Stay informed: Sign up for eNews Subscribe
Read Chapter 1


Biotic Uniqueness

An Overview

Endemism, or the confinement of species or other taxa to particular geographic areas, can be a slippery concept. Every species is confined to some place; for example, it has been estimated that more than 90 percent of the world's plant species are found in only one floristic province (Kruckeberg and Rabinowitz 1985). So when do species or places become interesting on account of their "endemism"? Islands with unique floras and faunas provide the clearest answer. It is no accident that the Galápagos were instrumental to Darwin's thinking. Long-distance colonization, the curtailment of gene flow with close relatives, adaptation to new biotic and abiotic conditions, and (in some cases) the survival of ancestral forms that have become extinct on mainlands can be seen and studied with exceptional clarity on islands that are rich in species found nowhere else. Similar evolutionary forces may be revealed to operate more subtly in regions and habitats with islandlike qualities. California is a good example of an islandlike area within a continent; it is a region of mediterranean climate completely surrounded by mountains, desert, and ocean hostile to much of its flora and fauna, and the nearest similar "islands" are far away, in Chile and the Mediterranean Basin.

The endemic-rich Californian flora has been an influential living laboratory for the study of plant adaptation and speciation. Two of the founders of modern plant evolutionary biology were G. Ledyard Stebbins (1906-2000; UC Berkeley and UC Davis), who first focused evolutionary theory on the study of plants with his Variation and Evolution in Plants (1950) and whose work called attention to the central roles of hybridization and polyploidy in plant speciation; and Jens Clausen (1891-1969; Carnegie Institution), who is best known for leading interdisciplinary experimental studies of genetic differentiation of plant populations along gradients and who wrote Stages in the Evolution of Plant Species (1951). Since the mid-twentieth century, there has been a flourishing tradition of using endemic-rich Californian genera such as Clarkia, Ceanothus, Limnanthes, Madia, and Mimulus as model systems in evolutionary biology (see Chapter 3).

Problems in Defining Endemism

Before discussing endemism, or geographic restriction, of species to either the state of California or the California Floristic Province (CFP), let us consider some of the issues that affect its definition.

Relationship to Rarity

In common with many other works, this book uses the term endemism to mean the condition of having a limited geographic range, regardless of whether a species can be considered rare. However, in the literature on the biology of rarity, the term is sometimes used in a narrower sense. For example, in a classic review of endemism in higher plants, Kruckeberg and Rabinowitz (1985) define endemics as species existing as only one or a few populations. They note that such species can nearly always be considered rare in the sense of having very small geographic ranges. Many endemics (as defined by these authors) are also rare in the sense of having narrow niches; the best-known examples are plants specialized on particular soils, often called "edaphic endemics." Endemism is uncorrelated with a third type of rarity, namely, low population density; these authors note that endemics are often locally abundant within their narrow geographic ranges or habitats.

Appropriate Spatial Units

Islands are natural units for defining and measuring endemism, because the boundaries of an island are clearly defined and obviously linked to the evolutionary processes giving rise to unique species. This is less true for almost any other kind of geographic unit. Political boundaries seem especially inappropriate since they are unrelated to biology, yet the majority of the world's biodiversity data are compiled by country, state, province, or other similar unit. In the United States, an important source of data is the Natural Heritage Network, a national program founded by the Nature Conservancy in the mid-1970s and now implemented by each state. Each member of the network-in California's case, its Department of Fish and Game-compiles occurrence records of imperiled species and other conservation elements such as natural communities and makes these records available in an interchangeable format. Analyses of these data (Stein et al. 2000), discussed in Chapters 3 and 4, point to California as the U.S. state with the highest number of total and endemic species, although Hawaii is higher in percentage endemism, as is often true of oceanic islands. The problem with this state-based approach is that it greatly understates the diversity of biogeographic regions that occur across many states. Appalachia is an important U.S. center of biodiversity and endemism that encompasses eight states, none of which ranks particularly high in state-level analyses.

