Utah State University DigitalCommons@USU

All Graduate Theses and Dissertations Graduate Studies

5-2011

Individualistic Response of Piñon and Juniper Tree Species Distributions to Climate Change in North America's Arid Interior West

Jacob R. Gibson Utah State University

Follow this and additional works at: https://digitalcommons.usu.edu/etd

Part of the Climate Commons, and the Ecology and Evolutionary Biology Commons

Recommended Citation Gibson, Jacob R., "Individualistic Response of Piñon and Juniper Tree Species Distributions to Climate Change in North America's Arid Interior West" (2011). All Graduate Theses and Dissertations. 908. https://digitalcommons.usu.edu/etd/908

This Thesis is brought to you for free and open access by the Graduate Studies at DigitalCommons@USU. It has been accepted for inclusion in All Graduate Theses and Dissertations by an authorized administrator of DigitalCommons@USU. For more information, please contact [email protected]. INDIVIDUALISTIC RESPONSE OF PIÑON AND JUNIPER TREE SPECIES

DISTRIBUTIONS TO CLIMATE CHANGE IN NORTH AMERICA'S

INTERIOR WEST

by

Jacob R. Gibson

A thesis submitted in partial fulfillment of the requirements for the degree

of

MASTER OF SCIENCE

in

Ecology

Approved:

______Thomas C. Edwards, Jr. James A. MacMahon Major Advisor Committee Member

______Terry L. Sharik Gretchen G. Moisen Committee Member Committee Member

______Byron Burnham Dean of Graduate Studies

UTAH STATE UNIVERSITY Logan, Utah

2011 ii ABSTRACT

Individualistic Response of Piñon and Juniper Tree Species Distributions to Climate

Change in North America's Arid Interior West

by

Jacob R. Gibson, Master of Science

Utah State University, 2011

Major Professor: Dr. Thomas C. Edwards, Jr. Department: Wildland Resources

Piñon and juniper tree species have species-specific climatic requirements, resulting in unique distributions and differential responses to climate change. Piñons and junipers co-dominate the arid woodlands of North America as groups with widespread hybridization. Two piñons, Pinus edulis; P. monophylla, and four junipers, Juniperus deppeana var. deppeana; J. monosperma; J. occidentalis; J. osteosperma, are endemic to the midlatitude interior west and form three groups of hybridizing sister species, P. edulis-P. monophylla; J. deppeana var. deppeana-J. monosperma; J. occidentalis-J. osteosperma. Recent droughts have caused widespread mortality among piñons, but have had less impact on junipers and indicate shifts in co-occurrence have already begun in response to global climate change. Within these groups hybridization likely plays an important role in such distribution changes.

The central objective of this thesis is to forecast the distributions of piñons and junipers endemic to the US under modeled climate change for the 21st century. Species iii distribution models are built with an emphasis placed on aligning the life cycle

dynamics of the species within the temporal and spatial resolution of predictor variables,

and within the modeling technique. Two concerns surrounding species distribution modeling are addressed. First, concerns regarding the extent to which species are at

equilibrium with the current climate are addressed by incorporating dispersal into the

model building process. Second, concerns regarding the potential role of hybridization between closely related species are addressed by building distribution models for each of the three sister species groups as well as the six component species.

Species distribution models exhibited individualistic responses to modeled climate change. Modeled areal loss was greater than gain for all species, which is reflected in changes of co-occurrence. Piñon-juniper richness is forecast to increase in the northern Colorado Plateau, eastern Great Basin, and Rocky Mountains. The sister-species models forecast greater areal gain, and less areal loss, along hybridization zones for P.

edulis-P. monophylla and for J. occidentalis-J. osteosperma, but forecast greater areal

loss along the periphery of the component species distributions. The sister-species model

for J. deppeana var. deppeana-J. monosperma forecasts overall greater areal loss than the

component species. In general, forecast changes in latitude and elevation are about one

third of the changes inferred, from the fossil record, to have occurred following the

transition to the current interglacial ~10,000 years ago.

(104 pages) iv ACKNOWLEDGMENTS

I would like to first thank my major advisor, Dr. Thomas C. Edwards, Jr., for creating a phenomenally heuristic environment, the BioDiversity Lab, for his students to learn and practice the techniques of research. Within this lab Tom provides cutting-edge

resources and ample space for exploration. I have worked in this lab since 2005, when

Tom took me on to work on a project with his colleagues Dr. Richard Cutler and Dr.

