - number of individuals reproductive spring of 2007 divided by total number of individuals for each species. Fisher’s exact test probabilities are shown. 34 species except D. p. pulverulenta. Reproduction – There were no significant differences between gardens for any of the species (Figure 7B). Nevertheless, the rare species tended to have a greater number of individuals reproduce in the coastal versus the inland garden (except D. b. blochmaniae which had more individuals with inflorescences in the inland garden and D. c. agourensis which had an equal proportion reproduce in both gardens). Common species showed the opposite trend with a greater proportion of individuals producing inflorescences in the inland garden than in the coastal garden, except D. p. pulverulenta, which had no individuals reproduce in either garden. All species had earlier inflorescence production in the coastal garden than in the inland garden (Appendix D Figure D-2). Differential Environmental Responses – The two gardens provide additional evidence for the limitation of the rare species’ ranges due to habitat characteristics. Seven individuals died after surviving transplantation (alive in August 2006 but later dead), and all were at the inland garden (4 ovat, 1 marc, and 2 parv; not including bloc). Considering vegetative size, there was a strong negative correlation when the total biomass dry weights of plants grown in the brick wall were plotted against the maximum difference in size between the plants grown in the coastal and inland gardens (r = -0.962, P < 0.001, n = 8: Figure 8A). Rare species (small biomass dry weights) had greater differences in size than common species (large biomass dry weights). There was also a strong negative correlation when total biomass dry weights of plants grown in a brick wall were plotted against the difference in the proportions of reproductive individuals in the coastal minus inland gardens (r = -0.910, P < 0.002, n = 8: Figure 8B). Rare plants had a greater or equal proportion of reproductive individuals in the coastal garden than the inland one. In 35 contrast, two of the three common species (D. lanceolata and D. caespitosa) had proportionally more reproductive individuals in the inland garden than in the coastal garden. The other common species (D. p. pulverulenta) was not reproductive in either garden. A B 1.3 0.2 parv 1.2 parv 1 % reproductive (Coast - Inland) max size (Costal - Inland) 1.1 0.15 ovat 0.9 marc 0.8 veri 0.7 agou 0.6 LANC 0.5 0.4 0.3 -0.75 PULV CAES r = - 0.962, P < 0.001 -0.25 0.25 0.75 1.25 marc 0.1 0.05 ovat agou PULV 0 veri LANC -0.05 CAES -0.1 r = - 0.910, P < 0.002 -0.15 -0.75 -0.25 0.25 0.75 1.25 log (Dry weight) log (Dry weight) Figure 8. Differential environmental responses. Sibling plants of each species were grown in a coastal and inland garden from June of 2006 to November of 2007. Other siblings were grown in a brick wall garden. These plants were grown between bricks in a thin layer of soil. Roots and rosettes were collected and dried. Dry weights of the roots and rosettes of the brick wall plants were added for a total biomass dry weight. A. The sizes of plants were measured by the log of the product of the longest leaf length and number of leaves + 1. The difference in size was calculated between plants of each species in each garden by subtracting inland garden size from coastal garden size. The maximum difference in size for each species was then found. B. The percent of individuals reproductive spring of 2007 was divided by the total number of plants in each garden for each species. The percent reproductive in the inland garden was subtracted from that of the coastal garden for each species. Note: D. p. pulverulenta is in gray because none of the plants were reproductive in either garden. 36 Discussion The distributions of rare and common species in the Santa Monica Mountains area reflect important life-history tradeoffs. With regard to reproduction and growth rare species are at one end of the tradeoff, allocating energy to reproduction while seemingly foregoing additional growth early in life. Common species are at the other end of the spectrum, allocating energy to growth and foregoing reproduction early in life. The outcome of this difference is that the rare species tended to be smaller in body size, whereas the common species were larger in body size. This tradeoff between growth and the amount of time it takes to reach reproductive maturity, in turn, had several effects. First, the reproductive outputs of the larger, and more common, species were greater (more fruits and seeds per fruit) than for the smaller, and rarer, species. Similar results were found in a study by Lavergne et al. (2004). Secondly, the common species tended to have larger seeds than the rare species which affected seedling survival, because the larger the seed, the better the chance of seedling survival (Morin and Chuine 2006, Mojonnier 1998). A third consequence, as seen in the comparison of coastal and inland gardens, was greater adult survival under stressful conditions for the common species (Appendix D). A fourth consequence of the tradeoff between body size and reproduction is that the large common species had taller inflorescences in the field than the small rare species, which would be expected to increase the distance that seeds disperse (Lloyd et al. 2003). Another factor that can influence a species’ prevalence is specialization to habitats that are found in low numbers or over a small geographic area. The rare species occur in unique terrains (for example, serpentine and volcanic soils and rock outcrops), 37 while the common species have more generalized habitat associations. This is in keeping with Gaston’s (1994) assertion that the sets of environmental conditions necessary for rare species exist less frequently than those for common species. Within habitats, conditions such as co-occurring vegetation, soil types, and climate can influence species’ distributions. The rare species seem to require the absence of other vascular plants and the presence of mosses, lichens, or Selaginella, more specific soil types, aspects that are closer to north facing, and climates that are cooler and more humid compared to the common species. Although beyond the scope of the results reported here, the morphological adaptations of the study species to their environment (having waxy leaves in the form of a basal rosette and low allocation of energy to roots) also appear to be tied to where they can live and, therefore, their distributions (Appendix A). The life-history characteristics of the rare and common species differ in such a way as to affect their geographic ranges and persistence in time (McKinney 1997, Brown et al. 1996). The tradeoff between reproduction and growth has apparently led to the rare species having smaller ranges and shorter expected times to extinction when compared with the common species. Life-history tradeoffs Tradeoffs can occur between pairs of life-history traits (Stearns 1992). In the current study, measured life-history traits were correlated in order to determine which were involved in a tradeoff. Those traits involved in or affected by tradeoffs will be discussed: size and age of first reproduction, number of seeds, size of seeds, and survival. There is a tradeoff between size and age of first reproduction and number of seeds 38 (reproductive output) but not between the size of seeds and survival. Differences between the rare and common species in size at first reproduction affect the other traits contributing to the smaller distributions of the rare species when compared to the common ones. The tradeoff between growth and reproduction is also reflected by relationships between size at maturity and number of offspring. Sterns (1992) reviewed the advantages and disadvantages of early versus delayed reproduction. The advantages of early maturation are a greater probability of surviving to maturity and higher fitness because offspring are produced early in life and those offspring reproduce earlier but at the possible cost of higher mortality rates of offspring than if reproduction were delayed. Theory generally suggests that there is strong selection for early maturity in the absence of tradeoffs. However, reproduction will be delayed if it results in greater lifetime reproductive success and there will be a tradeoff between this increase and lost fitness associated with longer generation times and lower survival to maturity. For example, if delayed maturity permits further growth, and if fecundity increases with size, then a benefit of delayed maturity would be higher eventual fecundity than would be attained if maturity were reached at a smaller size (Stearns 1992). This is the case comparing the Dudleya species in the Santa Monica Mountains area. For example, the small rare D. c. marcescens produced on average 10 fruits and 44 seeds per fruit. In contrast, the huge common D. p. pulverulenta produced on average 167 fruits and 551 seeds per fruit. Lavergne et al. (2004) found that rare species in many genera produce fewer seeds than congeneric common species. Not only does lower seed production negatively affect not 39 rates of population persistence, it limits opportunities for colonization of new habitats, which might also affect species’ distributions (Lavergne et al. 2004). Another tradeoff is between the number of seeds produced and the size of the seeds. This is a tradeoff between seed size and the likelihood of seed survival and seedling establishment. Plants can produce either a small number of large well provisioned seeds that are each likely to germinate and become established, or they can produce many small low quality seeds that, individually may (or may not) be successful, but are so numerous that they have a high combined chance of establishment (Gurevitch et al. 2006). For the Dudleya species evaluated for this study, the rare species tended to have smaller seeds than the common species (except for D. p. pulverulenta), and these seed size differences translated into differences in seed germination and seedling survival. As seed size