Ecoregions are units defined by biogeographers on the basis of shared climates, vegetation types, and major assemblages of species. Various classifications are used by conservationists (e.g., World Wildlife Fund, Nature Conservancy), resource managers (e.g., U.S. Forest Service, Environmental Protection Agency), and biological databases (e.g., The Jepson Manual [Baldwin et al. 2012]). Analyzing endemism by ecoregions seems more defensible than by states, but it has its pitfalls too, and California is a good example. In a global conservation assessment (Ricketts et al. 1999), California does not register high in either species diversity or endemism; as the authors acknowledge, this is because California is so diverse that it is divided into 13 ecoregions. If California's biological uniqueness results from a heterogeneous landscape, across which a common ancestral pool of species has diverged into many localized endemics, this approach underestimates the "true" diversity of California to the same extent that the state-based approach underestimates Appalachia.

Biogeographic units based on assemblages of related species are another alternative. In the most widely used system, the California Floristic Province forms part of the Madrean Region, which belongs in turn to the Holarctic Kingdom (Table 1; Takhtajan 1986). The majority of authors define the California Floristic Province as including all nondesert parts of the state of California, plus south-central Oregon and northwestern Baja California (Figure 4; see, e.g., Raven and Axelrod 1978; Conservation International 2011; Baldwin et al. 2012). Under a narrower definition, the wetter areas of northwestern California and southern Oregon may be considered part of the Rocky Mountain Province (Takhtajan 1986). The California Floristic Province broadly coincides with the mediterranean climate or mediterranean biome, as defined by rainy winters, dry summers, annual precipitation of 25 to 100 centimeters, and sclerophyllous vegetation (Dallman 1998). (Again, by some definitions, northwestern California and southern Oregon are too rainy, the Sierra Nevada too snowy, and parts of the Central Valley too arid to be considered mediterranean.) Under the broad definition, which is consistent with a floristic analysis of the West Coast (Peinado et al. 2009), there is 70 percent geographic overlap between the state of California and the California Floristic Province (Conservation International 2011). Thus it is reasonable to speak of endemism in California as a natural phenomenon and not just the product of a political boundary. This book uses the broad definition of the California Floristic Province (see Figure 4), in accordance with major works on the flora (Raven and Axelrod 1978; Baldwin et al. 2012), and a novel effort is made to compile data on its animal endemism.

Data on endemism in the state of California were generally obtained from published sources (plants, Baldwin et al. 2012; mammals, CDFG 2003; birds, Shuford and Gardali 2008; reptiles and amphibians, Jennings and Hayes 1994; fish, Moyle 2002; butterflies, Pelham 2008); these lists were updated for taxonomic and distributional changes by consultation with experts. Data on endemism in the California Floristic Province were harder to obtain. Remarkably, for plants there is currently no database from which the thousands of species endemic to the Floristic Province can easily be counted or identified, but a preliminary attempt is made in this book (see Appendix). For animals, the modest lists of Floristic Province endemics were obtained by visually interpreting range maps in atlases and by asking experts on each group.

Spatial and Taxonomic Scales

Systematic biases in the estimation of endemism arise from both spatial and taxonomic scales. Larger geographic units will tend to have more endemics than smaller ones. Converting numbers of species to species density (species/area), as is sometimes done, is not a valid correction for this bias because the expected number of species (S) does not increase linearly with area (A). Instead, it follows a logarithmic relationship, S = cAz, where the exponent z is typically 0.15-0.35 among islands or other units that share some of their species. A tenfold increase in area therefore results in only an approximate doubling of species, and species density (S/A) has a strong bias toward being higher on small islands. Among continents or other units sharing relatively few species, z may approach 1.0, reducing the bias in species density (Rosenzweig 1995). Still, the best way to correct diversity for variation in area is to use S/Az, where z is estimated from regressing ln(S) on ln(A). Another solution is to calculate diversity and endemism from species range maps that have been converted to equal-area polygons (e.g., Stein et al. 2000; CDFG 2003), as long as the underlying data are accurate enough.