increased, so did seed germination and seedling survival. For instance, the rare D. c. ovatifolia had seed projected areas averaging 0.254 mm2, 59% germination, and 40% seedling survival compared to the common D. lanceolata with seed projected areas averaging 0.369 mm2, 80% germination, and 75% seedling survival. Other studies have also found positive relationships between seed size and the probability of germination and establishment (Morin and Chuine 2006, Mojonnier 1998 and citations therein). These traits were not found to be involved in a tradeoff, but differences between species could contribute to differences in prevalence. The number of individuals that survive in each species can affect the chances of that species persisting, with high survival increasing the probability of persistence. When rare and common species are compared, the greater reproductive outputs and higher rates of seedling establishment for common species improves the likelihood of their population 40 size being augmented and reduces the chance of genetic diversity eroding. Large patch size can also increase the probability of dispersal of seeds of the common species to other suitable areas (Lavergne et al. 2004, Karron 1997, Gaston 1994), enabling common species to be more widespread. Morin and Chuine (2006) also found that life-history traits can affect species’ ranges. Specifically, those tree species with earlier reproduction, shorter heights, and lower seed mass had smaller ranges. Environmental dependence of species’ prevalences A species will increase its range until it no longer finds the environmental conditions it needs (Levin 2000). Conditions that can impede a species’ spread include the presence of other organisms as well as adverse soil and climate conditions (Kruckenberg and Rabinowitz 1985, Stebbins 1978). If habitats with the conditions necessary for the persistence of a species are uncommon or of limited size, then the distribution of species that occupy them will necessarily be low as well (Gaston 1994, Burgman 1989, Kruckeberg and Rabinowitz 1985). One element that can affect the distribution of species involves competition from (Ackerly 2003) or facilitation with (Kruckeberg and Rabinowitz 1985) various other organisms. There were differences between the rare and common Dudleya species in cooccurring vegetation. Rare species were associated mostly with mosses, lichens, and Selaginella, low stature plants, whereas the common species were often found in close proximity to taller plants. The ability of common species to co-occur with a wider range of plant growth forms may be because of their superior competitive abilities and tolerances associated with being of larger size and having a greater reproductive output 41 compared to the rare species (Gurevitch et al. 2006). A study by Lavergne et al. (2004) of congeneric pairs of narrow endemic and widespread species, also found that narrow endemics occurred in vegetation with a lower canopy and a smaller number of coexisting species, indicating reduced competitive abilities when compared to widespread congeners. The negative effects of competition – reduced growth, reproduction, and survival (Gurevitch et al. 2006) – may preclude rare species from existing in habitats with other vascular plants, thereby limiting the habitats in which they can persist. Riefner et al. (2003) and Riefner and Bowler (1995) have presented evidence that the presence of biological crusts, lichens, mosses, and/or Selaginella may facilitate seedling recruitment of the rare Dudleya species studied herein. Biological crusts fix nitrogen and thereby increase soil fertility, stabilize soil surfaces, improve percolation and soil moisture storage, and reduce erosion. Lichens collect nutrient-rich fog condensation, soil particles and moisture, and may provide protection from herbivory and decrease establishment of other vascular plants. Mosses and Selaginella collect moisture and soil. All of them produce sites for establishment on rocks that would not be present otherwise increasing space for seedlings to recruit and the chances of successful establishment to a larger size. Dependence on the presence of bryophytes in order to persist in a habitat may limit the abundance of those species with this requirement Riefner et al. (2003). Another feature of a habitat that can affect species’ distributions is soil type. Each of the Dudleya listed as threatened occurs in a different habitat with unique geological substrates that apparently limit its distribution. Dudleya a. parva is found on Conejo volcanics, D.verityi exists on volcanic rock outcrops, D. c. marcescens occurs on sheer 42 volcanic rock surfaces, D. c. ovatifolia grows on sedimentary conglomerate rock formations, and D. c. agourensis lives on dissected gravels from the late Pleistocene (USFWS 1999). Dudleya b. blochmaniae, the species of intermediate rarity, occurs in clay or serpentine soils that are rocky (Bartel in Hickman 1993). In contrast, the common species are found in rocky or coastal habitats in which rock/soil types vary. Dudleya lanceolata grows on rocky slopes, D. caespitosa lives coastally, and D. p. pulverulenta grows in a variety of rocky places (Bartel 1993). The rare species are adapted to geological formations that have smaller and more fragmented distributions than the common species. A third environmental factor that may play a role in species’ distributions is climate. All of the study species use the CAM photosynthesis pathway and therefore are limited to regions with regular seasonal rains (Luttge 2004). Also, Dudleya may, to varying degrees, require the cool humid climate available along the coast as evidenced by their mostly coastal distribution (Ulh 2004, 1994). Within the broad category of mediterranean-type climates, there can be levels of humidity and temperature to which the species may react differently. The rare versus common species’ performance in the mesic coastal and xeric inland gardens provide evidence for differences in physiological tolerances to stressful conditions. The rare species had smaller linear leaf sizes, rosette dry weights, a smaller proportion of individuals reproduce, and poorer survival in the inland garden compared to the coastal garden. For example, the linear leaf sizes for D. c. ovatifolia in April (when they would be reproductive) were1.943 in the coastal garden versus 1.289 in the inland garden. Broken down into its components, longest leaf lengths were 5.6 versus 2.55 and leaf numbers 16.69 versus 11 in the coastal versus inland 43 gardens, respectively. For rosette dry weights, D. verityi had the greatest difference in weight between gardens (1.400 g inland and 1.932 g coastal). With regard to reproduction D. a. parva had 81% reproduction in the inland garden and 97% in the coastal garden. Finally, of the few deaths attributable to climatic conditions, more occurred in the inland than coastal garden and all were of rare species. The inability of a species to survive and reproduce in a habitat keeps it from expanding throughout that habitat (Brown et al. 199). The decreased size of the rare Dudleya species in the inland compared to the coastal garden may be attributed not only to their poorer survivorship, it likely had a negative effect on reproduction. Though the decrease in the proportion of individuals that were reproductive in the inland compared to the coastal garden was not significant, the smaller size of plants in the inland garden plants likely led to smaller reproductive outputs compared to plants in the coastal garden. Combined these results suggest the rare species would not persist under dry hot conditions and, therefore, were limited in the habitats available for them to occupy. The common species, on the other hand, had smaller differences in linear leaf sizes and tended to have a greater proportion of individuals reproduce in the inland garden compared to the coastal garden suggesting they were less limited in the habitats they could occupy than the rare species. Conclusion The life-history tradeoff between growth and reproduction in rare and common Dudleya species has a macroevolutionary impact. As a result of this tradeoff the rare species have small body sizes engendering poorer abilities to reproduce, compete with other plants, and tolerate hot dry conditions than the larger bodied common species. 44 These limited abilities appears to be part of their specialization to very specific environmental conditions, where they could be successful by avoiding competition (Lavergne et al. 2004), but at the cost of their abilities to adapt to new habitats when the opportunity for colonization became available (Kunin and Gaston 1997). The common species, with their larger body size, have superior reproductive and competitive abilities and physiological tolerances to hot dry conditions as evidenced by their abilities to coexist with a broader range of vascular plants and perform better in the stressful xeric inland garden than the rare species. A consequence of the common species’ larger body sizes was their ability to become generalists, persisting in a wide variety of habitat conditions, relative to the rare species. Thus, the common species were better able to adapt to new habitats when colonization opportunities arose. The narrow habitat requirements of the rare species resulted in there being few suitable habitats for them to occupy making them more prone to extinction than the generalist common species with broad habitat requirements and geographic ranges (McKinney 1997). The selection processes that lead to the adaptation of the species to their habitats probably occurred at the level of individuals. The differences in life-history traits resulting from the tradeoff and its effect on other traits may have set the stage for later higher-level selection at the species level. The greater extinction probability of the rare species as compared to the common species in conjunction with their current distributions exemplifies how lowerlevel selection processes can shape higher-level ones (Gould 2002). To say, though, that rarity in some Dudleya species is because of life-history tradeoffs and dependence on habitat conditions would be an oversimplification. This study focused on only a few of the time scales, ecological processes, and factors that could have affected the 45 distributions of Dudleya species of the Santa Monica Mountains area. Also, the generalizations put forth in this thesis do not hold true for every Dudleya species. For example, D. p. pulverulenta, a common species, is one of the largest in body size and yet produces the smallest seeds. Exceptions are not limited to common species nor to lifehistory traits. Microhabitat requirements for the rare species are not all the same. Dudleya c. marcescens and D. c. ovatifolia have a habitat requirement for shade that is not found in the other species. In conclusion, there is more than one reason for rarity and when one reason is given for multiple species there will be exceptions. 