With regard to taxonomic scale, some data sources report endemism based on all named taxa (species, subspecies, and varieties); others report only full species. Logically, endemism in a given geographic area will always be higher among taxa of lower rank (Kruckeberg and Rabinowitz 1985). Taxa below the species level are described more often and on the basis of smaller differences in vertebrates than invertebrates, and in showier invertebrates (butterflies) than inconspicuous ones (most others). Examples from California suggest this leads to considerable bias. In kangaroo rats, 23 subspecies but only 5 full species are endemic to the state (Goldingay et al. 1997). In birds, 64 named taxa but only 2 full species are state endemics (Shuford and Gardali 2008). In plants, however, endemism is 34 percent for all named taxa and 28 percent for full species (Chapter 3, Table 3). Grasshoppers show endemism of 53 percent for full species plus subspecies and 51 percent for full species only (Chapter 4). The much smaller disparities for plants and grasshoppers than for kangaroo rats and birds suggests that subspecies and varieties are less often described in plants and invertebrates than in vertebrates. In the majority of invertebrates, in fact, surveys are too incomplete for even crude estimates of species-level endemism (Chapter 4). Full species are the focus of this book because of the extra subjectivity and bias introduced by subspecies.

Defining species remains a perennial source of debate in both plant and animal systematics (Mallet 2001). Traditionally, most taxonomists have sought consistent breakpoints in the variation of multiple traits, presumably reflecting a lack of gene flow, as a way to define the boundaries between related species (e.g., Oliver and Shapiro 2007). As molecular data have become increasingly available, one alternative that has gained popularity is that any unique trait can define a lineage as a species (Mallet 2001). In practice, these diagnostic traits are often variations in mitochondrial DNA, which evolves relatively fast in animals. Many existing species can be split up into multiple, small-ranged, and morphologically nearly identical new species under this concept (Agapow et al. 2004). The California raven, for example, could be its own species based on molecular variation, even though it does not differ in appearance or behavior from other North American ravens (Omland et al. 2000). Species numbers would more than double in plants and nearly double in most groups of animals under this "phylogenetic" or "diagnostic" species concept (Agapow et al. 2004), leading to even more substantial increases in endemism. This book accepts and includes all species that have been formally described by any method but does not deal with proposed new species of unclear status, nearly all of which are subdivisions of existing species.

Relative versus Absolute Values

Endemism may be reasonably expressed and compared either in percentages or numbers of species. It is worth remembering that percentages are more meaningful the greater the diversity as well as the higher the taxonomic rank of the group being examined. Thus 50 percent Californian endemism in the grasshopper family Acrididae (with 186 species in the state) means more for the state's biotic uniqueness than 50 percent endemism in the grasshopper families Eumastacidae and Tanaoceridae (4 and 2 species in the state), or even than 86 percent endemism in the 21 species of Timema (a genus of walking stick insects). Throughout this book, endemism is expressed in both numbers and percentages, in the belief that they provide complementary information.

Comparative information from other geographic regions is essential to characterizing and explaining Californian endemism. Acridid grasshoppers are one of the most endemic-rich groups in California, but they may be equally so in other parts of the mountainous western United States (Knowles and Otte 2000). Whether or not it is remarkable that 5 of 23 kangaroo rats (Dipodomys) or 21 of 22 slender salamanders (Batrachoseps)are endemic to California depends on whether ecologically similar groups are just as diverse in neighboring regions. It is challenging to find, for almost any group, either comparative data or interpretive analyses that place endemism in California in a larger context. This book relies on comparisons with other states and the other four mediterranean climate regions to provide a context for Californian endemism.

Large-Scale Patterns in Species Richness and Endemism

One of the best predictors of species richness at a global scale is plant productivity, which is determined at large scales by the abundance of water and solar energy. At low latitudes water exerts stronger control, whereas at high latitudes solar energy is a stronger limitation. There are consistently more species of plants and animals in the warm and wet parts of the world than the colder or drier ones, regardless of whether the latitude is tropical or nontropical (Figure 5a; Hawkins et al. 2003). Within the United States as a whole, plant and vertebrate animal diversity is higher in the warmer southerly states (Stein et al. 2000). Within California, in contrast, the diversities of plants, birds, mammals, and amphibians (although not reptiles) are highest in the rainier north (CDFG 2003). However, this is a case where the exception proves the rule, because California is a sunny but arid region in which water is the limiting factor governing plant productivity. Plant diversity in California is positively related to a remotely sensed index of productivity, which in turn is strongly related to rainfall but not to temperature (Figure 5b; Harrison et al. 2006).

Levels of endemism may follow geographic patterns different from total species diversity. Isolated islands, for example, are often high in endemism but low in total richness. Endemism on continents is harder to explain, but one recent analysis suggests that global patterns in endemism are best explained by climatic stability. Sandel et al. (2011) defined the climate change velocity of any given location as the ratio of climatic change over time at that location since the last glacial maximum, 22,000 years ago, to the average change in climate over space at the same location at present. This velocity represents how fast an organism has had to shift its distribution to keep pace with postglacial warming. It is slow, for example, in mild maritime climates that have undergone less change in temperature over time and in rugged regions where present-day temperatures vary sharply over short distances (e.g., from low to high elevations). Globally, animal endemism is higher where climate change velocity is lower, and this effect is stronger for sedentary amphibians than for mobile mammals and birds, suggesting that stable climates have promoted the persistence of sedentary species with small geographic ranges (Figure 6; Sandel et al. 2011).

Endemism has been an important concept in conservation, as manifested by efforts to identify hotspots of high numbers of species found nowhere else. In the most famous such analysis, Myers et al. (2000) defined global hotspots as regions with more than 1,500 endemic plant species and more than 75 percent loss of primary vegetation cover. The 25 hotspots thus identified make up only 1.5 percent of the world's terrestrial land surface but hold over one-third of the world's vertebrates in four major groups, as well as one-half to two-thirds of the plants and vertebrates on the IUCN Red List of Threatened Species. Nearly all hotspots are in the moist tropics or subtropics, contributing to their high overall diversity, and in the tropics they tend to be on islands or in islandlike mountain ranges, contributing to high endemism. Almost the only nontropical hotspots are found in the five mediterranean climate regions of the world, including the California Floristic Province. These five regions are also lower in vertebrate diversity than the other hotspots.

Global analyses of animal endemism generally find similar results to those of Myers et al. (2000), except where the mediterranean climate regions are concerned. For example, Rodrigues et al. (2004) used distributional maps for mammals, amphibians, turtles, tortoises, and endangered birds to identify the global regions with highest "irreplaceability," or numbers of species not found or protected anywhere else. The results highlighted many of the same tropical islands and mountain ranges identified by Myers et al. (2000) but not the five mediterranean climate regions. Likewise, Lamoreux et al. (2006) found that hotspots of amphibian, bird, mammal, and reptile endemism tended to coincide, but none of these concentrations occurred in the mediterranean regions. Nor did the mediterranean regions score as globally significant for total, endemic, or endangered birds (Orme et al. 2005)

Within the United States, including California, hotspots of endemism have been identified using a metric called rarity-weighted richness (Stein et al. 2000; CDFG 2003). A region is divided into equal-area polygons, within each of which the rarity-weighted richness is the sum of each species present in the polygon divided by the number of polygons occupied by that species. The output is a map showing high concentrations of narrowly distributed species (Figures 7, 8). The input data are often coarse and incomplete; in these examples, only Heritage Network-listed species are included, and their distributions are less than fully known. Also, the results may sometimes be dominated by small numbers of imperiled species with very tiny ranges; there is no single "correct" way to balance the contributions of number of species and range sizes in this type of analysis. Nonetheless, it provides a synoptic view of biodiversity that emphasizes endemism.

Within the United States, hotspots of rarity-weighted richness for Heritage Network-listed species (Figure 7) are the greater San Francisco Bay Area, part of coastal and interior Southern California, Death Valley, Hawaii, the Florida Panhandle, and the southern Appalachians. The first two reflect large numbers of plants threatened by both natural rarity and urbanization. Plants comprise 113 of 135 imperiled species in the San Francisco Bay Area and 72 of 81 in the Southern California hotspot (Stein et al. 2000). The Death Valley hotspot comprises fish, snails, insects, and plants confined to one or a few desert springs. Appalachia's imperiled species are mainly freshwater mussels, fish, and amphibians; Hawaii's include many birds; and Florida's include many reptiles and amphibians, although plants are also important in all three (Stein et al. 2000).

Within California, concentrations of rarity-weighted richness for nearly all groups of species (including mammals, birds, reptiles, and invertebrates) occur in the San Francisco and the Los Angeles-San Diego urban coastal regions (CDFG 2003). High levels of sampling effort and high degrees of endangerment may contribute to this pattern. For plants (Figure 8a), there are additional and presumably more natural hotspots in the Klamath-Siskiyous, Modoc Plateau, southeastern Sierra, Central Coast, and Mojave Desert. For amphibians (Figure 8b), the southern Sierra Nevada and Transverse Ranges stand out; and for fish (Figure 8c), the Modoc Plateau and Death Valley are especially rich in small-ranged species.