46 Literature Cited Ackerly, D.D. 2003. Community assembly, niche conservatism, and adaptive evolution in changing environments. International Journal of Plant Sciences 164:S165-S184. Brown, J.H., G.C. Stevens, and D.M.Kaufman. 1996. The geographic range: Size, shape, boundaries, and internal structure. Annual Review of Ecology and Systematics 27:597-623. Burgman, M.A. 1989. The habitat volumes of scarce and ubiquitous plants: A test of the model of environmental control. The American Naturalist 133:228-239. Cavender-Bares, J., K. Kitajima, and F.A. Bazzaz. 2004. Multiple trait associations in relation to habitat differentiation among 17 Floridian oak species. Ecological Monographs 74:635-662. Cleavitt, C.L. 2002. Stress tolerance of rare and common moss species in relation to their occupied environments and asexual dispersal potential. Journal of Ecology 90: 785-795. Dorsey, A. 2007. Dudleya, with special reference to those growing in the Santa Monica Mountains. Pages 93-107 in D.A. Knapp, editor. Flora and ecology of the Santa Monica Mountains: Proceedings of the 32nd annual Southern California Botanists symposium. Southern California Botanists Special Publication No. 4, Fullerton. CA, USA. Elzinga, C.L., D.W. Salzer, and J.W. Willoughby. 1998. Measuring and monitoring plant populations. Bureau of Land Management, Denver, Colorado, USA. Gaston, K. 1994. Rarity. Chapman & Hall, London, UK. Gould, S.J. 2002. The structure of evolutionary theory. Harvard University Press, Cambridge, Massachusetts, USA. Gurevitch, J., S.M. Scheiner, and G.A. Fox. 2006. The ecology of plants. 2nd ed. Sinauer Associates, Inc. Sunderland, Massachusetts, USA. Hickman, J. C. 1993. The Jepson Manual: Higher Plants of California. University of California Press, Berkeley, California, USA. Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54:187–211. Jongejans, E., H. de Kroon, and F. Berendse. 2006. The interplay between shifts in biomass allocation and costs of reproduction in four grassland perennials under simulated successional change. Oecologia 147:369-378. Karron, J.D. 1997. Genetic consequences of different patterns of distribution and abundance. Pages 174-189 in Kunin, W.E. and K.J. Gaston, editors. The biology of rarity. Causes and consequences of rare-common differences. Chapman and Hall, London, UK. Kruckeberg, A.R. and D. Rabinowitz. 1985. Biological aspects of endemism in higher plants. Annual Review of Ecology and Systematics 16:447-479. Kunin, W.E. 1997. Introduction: On the causes and consequences of rare-common differences. Pages 3-11 in Kunin, W.E. and K.J. Gaston, editors. The biology of rarity. Causes and consequences of rare-common differences. Chapman and Hall, London, UK. Kunin, W.E. and K.J. Gaston. 1997. The biology of rarity. Causes and consequences of rare-common differences. Chapman and Hall, London, UK. 47 Lavergne, S., J.D. Thompson, E. Garnier, and M. Debussche. 2004. The biology and ecology of narrow endemic and widespread plants: A comparative study of trait variation in 20 congeneric pairs. Oikos 107:505-518. Levin, D. A. 2000. The origin, expansion, and demise of plant species. Oxford University Press, New York, New York, USA Levin, G. and T. Mulroy. 1985. Floral morphology, nectar production, and breeding systems in Dudleya subgenus Dudleya. Transactions of the San Diego Society of Natural History 21:57-70. Lloyd, K.M., J.B. Wilson, and W.G. Lee. 2003. Correlates of geographic range size in New Zealand Chionochloa (Poaceae) species. Journal of Biogeography 30:17511761. Lüttge, U. 2004. Ecophysiology of crassulacean acid metabolism (CAM). Annals of Botany 93:629-652. McAuley, M. 1996. Wildflowers of the Santa Monica Mountains. Canyon Publishing Company, Canoga Park, California, USA. McKinney, M.L. 1997. How do rare species avoid extinction? A paleontological view. Pages 110-129 in Kunin, W.E. and K.J. Gaston, editors. The biology of rarity. Causes and consequences of rare-common differences. Chapman and Hall, London, UK. Marchant, T.A., R. Alarcon, J.A. Simonsen, and H. Koopowitz. 1998. Population ecology of Dudleya Multicaulis (Crassulaceae); a rare narrow endemic. Madroño 45:215220. Martorell, C. and E. Ezcurra. 2002. Rosette scrub occurrence and fog availability in arid mountains of Mexico. Journal of Vegetation Science 13:651-662. Martre, P., G.B. North, E.G. Bobich, and P.S. Nobel. 2002. Root deployment and shoot growth for two desert species in response to soil rockiness. American Journal of Botany 89:1933-1939. Metcalf, C.J.E., M. Rees, J.M. Alexander, and K. Rose. 2006. Growth-survival trade-offs and allometries in rosette-forming perennials. Functional Ecology 20:217-225. Mojonnier, L. 1998. Natural selection on two seed-size traits in the common morning glory Ipomoea purpurea (Convolvulaceae): Patterns and evolutionary consequences. American Naturalist 152:188-203. Moran, R. 1951. A revision of Dudleya (Crassulaceae). Ph.D. thesis, University of Berkeley, California, USA. Morin, X. and I. Chuine. 2006. Niche breadth, competitive strength and range size of tree species: A trade-off based framework to understand species distribution. Ecology Letters 9:185-195. Mulroy, T.W. 1979. Spectral properties of heavily glaucous and non-glaucous leaves of a succulent rosette-plant. Oecologia (Berlin) 38:349-357. Nakai, K. M. 1983. A New Species and Hybrid of Dudleya (Crassulaceae) From the Santa Monica Mountains, California. Cactus and Succulent Journal (US) 55:196200. __. 1987. Some New and Reconsidered California Dudleya (Crassulaceae). Madroño 34: 334-353. 48 Nobel, P.S and B.R. Zutta. 2007. Carbon dioxide uptake, water relations and drought survival for Dudleya saxosa, the ‘rock live-forever’, growing in small soil volumes. Functional Ecology 21:698-704. Nobel, P.S., E. Quero, and H. Linares. 1989. Root versus shoot biomass: Responses to water, nitrogen, and phosphorus applications for Agave lechuguilla. Botanical Gazette 150:411-416. Pavlik, B.M. 1997. Perspectives, tools, and institutions for conserving rare plants. The Southwestern Naturalist 42:375-383. Pugliese, A and J. Kozlowski. 1990. Optimal patterns of growth and reproduction for perennial plants with persisting or not persisting vegetative parts. Evolutionary Ecology 4:75-89. Rabinowitz, D. 1981. Seven forms of rarity. Pages 205-217 in H. Synge, editor. The biological aspects of rare plant conservation. Wiley, New York, USA. Riefner Jr., R.E. and P.A. Bowler. 1995. Cushion-like fruticose lichens as Dudleya seed traps and nurseries in coastal communities. Madroño 42:81-82. Riefner Jr, R.E., P.A. Bowler, T.W. Mulroy, and C. Wishner. 2003. Lichens on rock and biological crusts enhance recruitment success of rare Dudleya species (Crassulaceae) in Southern California. Crossosoma 29:1-36. Stearns, S.C. 1992. The evolution of life histories. Oxford University Press, New York, New York, USA. Stebbins, G.L. 1978. Why are there so many rare plants in California? I. Environmental factors. Fremontia 5:6-10. __. 1984. Mosaic evolution, mosaic selection and angiosperm phylogeny. The Botanical Journal of the Linnean Society 88:149-164. Ting, I.P. and L. Rayder. 1982. Regulation of C3 to CAM Shifts. Pages193-207 in I.P. Ting and A. Gibson, editors. Crassulacean Acid Metabolism:Proceedings of the fifth symposium in botany, Waverly Press, Baltimore, Massachusetts, USA. Uhl, C.H. 1994. Intergeneric hybrids in the Mexican Crassulaceae. II: Dudleya. Cactus and Succulent Journal (U.S.) 66:74-80. __. 2004. Whence came Dudleya? Cactus and Succulent Journal (U.S.) 76:242-247. Uhl, C.H. and R. Moran. 1953. The cytotaxonomy of Dudleya and Hasseanthus. American Journal of Botany 40:492-502. U.S. Fish and Wildlife Service (USFWS). 1999. Recovery plan for six plants from the mountains surrounding the Los Angeles Basin. Portland, Oregon, USA. Westoby, M., M.R. Leishman, and J.M. Lord. 1995. On misinterpreting the ‘phylogenetic correction’. Journal of Ecology 83:531-534. Whittaker, R.J., K.J. Willis, and R. Field. 2001. Scale and species richness: towards a general, hierarchical theory of species diversity. Journal of Biogeography 28:453470. Willis, K.J. and R.J.Whittaker. 2002. Species diversity-scale matters. Science 295:12451248. 49 Appendix A: Interpretative background on Dudleya The thesis is best understood if one has contemplated the natural history of Dudleya in general and more particularly the nine species that were studied. Morphology, physiology, phenology, and details of the geographic range come together in a speculative interpretation of adaptations, and the interpretation suggests hypotheses that could, in principle, be tested. Much of this interpretation, however, was done after, not before, the project was set up; thus, the following discussion may be useful in setting up additional studies, as well as pondering the implications of the current one. Dudleya is a genus of perennial succulents in the family Crassulaceae, the Stonecrop Family (Bartel in Hickman 1993), that uses the crassulacean acid metabolism (CAM) photosynthetic pathway (reviewed by Dorsey 2007). Dudleya grow on rock outcrops in cracks, on saxicolous mosses or lichens, with Selaginella, or in shallow soil deposits on rocks and nutrient-poor mineral soils often surrounded by grasses (Riefner et al. 2003, Riefner and Bowler 1995, pers. obs.). The growing season of Dudleya starts with winter rains and extends through the spring. Inflorescences can start to form as early as December. Some species bloom as early as March, others as late as July. They stay in bloom for two to three months (McAuley 1996). Over the summer, Dudleya become physically dormant: inflorescences dry out and may break off, the leaves of the rosette close, and the outermost leaves become dry but in most species remain intact. Dudleya a. parva and D. b. blochmaniae have leaves that are summer deciduous. The fruits ripen over the summer. Seeds are dispersed starting in late summer when the follicles become dry and open. Most seeds fall close to the parent plant (Marchant et al. 1998), but some are dispersed by the wind. Seed germination is activated by rain (or persistent moisture). 