Origins of Endemic Species

Paleoendemism and Neoendemism

As pointed out by Stebbins and Major (1965), species may be restricted to a narrow geographic area for two general reasons: either they were once widespread and are now confined to a subset of their former ranges (paleoendemism), or they evolved recently and have had inadequate time to spread (neoendemism). (Some authors consider a third category, "holoendemics," or species restricted by their habitat requirements.) Although humans have had enormous impacts on biological diversity, there are relatively few "anthropoendemics" whose small ranges are attributable to humans (Kruckeberg and Rabinowitz 1985).

Paleoendemics or relictual taxa may have fossil records far beyond their contemporary ranges, like ironwood (Lyonothamnus), found in western Nevada fossil beds but now confined to the Channel Islands. They often occur in discontinuous populations, like the coastal redwood (Sequoia sempervirens) and giant sequoia (Sequoiadendron giganteum), thought to be remnants of once-broader distributions. Usually their habitats are more benign than the surroundings, with higher-than-average summer rainfall and/or mild summer temperatures. Paleoendemics are also classically diagnosed by lacking close relatives nearby, since they represent old and shrinking lineages. The closest relatives of many Californian paleoendemics are found in eastern North America or eastern Asia (Stebbins and Major 1965; Raven and Axelrod 1978). By contrast, neoendemics belong to lineages undergoing recent speciation, so they may occur in complexes of closely related and adjacent species and may be poorly differentiated in morphology, genetics, and/or reproductive compatibility. Examples are discussed in Chapters 3 and 4.

The concepts of paleoendemism and neoendemism have also been applied to edaphic (soil) endemism. Paleoendemism is illustrated by species such as leather oak (Quercus durata), which is confined to serpentine soils but scattered across California and is believed to have once occurred widely on other soils but to have become restricted to serpentine through changes in the climate and/or competitive environment. Neoendemism is exemplified by species such as Layia discoidea, which appears to have arisen recently from a nonserpentine ancestor and to have been restricted to serpentine ever since it evolved (Baldwin 2005).

The distinction is not absolute, of course. Although Lyonothamnus is paleoendemic as a genus, its extant representative, L. floribundus, with its two distinctive subspecies, probably evolved recently on the Channel Islands (Erwin and Schorn 2000). The Streptanthus glandulosus complex appears to consist of a widespread species that became fragmented (similar to a paleoendemic), but some of its members have speciated recently as a result of their restriction to serpentine soil (Mayer et al. 1998). Although the prefixes paleo- and neo- imply differences in age, it should not be assumed that paleoendemic lineages are older unless (as is seldom true) this is actually tested.

It is usually assumed that neoendemism accounts for most of the wealth of endemism in California. It is certainly true that most of the studies of plant endemism in the state have focused on the evolutionary processes giving rise to new species in the region. As a background to the following chapters, the rest of this section briefly reviews modes of speciation by which new endemics evolve. The fundamental challenge is to understand how a single lineage can give rise to two (or more) descendant lineages that remain on separate evolutionary pathways rather than lose their integrity through gene flow. Geographic barriers, natural selection, hybridization, chromosomal rearrangements, and genetic architecture play roles that have been studied and debated for decades.

Geographic Speciation

One basic distinction is between geographic and ecological speciation, where the former implies a strong role for external barriers to gene flow as opposed to natural selection. In the classic model of gradual allopatric divergence, anancestral lineage becomes subdivided by a new mountain range, water body, or similar obstacle. Sequential fragmentation may result in complexes of species with parallel patterns of genetic distance, morphological variation, and variable interfertility. Relatives that co-occur tend to have diverged longer ago than relatives that do not, suggesting that isolation promoted their initial divergence (Baldwin 2006). This is a classic and uncontroversial mode of speciation. It has been studied using the methods of biogeography, where the distributions of closely related taxa are interpreted in light of geologic and climatic events, and more recently using phylogeography, where the genetic patterns within lineages are similarly correlated to historical earth surface events. Biogeographic and phylogeographic evidence suggest that nearly all North American mountain chains are "suture zones," that is, places where plant and animal lineages have diverged and sometimes come into secondary contact (Swenson and Howard 2007).