50 After consistent rains, the rosettes open revealing the living inner leaves, and growth resumes. For the species with summer deciduous leaves, new growth is from a corm (D. b. blochmaniae) or caudex (D. a. parva). The above and below ground morphologies of Dudleya species enable them to live in dry rocky habitats. A basal rosette allows for the maximization of leaf area for photosynthesis. Because leaves are basal, little energy needs to be put into developing a supporting stem and its low stature protects the plant from wind and other mechanical damage (Stebbins 1984). The rosette habit of Dudleya species may allow them to take advantage of small amounts of available moisture in the form of fog. Water from the fog collects on the leaves and is channeled to the roots by flowing down the stem. This ability is even greater when plants have a waxy epidermis, such as agaves, that can channel as little as 1 mm of rain (Martorell and Ezcurra 2002). Dudleya are likely to be able to do this as well, as indicated by their ability to grow roots in response to high levels of humidity (Riefner et al. 2003). Of the study species, all but three (D. b. blochmaniae, D. c. marcescens, and D. c. ovatifolia) have glaucous leaves (Bartel 1993, Nakai 1987, Moran 1951). Having wax on the leaves may not be important for D. b. blochmaniae and D. c. marcescens because their leaves are not living during the hot dry summer (Moran 1951, pers. obs.) and for D. c. ovatifolia because they occur where it is shaded (USFWS 1999; Figure 4). This ability to take advantage of small amounts of condensation may help Dudleya endure the long summer drought, but it may limit the species that need this source of water to coastal or shaded inland habitats (where D. c. marcescens and D. c. ovatifolia occur; USFWS 1999) as evidenced by the rare species’ reduced size, 51 reproduction, and survival when in a drier hotter inland environment than when in a cooler moister coastal one. An additional adaptive feature related to water stress is the ability of rosetteforming species to let their outer leaves dry and transfer the water and nutrients to the inner leaves (Luttge 2004, Mulroy 1979). Being able to retain leaves year round allows plants to maximize rainfall by being able to store water and resume active photosynthesis quickly as opposed to having to grow new leaves first (Ting and Rayder 1982, Mulroy 1979). This is especially important when timing and amount of rainfall varies from year to year. There can be other advantages to only retaining below-ground structures. Those species with leaves living only part of the year (D. b. blochmaniae, D. a. parva, and D. c. marcescens) are able to escape from the problems of aging because new parts are grown each year, and from harsh conditions, so energy does not have to be put into adaptations to those conditions which might decrease physiological performance (Pugliese and Kozlowski 1990). The environment in which a plant occurs can also influence its below ground morphology. The study species have thick branched main roots and fine lateral roots (pers. obs.), a morphology ideal for dry rocky habitats. The thick part of the roots can be used for storage to help the plant withstand the long summer drought. The branching can facilitate growth around rocks or in different directions in order to cover more completely the surrounding available area than if the roots did not branch. The fine roots allow for growth in narrow spaces between the rocks or in cracks. Energy allocation to roots in the study species (Figure C-3) may be low relative to many other kind of plants. Root growth in Dudleya saxosa (Nobel and Zutta 2007) and desert succulents (Nobel et al.1989) has 52 been found to be optimized to only the volume of soil that would be necessary to absorb enough water to fill the storage capacity of the plant. Root growth may be limited by the rockiness of the soil as well. In a study of Agave deserti, a slow growing CAM succulent, Martre et al. (2002) showed that as the rockiness of the soil increased the volume of roots decreased. 53 Appendix B: Nested ANOVAs and correlations In the Results, significance tests are presented as to whether or not there was variation among species for a given feature, and whether or not there was an added difference between the three common species and the five (to six) rare species. Table B-1 presents the nested ANOVAs in more detail. Later in the Results, some correlations were presented between one trait and another, with n equal to 8 or 9 species (Figure 3). Table B-2 presents the means of the traits by species. Table B-3 presents the correlations among traits in a more exhaustive manner than in Figure 3. Table B-1. Nested ANOVAs. For various traits, individuals at 2-3 sites were pooled for each “species,” species-within-rarity was tested over the error, and rarity was tested over species-within-rarity (the latter being a weak test because there were only 3 common species and 5 rare species). Trait error MS (df) 20519 (261) 633 (261) 1316 (261) 0.036 (203) 0.359 (203) 0.020 (261) 0.616 (261) 0.004 (254) 0.00096 (87) 0.096 (514) 0.055 (514) Average distance to nearest neighbors, cm Microsite slope, degrees Degrees from due north Log(Leaf length, cm) Sqrt(Leaf number) Log(Inflorescence height, cm) Cubed root(Number of fruits) (Seeds per fruit) - 1/6 Seed projected surface area, mm Arcsine(%Germination) % Survival NS P>0.1; †P<0.1; *P<0.05; **P<0.01; ***P<0.001 54 spp ⊂ rarity MS (7 df) 767358*** rarity MS (1 df) 889790 NS 12772** 823.869 NS 36918** 27449 NS 1.056** 18.162** 54.238*** 157.853 NS 0.641*** 30.023*** 13.196** 176.843** 0.107*** 0.779* 0.01636** 0.20778** 1.044* 1.627 NS 1.167** 1.446 NS Table B-2. Means of selected traits for each species. See Figures 3 and 4 with associated text for explanation of traits. # fruits (1/3) seeds per fruit –(1.6) seed projected area arcsine germination % seedling survival log inflorescence height nearest neighbors distance degrees from north slope log (root+shoot) in brick wall log(root/shoot) reproduction out of 3 springs log reproductive output in the wild bloc 2.429 -0.633 0.298 0.851 60.610 1.009 7.598 98.000 23.733 - parv 2.743 -0.562 0.330 0.932 70.520 1.110 10.775 20.000 24.000 -0.499 -0.247 2.898 2.936 veri 3.222 -0.574 0.247 0.668 47.070 1.060 6.255 37.333 39.100 0.321 -0.474 2.825 3.142 marc 2.093 -0.570 0.246 0.559 29.720 0.987 6.630 13.000 69.600 0.056 -0.444 2.792 2.615 ovat 2.539 -0.523 0.254 0.647 39.980 1.022 25.786 20.333 66.500 0.086 -0.597 2.534 2.998 agor 4.037 -0.522 0.288 0.715 55.770 1.390 68.256 100.000 57.167 0.337 -0.409 2.454 3.659 LANC 4.253 -0.455 0.369 0.966 75.290 1.813 168.001 44.667 30.833 0.579 -0.287 1.828 4.033 CAES 4.200 -0.538 0.427 0.855 66.740 1.663 50.815 22.167 72.000 1.046 -0.415 0.621 3.679 PULV 5.230 -0.351 0.185 0.720 43.550 1.934 501.399 13.333 48.333 0.868 -0.895 0.026 4.964 Table B-3. Correlations between traits, calculated across 9 species, except those involving the last four variables which were calculated across 8 species (without D. b. blochmaniae). –(1.6) seeds per fruit seed projected area arcsine germination % seedling survival log inflorescence height nearest neighbors distance degrees from north slope log (root+shoot) in brick wall log(root/shoot) reproduction out of 3 springs log reproductive output in the wild # fuits (1/3) 0.820 seeds per fruit –(1.6) 0.073 -0.271 seed projected area 0.275 0.036 0.713 arcsine germination 0.293 -0.041 0.793 0.968 % seedling survival 0.946 0.835 0.210 0.394 0.368 log inflorescence height 0.793 0.925 -0.390 0.040 -0.078 0.790 nearest neighbors distance -0.039 -0.362 0.111 0.190 0.297 -0.137 -0.240 degrees from north 0.047 0.125 -0.045 -0.633 -0.558 0.048 -0.027 -0.339 slope 0.787 0.551 0.182 0.078 0.096 0.783 0.543 0.042 0.354 log(r+s) -0.404 -0.664 0.714 0.512 0.610 -0.301 -0.731 0.296 -0.327 -0.437 -0.834 -0.767 -0.083 -0.224 -0.140 -0.869 -0.787 0.231 -0.202 -0.846 0.966 0.924 -0.092 0.304 0.238 0.934 0.914 0.105 -0.152 0.730 55 log(r/s) 0.584 -0.551 repro3 -0.852 Appendix C: The brick wall garden The brick wall garden consisted of plants grown between clay bricks with a thin layer of soil. A 10 × 4’ wooden structure with 3’ perpendicular sides made of 1/2” plywood and 2 × 4s set at an angle was used as a support structure for the bricks to rest against. The wooden backing was supported by 2 × 4s the ends of which were cut at an angle so they were flush with the plywood and the ground. To increase the support from the ground, the area where the 2 × 4s rested was dug up and a brick was put under each 2 × 4. Pieces of rebar were pounded into the ground behind the plywood and between the 2 × 4s to give added support to the plywood. Concrete bricks were placed along the back side of the plywood at ground level to keep the base in place. Four by fours were placed in front of the structure. One end of the bottom row of bricks was placed on the 4 × 4 and the other on the ground so the back side of the brick was flush against the wooden support. Bricks were laid lengthwise, perpendicular to the support structure. An 8” × 4’ piece of hardboard was vertically placed between each brick column. As each brick was laid a 1/4 inch thick layer of one part potting mix used with the potted plants and one part potting soil was placed on it. A randomly chosen individual of a pre-chosen species was placed in the soil, a drip irrigation hose was placed over the plant, secured with wire to the hardboard, and another brick was put on top. This was repeated until one row of plants was in place. The next row of bricks was constructed the same way. The species were arranged such that from column to column the order of the species from top to bottom was shifted by one. This allowed for variation in the location of a given species’ individuals both left to right across rows and up and down across columns. There were 9 rows and 28 columns total. The garden was constructed from 7 – 17 July 2006. The 56 brick-wall garden was north-facing on the California State University, Northridge campus (34.239o N, 118.531o W). The campus is in the hot dry valley well inland of sites that elsewhere are referred to as “inland.” Figure C-1. Photo of brick wall garden shortly after planting. Plants were watered every 4 to 5 days for a total of 4 times after transplantation. Watering was resumed 11 November 2006 to 12 July 2007, weekly. From July 2006 to September 2007, every two months, data on the number of leaves, longest leaf length, and number of inflorescences of each individual were logged. Buds and flowers were cut from the inflorescences as necessary to limit the amount of energy allocated to reproduction (Jongejans et al. 2006) and control for energy allocation differences between reproductive and non-reproductive individuals. Plants grown in the brick-wall garden were intended to approximate more natural growing conditions than plants grown in a pot. In all of the species, sizes (as measured by longest leaf length multiplied by leaf number) decreased in late summer and fall, when plants started to show signs of physical dormancy, and increased in winter and spring, during the growing season (Figure C-2). The common species tended to be larger in size than the rare ones. The exceptions being D.verityi (rare) and D. lanceolata (common) 57 Between subjects Species Error df 7 204 MS 29.535 0.629 F 46.968*** Within subjects Date Date × Species Error ***P<0.001. 7 49 1428 11.308 0.374 0.04253 265.851*** 8.786*** 3 2.5 CAES log(linear size + 1) PULV 2 veri LANC 1.5 agou ovat marc 1 parv bloc 0.5 0 0 10 20 30 40 50 60 Week in brick wall Figure C-2. Changes in size of plants over time. Size measured as the log(longest leaf length × number of leaves + 1). Week 0 was 17 July 2006. Inset repeated measures ANOVA done without bloc. both of which were very close in size, with one being larger than the other or visa versa over time. Dudleya lanceolata tended to have longer leaves than D. verityi, but D. verityi, a clonal species that forms multiple rosettes, tended to have more leaves than D. lanceolata. From September 2007 to November 2007, plants were removed from the brick wall garden. At the time of removal the wire attaching the irrigation drip hose was removed, the brick above the plant was removed, the drip hose was moved out of the 58 way, the brick the plant was resting on was lifted taking care not to break any roots, pressure was applied to the soil mass from the sides, and the rosette was lifted up. Once the soil was loose, the plant and intact roots were removed. The rosette was cut from the roots below the leaves (at the root crown). All of the dried inflorescences were removed from the rosette, and an ink imprint was taken of the cross-section. Rosettes were placed in labeled paper bags. The roots were held up near the cut to allow loose dirt to fall away and were then placed in a bucket two-thirds full of water. As much soil as possible was removed from the roots by flexing the roots slightly, separating roots from each other, and gently rubbing the soil off the roots. The roots were then placed in another bucket that was half filled with water. The roots were gently swirled in the water and the above process was repeated to remove as much remaining soil as possible. The very finest roots were lost in this process as they disintegrated when the soil was removed. Remaining larger pieces, such as perlite and peat were picked or pushed out of the roots, using a dissection-kit needle probe. The roots were then laid on a brick and allowed to dry before being placed in a labeled paper bag. A strainer was used to remove as much soil from the first bucket as possible after each plant. When the water became dirty it was replaced. At this time the second bucket became the first and visa versa. After a whole column was completed, the wires were removed from the drip irrigation hose and the now exposed hardboard. The hardboard was removed to gain access to the next column of bricks. These steps were repeated until all of the plants had been removed. Bagged roots and rosettes were placed in plastic full-sized flats. The flats were put in a drier set at 37.8o C until weights stopped decreasing, which took months of drying. 59 14 Dry weight (g) 12 Rosette Root 10 8 6 4 2 0 parv ovat marc agou veri LANC PULV CAES Figure C-3. Root and shoot dry weights. Plants of each species were grown in a thin layer of soil between vertically stacked bricks from July 2006 until November 2007. When the garden was dismantled roots were separated from rosettes and cleaned. Roots and rosettes were dried until weights started to increase. Sample sizes n = 21 to 29. 60 Appendix D: More information on coastal versus inland gardens In the Results comparing coastal and inland gardens, data were presented on how leaf size changed through the seasons, final rosette weight, and the percentage of plants that bolted. Corresponding ANOVA tables are here presented along with data on root crown cross-sectional surface areas, which is another measure of size. When plants were harvested, they were cut at the base of the rosette. An ink print was taken of the crosssection. From the ink prints of the root crowns, the longest diameter and the one perpendicular to it were measured using a ruler. The cross-sectional area of the roots were calculated using the formula for the surface area of an ellipse (SA = πr1r2). In the case of multiple rosettes or root branching, measurements were taken of each crosssectional area and areas were added. The largest cross-sectional surface area was measured for the common species, D. caespitosa, and the smallest for the rare species, D. a. parva. Dudleya caespitosa is one of the largest study species and has the ability to produce multiple rosettes. Dudleya a. parva, is one of the smallest study species with living leaves present only part of the year. Table D-1. Repeated measures ANOVA of log([maximum leaf length × number of leaves] + 1) between individuals rarity garden rarity × garden spp ⊂ rarity garden × spp ⊂ rarity error df 1 1 1 6 6 426 MS 158.6166 159.3713 16.90972 33.84762 1.49176 0.56536 F 4.686 106.834 11.335 no test no test P 0.074 <0.001 0.015 within individuals date date × rarity date × spp ⊂ rarity date × garden date × rarity × garden date × garden × spp ⊂ rarity error 7 7 42 7 7 42 2982 13.904 1.31737 0.43258 3.57332 0.54368 0.22784 0.03275 3.21E+01 3.045 no test 15.683 2.386 no test <0.001 0.011 61 <0.001 0.038 Table D-2. Split-plot ANOVA of square root(rosette weight). Effects of coastal versus inland garden, rarity class, and species within rarity class. Total n = 423. R2 = 0.692. df 1 6 1 1 6 407 rarity species ⊂ rarity garden garden × rarity t×s⊂r error MS 130.76297 6.58217 15.93847 1.30727 0.63795 0.21054 F 19.86624 31.26378 24.98389 2.04917 3.03011 P 0.004 <0.001 0.002 0.202 0.007 Table D-3. Split-plot ANOVA of log(root crown cross-sectional area + 1). Effects of coastal versus inland garden, rarity class, and species within rarity class. Interaction terms dropped sequentially when P > 0.25. Total n = 423. R2 = 0.592. df 1 6 1 414 rarity spp ⊂ rarity garden error MS 10.82183 3.30078 0.39659 0.05306 F 3.279 62.205 7.474 P 0.120 <0.001 0.007 0.45 coastal log(cross sectional area + 1) 0.4 inland 0.35 0.3 0.25 0.2 0.15 0.1 0.05 parv veri marc ovat agou LANC CAES PULV Figure D-1. Coastal and inland root-crown size by species. Sibling plants were grown in a coastal and an inland garden from June 2006 to late November (inland) or mid-December (coastal). When the plants were removed the rosettes were collected and ink imprints of the root-crown were taken. Standard error bars are shown. For sample sizes, see Figure 7A. 62 Proportion bolting 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 veri parv bloc agou marc ovat LANC CAES PULV coastal inland 0 20 40 60 80 0 20 40 60 80 0 20 week in garden Figure D-2. Bolting through the seasons in the coastal and inland gardens. See Figure 6 for timeline. 