Ecological Speciation

Natural selection in response to new ecological opportunities is central to ecological speciation. The most spectacular examples are the adaptive radiations that sometimes occur on newly formed islands or in islandlike habitats (e.g., Price 2008). Speciation into new niches also takes place in habitats already full of species, but formidable obstacles make its success unlikely (Levin 2004). Ecological speciation begins with a small lineage colonizing a habitat in which it is ill adapted. It is more likely to go extinct than to continue evolving because of its small population size, its low genetic variability, and the reduction in its potential population growth implied by the existence of strong selective pressures. Also standing in the way of successful adaptation are gene flow from populations in the ancestral habitat and correlations between adaptive and nonadaptive genetic traits. Even if the fledgling species becomes well adapted to its new environment, it risks being genetically swamped by its progenitor species until intrinsic reproductive isolation evolves, which may take longer than adaptation to the habitat. Hybrid origins and allopolyploidy help overcome these obstacles by creating immediate reproductive barriers between a newly adapted species and its ancestors; however, the new species may be ecologically outcompeted by its relatives unless it inhabits a novel niche (Levin 2004). Other pathways to reproductive isolation include strong selection against hybrids (Kay et al. 2011), selection that incidentally favors differences in reproductive traits (e.g., flowering phenology, floral morphology), and linkage or epistatic interactions between genes involved in adaptation and genes that confer reproductive isolation (Wu et al. 2007).

Progenitor-Derivative Speciation

Another basic distinction is whether speciation results in sister taxa with roughly equal initial population sizes, geographic ranges, and genetic diversity, as in classic gradual allopatric divergence, or whether it involves a small population budding off within the range of a widely distributed species. The latter case, called peripheral isolate formation or progenitor-derivative speciation, leads to a localized neoendemic species that is phylogenetically embedded in its ancestral species. The ancestor then becomes paraphyletic; that is, it does not include all descendants of a single common ancestor. Progenitor-derivative speciation appears common in plants (Grant 1981; Gottlieb 2003; Baldwin 2006). It is related to the classic idea of "catastrophic speciation," in which a sudden event causes a population decline in a widespread species, allowing a random chromosomal arrangement to become rapidly fixed in a small population that evolves into the derivative species. Chromosomal alterations, novel habitats, breeding system changes, and adaptive morphological differences may contribute to reproductive isolation between progenitor and derivative species. Changes appear to be moderate and to involve a small number of genes, and overall genetic similarity between progenitor and descendant remains high (Gottlieb 2003).

Hybrid and Polyploid Speciation

The origin of new species through hybridization and/or changes in chromosome number is sometimes detected in animals but is central to evolution in plants. In his classic Plant Speciation (1981), Grant argued that patterns of relatedness among plant lineages are so often networklike, due to both modern and ancient hybridization events, that the standard view of evolution as a branching "tree of life" may not apply. Hybridization generates adaptive potential because it increases genome size and allows various duplicate genes to be turned on or off in hybrid progeny. Hybridization between species with different chromosome numbers or structures initially produces progeny of low fertility because of chromosomal incompatibility, but chromosomal doubling or other rearrangements may restore fertility, in addition to causing reproductive isolation between hybrid and parental lineages. The resulting allopolyploid hybrids are thought to be particularly capable of rapid evolution because they have the full chromosomal complement of both parents. Hybridization between parental species with the same chromosome numbers and structure produces hybrids with the same chromosome number as the parents. These homoploids face lower initial barriers to fertility than allopolyploid hybrids but are more likely to be genetically swamped by the parental lineage. For reproductive isolation to develop in homoploid hybrids, within-chromosome rearrangements and geographic or ecological isolation may be necessary (Rieseberg 2006). Both allopolyploid and homoploid hybrids may exceed the parental species in the values of quantitative traits, and such "transgressive hybridization" may facilitate the invasion of new niche space (Grant 1981; Rieseberg 2006). Autopolyploids are products of within-species gene duplication rather than hybridization. Like allopolyploids, they enjoy the adaptive benefits of larger genome sizes (Soltis et al. 2004).