63 40 60 80 Appendix E: Watering treatments When the thesis project was being set up, there was a large portion of the research that was devoted to trying to find physiological differences between the species and how they might depend on air temperature and humidity and soil moisture. In order to determine if stressful habitat conditions could affect the physiological performance of the study species, three sets of plants were subjected to three watering treatments in an experimental garden in which they were exposed to hot dry summers: a group with no water added, a group that was misted, and a group that was watered weekly. As transplantation of seedlings was being done, two 1 × 1” travertine tiles were placed to either side of the seedling in these three treatments. The tiles were intended to assure that there would be a space below the plant’s rosette for an infrared-gas-analyzer chamber to be cuffed on, so photosynthesis readings could be taken. However, the plants quickly grew too big for the chamber to be used, so gas exchange was never measured. Photosynthetic activity was measured, more ambiguously, by light-adapted fluorescence, which does not allow one to interpret whether plants are doing CAM photosynthesis or regular C3 photosynthesis. There were more serious problems, though, with the wateringtreatments experiment, so limitations in the photosynthesis related variables turned out to be a moot point. The three sets of potted plants were kept on the California State University, Northridge campus, where summers are as hot as the hottest areas in the Santa Monica Mountains. The water group was watered weekly unless it rained, in which case they received no additional water. The misted group was intended to be cooled by misting without having water added to the soil. For the first 20 days, misting was controlled by a 64 timer capable of being set “on” or “off” only, so plants were misted constantly for a set period of time; thereafter, the mist group was misted every 15 minutes for three or five seconds. Misting was done during the 4 warmest hours of the day, from 12:00 noon to 4:00 p.m. May through October and 11:00 a.m. to 3:00 p.m. December through February. The duration of the misting was longer June through November (5 seconds) and was shorter December through May (3 seconds). The group was misted even if it rained. The last group, the no-water group, received natural precipitation only, which meant basically no water from June to November. The treatments were started in early June. Each treatment group was kept in a separate bay. The three bays were constructed under a large shade-cloth awning. The walls between the bays consisted of a 2 × 4 frame over which additional shade cloth was secured. The walls were 6 feet high and 7 feet long. The mister system was constructed from 2 × 2s in the form of a cube frame raised to a height of 5 feet above the pots. The mister hose and nozzles were secured along the frame. The frame was centered over the plants. The potted plants were kept in flats set on wooden pallets. The flats in each treatment group were rotated weekly, and each group was moved to a different bay monthly. It should be noted, however, that plants in a treatment were kept next to one another, so there was spatial pseudoreplication in the application of the treatments, although not in the assigning of seedlings to treatments. One HOBO meter was placed in each treatment bay starting on 8 June 2006 to log temperature, dew point, and humidity. The HOBO for each treatment group was moved to follow the treatment group from bay to bay as the flats for their treatments were rotated each month. 65 Every three months, the sizes of plants were determined by placing white pieces of paper underneath the leaves, placing a ruler next to the plant, taking a photograph of the plant with a camera on a tripod centered over the plant, and using NIH Image software to find the projected area of the plant. Any dry inflorescences, buds, or flowers were removed from the plant before taking the pictures so only photosynthetic plant parts were included. Buds and flowers were removed in order to limit allocation of energy to reproduction (Jongejans et al. 2006) and to control for energy allocation differences between reproductive and non-reproductive individuals. Pictures were taken every three months from May 2006 to July 2007. At the end of the experiment the projected surface areas of all species in the three watering treatments in April 2007 were compared using an ANOVA. Tukey pairwise comparisons were done to compare the projected surface areas of each species grown in the three watering treatments. Photosynthetic activity was determined by using a PAM fluorometer (PAM-2000, Walz, Effeltrich, Germany; Cleavitt 2002). In order to minimize the effects of temperature throughout the seasons, flats of pots were placed in a greenhouse for at least 2 1/2 hours before fluorometer readings were taken. On a few occasions, plants were left in the greenhouse overnight. The readings were taken in the greenhouse. Temperature and light level readings for the greenhouse were recorded hourly until two hours before sunset when levels were recorded every 1/2 hour. Six readings were taken on each plant: spots on the upper and middle part of an outer leaf, the upper and middle part of a middle leaf, and the upper and middle part of an inner leaf. Data analyses were done using the 2nd from the maximum reading in order to show the highest potential yield while eliminating possible outliers. Readings were of light-adapted effective quantum yield. 66 Readings were taken over three consecutive days, one treatment group at a time, every two weeks, from August 2006 to January 2007, then monthly for the next two months. Presence of inflorescences was also recorded. Unfortunately, the three watering treatments experienced fairly similar temperature and humidity regimes (Figure E-1); misting did not lower the temperature nor did it raise the humidity in a way comparable to the difference between the coastal and inland gardens. Nevertheless, comparing the temperature (Figure E-1A) in the hot season, the no-water treatment temperature was hotter than the water and mist treatments, which were often similar. In late October and parts of November, the mist treatment had the highest temperatures followed by water and no-water, which were mostly similar. From the end of December through February, temperatures in all three treatments were mostly the same. Comparing the humidity (Figure E-1B) from August through the end of November, the water treatment had the lowest percent humidity. During this time period the mist group had higher percent humidity than the no-water group with differences being the greatest in August and early September. In December the percent humidity for all treatments was essentially the same. January through March the mist and no-water groups had similar percent humidity with the mist treatment often being slightly higher. Periods of low humidity, as measured in the no-water group daily averages from 12:00 – 3:00 p.m., were from August through mid-September (below 30%), late October (below 30%), early to mid-November (20-30%), early December (below 20%), January (below 20%), and early March (below 20-30%). 67 A mist 50 Temperature C (averaged noon-3 pm) water no water 40 30 20 10 0 13 17 8 A 13 S 17 O 22 26 31 35 40 100 % Humidity (averaged noon-3 pm) B 8 80 60 40 20 0 22 26 31 35 40 N D J F M Week of Treatment Figure E-1. The average temperature (A) and percent humidity (B) of three watering treatment areas from 12:00 pm to 3:00 pm over time. HOBOs were placed in each of the three water treatment areas (mist, water, and no water) and readings were taken every 15 minutes from June 2006 to July of 2007. Graphs show data from August 2006 to March 2007 when photosynthesis readings were being taken. Sibling plants from each species were randomly placed into each of the three treatments. Mist plants received misting for 3-5 seconds every 15 minutes from 11:00 am to 3:00 pm when temperatures were cool and from 12:00 pm to 4:00 pm when temperatures were hot. Water plants received water once a week. No-water plants received natural precipitation only. Figure B note: the humidity sensor in the water treatment HOBO stopped working properly at the end of December. 68 C 45 Temperature C 35 temp max mist 25 temp min mist temp max water 15 temp min water temp max no water 5 temp min no water -5 8 S O N D J F 13 17 22 26 31 35 M 40 Week of Treatment Figure E-1 continued. C. Temperature maximums and minimums for all water treatment groups with PAM reading dates. Sibling plants were randomly placed in one of three watering treatments: water, mist, and no-water from June 2006 through June of 2008. The water group received water or natural precipitation once a week. The mist group was misted for the four warmest hours of the day and received natural precipitation. The no-water group received natural precipitation only. HOBOs were placed in each of the treatment areas from June 2006 through July of 2007 to measure temperature and percent humidity. The gray bars show the dates when quantum effective yield readings were taken using a PAM fluorometer. Although the treatments were similar, temperature strongly changed through the seasons (Figure E-1C), so season × species interactions may be interesting. The highest maximum daily temperatures were from August through mid-September ranging from 30 – 45 oC. There were also high temperatures (low 30s) for a few days in late October, early and mid-November, and early March. In late December temperatures dropped to 0 and 1 oC for two consecutive evenings. Temperatures dropped again in mid-January to -4 C and three nights later to 1 oC. Even as late as the end of February there were three nights where temperatures were 2 oC. 69 Rare species had plants with smaller projected areas than common species (Figure E-2). Also, rare species’ projected areas were largest in the mist treatment with water and no-water plants having similar areas. In contrast, the common D. caespitosa and D. p. pulverulenta tended to have larger projected areas when watered (for D. lanceolata, mist and water projected areas were similar) and the common species had the smallest projected areas when not watered. This difference in effect of the water treatments on the species was significant (P < 0.001, F = 13.474, df 16, 712). A two-way ANOVA was done to test for effects of treatment and species on plant size. The treatment × species interaction was significant (P < 0.001). The analysis was further broken down by doing separate ANOVAs for each species comparing the effect of treatments in pairwise comparisons (mist – water; water – no water; mist – no water). For D. p. pulverulenta all combinations of treatments were significantly different from each other. For D. verityi the only treatments that were not significantly different from each other were between water and no water. For D. caespitosa the only treatment combination that was not significantly different was mist and no-water. For D. c. marcescens and D. c. agourensis, only the mist and no-water treatments were significantly different from each other. For D. a. parva the only significant differences were for the mist and water treatments. Finally, for D. b. blochmaniae, D. c. ovatifolia, and D. lanceolata none of the treatments were significantly different. 70 mist – water .065 water – no water .564 mist – no water .346 14 sqrt projected surface area .001 .428 .028 .348 .478 .035 <.001 .700 <.001 .080 .745 .012 <.001 <.001 <.001 .706 .429 .104 .001 <.001 .084 no water mist 12 a .171 .098 .001 water 10 8 6 4 df MS Treatmt 2 77.561 Species 8 524.139 T*S 16 22.649*** Error 712 1.681 2 0 bloc marc parv ovat veri agou n = 87 84 90 84 75 90 PULV LANC CAES 78 87 87 Figure E-2. Plant size affected by treatment and species. Projected surface areas of Dudleya plants sown from wild collected seeds grown from summer 2006 to spring 2008 in three different watering treatments: no-water plants received only natural precipitation, mist plants received misting during the four warmest hours of the day in addition to natural precipitation, water plants were watered weekly. Data are from April 2007 and were square root transformed. Tukey probabilities are shown above. Inset: two-way ANOVA (***P < 0.001). All of the species showed a decline in yield values for the no-water group from weeks 2 to 6 (late August to late September: Figure E-3). Dudleya a. parva showed the greatest decline (this is the time of year when leaves would not be present in wild populations of this species). In the mist group only D. a. parva showed a lesser but similar decline in yield readings, and D. verityi and D. p. pulverulenta showed a small decline from week 4 to 6. The water group yield readings showed no overall decline from weeks 2 to 6, though, D. caespitosa and D. p. pulverulenta did show very slight decreases from week 2 to 4, but their yield readings increased from week 4 to 6. From weeks 10 to 12 (late 71 October to early November) all of the species and treatments except for the mist groups of D. c. ovatifolia and D. lanceolata showed a decline in yield readings. All treatment groups for all of the species showed a decrease in yield readings from weeks 14 to 16 (mid-November to early December) and in the mist group also from weeks 16 to 18 (early to mid-December). From weeks 22 to 30 (January through March) all of the species in all of the groups showed a steep downward trend in yield values. Otherwise the species responded differently to the different treatments and over the yield reading period. The watering treatments affected the yield readings differently over time (interaction P < 0.001, F = 24.295, df 26, 182). Yield readings differed over time (P < 0.001, F = 50.278, df 13, 91) and by treatment (P < 0.001, F = 76.163, df 2, 14). There were no clear differences in the responses of the rare species compared to the common species to any of the watering treatments. In interpreting this experiment, it would be important to know whether misting provided water to the plants’ roots or only cooled and humidified the air. A side experiment was done to study this. Four-inch plastic pots were filled with the same potting mixture the plants were potted in. They were weighed, placed in a drier at 100o F for two days, and then weighed again. Eleven pots were placed randomly in the mist group flats, and 11 in the no-water group flats. Pots were re-weighed in the evening 3, 8, and 10 days later. The pots were then placed in the drier to dry for 4 days, were weighed, and allowed to dry 3 more days when a final weight was taken. The weights of the soilfilled pots were normalized by dividing weights measured later by initial weights. t-tests were used to determine if mist group soil was becoming wet by comparing the 72 Effective Quantum Yield 100 90 80 70 60 50 40 mist water A parv no water B veri 30 Effective Quantum Yield 100 90 80 70 60 50 40 C marc D ovat E agou F LANC G CAES H PULV 30 Effective Quantum Yield 100 90 80 70 60 50 40 30 Effective Quantum Yield 100 90 80 70 60 50 40 S A O N D J F M S A O N D J F M 30 0 2 4 6 8 10 12 14 16 18 20 22 26 30 week of measurement 73 0 2 4 6 8 10 12 14 16 18 20 22 26 30 week of measurement Figure E-3. Effective quantum yield readings over time by species in three watering treatments. Siblings of each species were randomly placed in three different watering treatments. Mist plants received misting during the warmest four hours of the day and natural precipitation. Water plants received water weekly. No-water plants received only natural precipitation. Effective quantum yield readings were taken after plants had been placed in a greenhouse for at least 2 ½ hours. Most or all of the readings for each treatment were taken on the same day with different treatments being measured on subsequent days over a period of two to three days. Readings were taken from August 2006 to March 2007. Each reading consisted of six measurements on different parts of the rosette. The graphs show the second highest reading averaged for each species and the week readings were taken. 74 proportional pot weights between the mist and no-water group (Figure E-4). There were no significant differences in the proportional weights of pots in the two water treatments, although inexplicably the misting treatment did seem to be more variable than the nowater treatment. 0.998 misted not watered mass / starting mass 0.996 0.994 0.992 0.99 0.988 0.986 am pm pm pm pm day 1 day 1 day 3 day 8 day 10 dried Figure E-4. Proportions of soil filled pot weights placed in the mist and no-water treatment bays. Pots (n = 11 for each treatment) were filled with soil, placed in a drier to dry, weighed, and then placed randomly in flats in the treatment areas. Weights were recorded in the morning and evening of day 1, the evenings of day 3, 8, and 10, and after the pots were placed in a drier and had completely dried. Proportions were calculated by dividing the weights over time by the starting weights. Standard error bars are shown. 75 Appendix F: Differences between confusable taxa Table 1 in the thesis describes some characteristics that differentiate the species. Petal color is often helpful for species identification. Colors range from white to shades of yellow to red, some with distinguishing stripes of a color different from the background color or glaucous keels. Other distinguishing features are the time of year when leaves are present, the number of rosettes, and formation of protective coatings on the leaves (Nakai 1987, Nakai 1983, Moran 1951, pers. obs.). Dudleya b. blochmaniae has the habit of dying back during the summer to underground corms, and thereby belongs to the subgenus Hasseanthus, whereas the other eight species belong to the subgenus Dudleya. Among those eight, the most obvious distinction is between D. p. pulverulenta, with its large body size and chalky covering, versus the other seven. Among those seven, there are two cases where each species’ distinctiveness has been questioned. (1) Dudleya c. agourensis has been variously considered to be the same as D. c. ovatifolia (Uhl and Moran 1953), a form of D. c. ovatifolia (Nakai 1983), or a subspecies (Nakai 1987). Dudleya c. ovatifolia as narrowly defined is distinguished from D. c. agourensis by its leaf undersides being maroon in color, having ovate to elliptic shaped leaves, an unbranched caudex, petal apices that spread 90o or more, and slightly longer pedicels (Nakai 1987). (2) Dudleya lanceolata and D. caespitosa appear very similar and belong to a polyploidy complex that might intergrade. Dudleya lanceolata is distinguished from D. caespitosa by having a simple or little-branched caudex rather than a branching one that can form over 100 rosettes, a different leaf shape, having a gently curved rather than an 76 obpyramidal shaped inflorescences (Moran 1951), and 34 chromosomes rather than 51 (Uhl and Moran 1953). In the course of doing the thesis, data were gathered for many characters that might differ between these confusable species. For continuous characters, like various aspects of size, two-sample tests were done in the spirit of planned comparisons using only the data on D. c. ovatifolia versus D. c. agourensis, and then only the data on D. lanceolata versus D. caespitosa. This was done for the many variables for which larger analyses mentioned in the thesis had been performed, e.g., nested ANOVAs. Replicated G2 tests-of-independence were done to compare species for % reproductive. Analyses that involved interactions such as the coastal versus inland analyses, were re-run with only two species, and the species × site interaction or the species × site × date interaction are reported. The root cross-sectional area of D. caespitosa #25 in the coastal garden was winsorized. Table F-1 reports on the two “planned” comparisons. As regards the habitat differences between D. c. ovatifolia and D. c. agourensis, there were no significant differences in slope or in nearest neighbor distance but there was in aspect. The life-history traits longest leaf length, leaf number, inflorescence height, fruit number, seed projected area, and percent seedling survival were significantly different but seeds per fruit, and percent seed germination were not. Dudleya c. ovatifolia had smaller and fewer leaves, shorter inflorescences, fewer fruits, smaller seed projected area, and lower percent seedling survival than D. c. agourensis. Leaf size of plants grown in the coastal and inland gardens differed significantly between gardens and over time, but rosette weights and root-crown cross-sectional areas of plants grown in the 77 Table F-1. Planned comparisons of species sometimes considered indistinct. For interactions, numbers reported are differences between means, coastal - inland; for example, this entry represents Y’ of coastal D. c. ovatifolia – Y’ of inland D. c. ovatifolia versus Y’ of coastal D. c. agourensis – Y’ of inland D. c. agourensis. D. c. ovatifolia vs. D. c. agourensis Habitat variables Slope, degrees Nearest same species neighbor distance, cm Aspect Life-history variables Longest leaf length (log) Leaf number (sq rt) Inflorescence height (log) Fruit number (cube rt) Seeds per fruit (-1/6) Seed cross-sectional surface area, mm % germination (arcsine) % seedling survival root:shoot (log) 10-90 vs. 25-90 25.8 vs. 68.3 0-180o vs. 90-0340o 1.8 vs. 2.6 5 vs. 10 11 vs. 26 cm 19 vs. 74 52 vs. 62 0.254 vs. 0.288 59 vs. 64 81 vs. 92 -0.597 vs. -0.409 Coast/inland interactions with species Rosette weight g, 0.2844 vs 0.3288 sqrt(Y+.5) Root crown x-sectional 0.0252 vs. 0.0167 