Recent molecular studies have confirmed the conclusions of classic authors that many plant lineages are of polyploid origin. Individual polyploid "species" may arise multiple times. Autopolyploidy and genome-wide duplication events are more common relative to allopolyploidy than was once believed. Rapid chromosomal rearrangements, genomic downsizing, and changes in gene expression following polyploid origins are beginning to be studied, as are the relationships of these genomic changes to pollination, reproductive biology, and ecological traits. A growing number of comparative studies illustrate the potential for polyploid hybrids to have new, broader, or more finely partitioned niches than their ancestors. Polyploid lineages may diverge ecologically or reproductively through the loss or silencing of alternative duplicate genes, contributing to their evolutionary dynamism (Soltis et al. 2004).

Hybridization and polyploidy have been credited with important roles in the rapid evolution of Californian flora (Stebbins and Major 1965). A common pattern in the California flora is for close relatives to hybridize intermittently, perhaps in certain zones, yet to remain distinct over their core geographic distributions. This pattern, seen in Arctostaphylos, Chorizanthe, Eriogonum, Monardella, and other taxa, is informally known to botanists as the "California pattern" (Skinner et al. 1995).

Traits of Endemic Species

Rare species are sometimes found to have lower genetic variability, higher rates of selfing, lower reproductive investment, poorer dispersal, higher susceptibility to natural enemies, or less competitive ability than common ones (Kruckeberg and Rabinowitz 1985; Lavergne et al. 2004). These traits are interpreted as factors that may cause rarity, that is, prevent species from achieving higher range sizes or abundances. Other studies find that rare species inhabit less competitive (e.g., rockier) or more benign (e.g., rainier, less seasonal, less fragmented) environments than their common relatives (Lavergne et al. 2004; Harrison et al. 2008); such extrinsic differences may be interpreted as factors that have helped species persist, given that they are rare for other reasons. Finally, some traits such as higher inbreeding and lower genetic variability could be interpreted as either causes or consequences of rarity. Not many strong generalizations have arisen from the literature on the biology of rarity, and the usual conclusion is that rarity is too complex to have a single cause.

To understand high endemism within a geographic region such as California, it would be interesting to ask whether endemics have any consistent attributes, either intrinsic or environmental, that explain their diversity relative to other taxa and other regions. For example, in the Cape flora of South Africa, it has been proposed that plant adaptations to nutrient-poor soils, including fine, oil-rich, flammable foliage and ant-dispersed seeds, produce high rates of speciation by conferring short generation times and low dispersal, thereby leading to exceptional diversity (Cowling et al. 1996). In California, Wells (1969) proposed that the diversification of Arctostaphylos and Ceanothus is linked to the loss of resprouting in these two shrub genera, which regenerate by seeds after fire and thus have shorter generation times than their resprouting relatives. The annual life form is a widespread adaptation (or preadaptation) to California's summer drought that may similarly facilitate rapid speciation (Raven and Axelrod 1978). There are now formal phylogenetic methods for testing the relationship between such possible "key innovations" and rates of diversification, but they have not yet been applied to California endemism. However, specialization to serpentine in 23 genera of California plants was shown to be the opposite of a key innovation; when lineages become endemic to serpentine, their diversification rates decline (Anacker et al. 2011).

Endemism is biologically interesting as long as the geographic unit being studied is meaningful in terms of processes that create and maintain diversity. Californian endemism is best studied at the scale of the California Floristic Province, a natural biotic unit defined by a flora with a shared evolutionary history, but this book also considers state-level endemism due to data constraints. Taxonomic scale also influences the study of endemism. This book focuses on full species because some groups of organisms are much more extensively split than others into units below the species level, which by definition have higher levels of endemism.

Species richness is highest in parts of the world where water and solar energy are abundant. The richness of endemic species is also high on islands and in historically stable climates. The mediterranean regions of the world are unusual in being rich in plant endemism, yet not as obviously rich in animal endemism, whereas other global hotspots (nearly all of which are tropical) do not show this disparity.

Endemics arise through the shrinkage of widespread geographic ranges (paleoendemism) and the evolution of new species (neoendemism). Pathways to neoendemism include allopatric divergence, ecological speciation, peripheral isolate formation, and hybridization, all of which are well known in the Californian flora. There is not yet a predictive understanding of the traits that make lineages likely to diversify in particular environments such as California's.