area cm2, log (Y+1) % Reproductive 4.8 vs 0.1 P F (df) P = 0.148 2.154 (1, 58) P = 0.436 0.617 (1, 58) P < 0.001 113.506 (1, 58) P = 0.001 11.323 (1, 58) P < 0.001 76.418 (1, 58) P < 0.001 92.161 (1,58) P < 0.001 65.415 (1,58) P = 0.966 0.002 (1,57) P = 0.012 7.619 (1,20) P = 0.137 2.242 (1, 115) P < 0.001 17.579 (1,115) P = 0.004 8.867 (1,51) P = 0.580 0.309 (1,101) P = 0.505 0.448 (1,100) Test not reliable Leaf size log(Y+1) see Figure 6 P < 0.001 Spp x garden x date 8.427 (7, 749) Watering treatment interactions with species Projected SA, sqrt cm2 0.330 vs 0.249 P = 0.859 Water – no_water 0.032 (1, 109) Projected SA, sqrt cm2 0.719 vs 0.990 P = 0.525 Mist – no_water 0.406 (1,110) Photosynthetic activity see Figure E-3 P = 0.042 Spp x treatment x date 1.529 (26, 2145) 78 D. lanceolata vs. D. caespitosa 0-90 vs. 0-90 168 vs. 50.8 50-0-140o vs. 230-0320o 3.8 vs. 6.4 7 vs. 16 69 vs. 49 83 vs. 88 123 vs. 58 0.369 vs. 0.427 80 vs. 73 96 vs. 94 -0.287 vs. -0.415 0.2297 vs 0.5059 -0.0188 vs. -0.009 -7 vs -10 see Figure 6 0.533 vs 2.628 0.874 vs 0.965 see Figure E-3 P F (df) P < 0.001 27.996 (1, 58) P = 0.028 5.103 (1, 58) P = 0.089 2.994 (1, 58) P < 0.001 56.203 (1, 58) P < 0.001 92.982 (1, 58) P < 0.001 14.540 (1,58) P = 0.820 0.052 (1, 58) P < 0.001 24.515 (1,57) P = 0.002 12.841 (1, 20) P = 0.035 4.550 (1, 113) P = 0.033 4.642 (1, 113) P = 0.005 8.362 (1, 55) P = 0.186 1.769 (1,114) P = 0.843 0.039 (1,114) P = 0.769 G2hetero=0.086 (1) P = 0.002 3.210 (7,798) P = 0.002 10.489 (1,108) P = 0.872 0.026 (1,112) P = 0.233 1.190 (26, 2184) gardens were similar. Differences between percentages of individuals reproductive in each of the gardens were not tested because the percentages were near 100%. The size of plants grown in the different watering treatments, as measured by projected surface areas, did not differ between the water and no-water treatments nor between the mist and nowater treatments. In terms of reproduction in the watering treatment, there was no significant difference in the first spring. The second and third springs were not tested because the test is not reliable for percentages near 100%. Comparing D. lanceolata versus D. caespitosa, substrate slope and nearest neighbor distance were significantly different, but aspect was not. Dudleya lanceolata and D. caespitosa showed significant differences in longest leaf length, leaf number, inflorescence height, seeds per fruit, seed projected area, percent seed germination, and percent seedling survival but not number of fruits. Dudleya lanceolata had smaller and fewer leaves, longer inflorescences, more seeds per fruit, smaller seed projected areas, and greater percent seed germination and percent seedling survival than D. caespitosa. Plants grown in the coastal and inland gardens showed significant differences in linear leaf lengths between gardens over time, but there were not differences between rosette weights, root-crown cross-sectional areas, or percent of individuals reproductive. Projected areas of plants grown in the different watering treatments differed between the water and no-water groups. Dudleya caespitosa, which was not prone to flowering, responded to water much more strongly in terms of projected area than D. lanceolata, which were more likely to flower. There was no significant difference between the mist and no-water groups. Differences between D. lanceolata and D. caespitosa were heterogeneous in bolting behavior between the watering treatments in the third spring, 79 with D. lanceolata being more likely to bolt (Figure F-1). Differences in bolting behavior in the first and second springs were not tested because percentages were close to 0 or 100%. 1 mist 0.9 water 0.8 nowater % Reproductive 0.7 0.6 G2 df P mist 25.22 1 <0.001 water 10.37 1 0.001 0.3 no water 52.14 1 <0.001 0.2 total 87.72 3 <0.001 0.1 pooled 75.13 1 <0.001 heterogeneity 12.60 2 0.002 0.5 0.4 0 LANC CAES Figure F-1. Bolting in the third spring for D. lanceolata and D. caespitosa in the three watering treatments. Sibling plants were randomly placed in three different watering treatments: water, mist, and no-water. Water treatment plants received natural precipitation or water weekly. Mist group plants received misting the four warmest hours of the day and natural precipitation. No-water plants received only natural precipitation. Presence of inflorescences was recorded each of three springs. Percent of reproductive individuals was calculated by dividing the number of individuals per species in each watering treatment with inflorescences by the total number of individuals in each species in each watering group. All n = 29 plants per species per treatment. Considering all of the habitat and life-history characters measured, each species was distinct from every other species in multiple traits. The comparisons between D. c. ovatifolia versus D. c. agourensis and D. lanceolata versus D. caespitosa support the view that each species is unique. In contrasting D. c. ovatifolia and D. c. agourensis, there were significant differences in one of the three habitat characteristics and six of the eight life-history characteristics measured. In contrasting D. lanceolata and D. caespitosa, two of the three habitat characteristics and six out of eight life-history 80 characteristics were significantly different between the species. True, statistical significance is not the same as 100% discriminate distinctiveness, but the species can be identified based on single characters (such as color and amount of branching), and the large number of other characters that tend to differ confirm that these key characters are indicative of differentiation between lineages. 81 Appendix G: Management implications and recommendations This thesis is an important step in learning about the life-histories and other factors that affect the prevalence and persistence of rare Dudleya species in the Santa Monica Mountains region. However, more needs to be known in order to improve plans for their conservation. Factors that have been found to influence the prevalence of the rare species are that they produce fewer fruits, fewer seeds per fruit, smaller seeds, are smaller in size, have shorter inflorescences, require co-occurring vegetation to be of a lower stature, milder microclimates, and aspects facing closer to north than the common species. The rare species also seem to depend on the presence of mosses, lichens, and/or Selaginella. The life-history traits of the rare species affect their ability to increase population size and disperse into other habitats. Therefore, a recommended follow-up to this study should be the establishment of a long-term demographic monitoring program. Monitoring is necessary in order to ascertain if the rare species’ populations are increasing, stable, or in decline. Monitoring would also allow for the determination of which life stages are most critical so that conservation efforts could be concentrated on those stages (Elzinga et al. 1998). Specifically, it would be useful to know if mortality is mainly in the first few months after germination, during the remaining time period until first-reproduction, or between subsequent reproductive bouts. It would also be desirable to determine whether factors associated with mortality differ among the various rare and common Dudleya species. In conjunction with the monitoring, a seed longevity study should be conducted in order to determine if these species have a seed bank and the number of years the seeds are viable (Elzinga et al. 1998). This information will make it possible to perform 82 population projections alerting mangers of potential declines before they occur (Elzinga et al. 1998). Once there is better understanding of the rare species’ demographics, supplemental research projects could be done. If populations are in decline, it would be important to know if, in a metapopulation sense, some were sources and others sinks. It would be a good idea to look at the genetic make-up of each grouping of species to determine population and possibly meta-population structure. If number of viable seeds being produced or seedling survival were critical for population stabilization then a study of pollination services could determine the need for supplemental manual pollination (Levin and Mulroy 1985). If it were determined that seedling establishment is a critical stage for population persistence, then seeds from declining populations could be collected and sown in captivity. The resultant plants could be cross pollinated with each other (derived from the same population) and those seeds could be put back into the natural population to augment seed production and improve seedling establishment (Pavlik 1997). A better understanding of the rare species’ habitat requirements could also be used to identify suitable habitats for the establishment of new populations. This study and personal observations indicate that population size is dependent on the availability of places for seedling establishment and probability of recruitment. Understanding the importance of bryophytes, lichens, and/or Selaginella for seedling establishment is critical. Observations could be made to verify this dependence. Furthermore, the rare species’ populations are denser than those of the common species in part because of the increase in suitable habitat where mosses, lichens, and/or Selaginella are present. It would be important to know if there were a tradeoff between density and interspecific 83 competition. If density is found to have a significant effect, determination of density thresholds for each species would be valuable. In this way population sizes could be maximized if seed augmentation or introduction projects were undertaken. Habitat protection efforts should also be enacted. Species that are locally abundant and adapted to specialized habitats may be able to persist as long as those habitats exist (Levin 2000). It will be important, then, to monitor usage of the habitats where these populations occur in order to prevent destruction (Lavergne et al. 2004, Lloyd et al. 2003) that could negatively impact the rare Dudleya species growing there. 84