Karen Beard, who I would also like to thank for treating me as a colleague and

introducing me to the world of research.

Tom's ability to challenge the understanding of his students is uncanny. I will

always remember the early stages of this degree when I would work for a week or more

on some idea to which Tom would respond with a simple question, transforming my

perspective and launching me on new avenues of discovery. Adding to all of this, Tom

provided indispensable support for me as I recovered from some of life's troubling curve-

balls.

I would like to thank Tom's former students whom I had the chance to share the

lab with. Russell Norvell brought me into the lab as an undergraduate in 2005. Tammy

Wilson and Adam Brewerton graciously shared the lab space. Both Russ and Tammy

provided much moral support, encouragement, and guidance as fellow students

navigating graduate school. I would also like to thank Phoebe Zarnetske who was finishing her master's as I entered the lab and immediately became my role model of a successful master's student. v My committee members have each taught me fundamentals of research in their own way. Dr. Jim MacMahon enlightened me to the field of biogeography in my final semester as an undergraduate and has always provided a genuinely human ear to me through this whole process. Dr. Terry Sharik fueled my passion for the investigation of the arborescent habit with his dendrology course. In addition to the technical support Dr.

Gretchen Moisen provided, she also taught me much in the way of working with others.

Among the numerous others who have provided support in various ways, I would like to thank Tracey Frescino of the Ogden FIA. Tracey has helped me with crucial bits of R code as well as patiently assisting my preparation of presentations. I would like to also thank those here at USU that have helped in so many ways. Ben Crabb encouraged my efforts all along and helped keep me light hearted. Marsha Bailey, Lana Barr, and

Cecelia Melder have kept track of me and provided indispensable assistance with the realities of paper-work; I would have been toast without these three.

I would also like to thank my blood-family and my greater friend-family. Both have been comprehensively supportive of my unusual work habits which often left me out of touch for weeks at a time. Yet they seemed always to show up with nutritious meals and open doors just when I needed it most. And finally, I would like to thank

Jaynan Chanceler and Travis Sapp of the Da'wi drum and dance group, who provide a solid rhythm in my life.

Jacob Gibson vi

CONTENTS

Page

ABSTRACT...... ii

ACKNOWLEDGMENTS ...... iv

LIST OF TABLES...... viii

LIST OF FIGURES ...... ix

BACKGROUND...... 1

Overview...... 1 Phylogeny and distribution of piñons and junipers...... 1 Life cycle and coevolutionary relations of piñons and junipers ...... 3

Fertilization...... 4 Dispersal ...... 5 Establishment...... 7 Persistence...... 9

The integrated distribution...... 12 Holocene and current distribution of piñons and junipers...... 14

INTRODUCTION ...... 17

METHODS ...... 23

Study area...... 23 Study species...... 25 Dispersal ...... 27 Modeling domains ...... 30 Data Structure ...... 32 Predictor variables ...... 33 Distribution model ...... 36

RESULTS ...... 38

Model performance...... 38 Elevation shifts in distribution...... 39 Areal shifts in individual distributions and their co-occurrence...... 43 vii Geographic shifts in distributions ...... 54

DISCUSSION...... 57

Variables driving distribution shifts...... 57 Forecast distribution shifts in relation to recent and historic shifts ...... 59 Interpretation of forecast distribution models...... 60

CONCLUSION...... 62

LITERATURE CITED ...... 63

APPENDICES...... 77

Appendix A Demographic literature...... 78 Appendix B Climate variables...... 81 Appendix C Range of species climatic variables within modeling domains...... 86 Appendix D Variable importance plots ...... 93

viii

LIST OF TABLES

Table Page

1. Observed and inferred dispersal distances of piñon and juniper species...... 29

2. Prevalence within sampling frames ...... 33

3. Predictor variables ...... 35

4. Thresholds for modeled presence and absence...... 37

5. Accuracy metrics for Random Forests models ...... 40

6. Elevation shifts of modeled B2a forecast ...... 41

7. Elevation shifts of modeled B2a forecast ...... 42

8. Absolute areal shifts of modeled species A2a forecasts ...... 47

9. Absolute areal shifts of modeled species B2a forecasts ...... 48

10. Absolute areal shifts of modeled sister-species A2a forecasts ...... 49

11. Absolute areal shifts of modeled sister-species B2a forecasts...... 49

12. Relative areal shifts of modeled species A2a forecasts ...... 50

13. Relative areal shifts of modeled species B2a forecasts ...... 51

14. Relative areal shifts of modeled sister-species A2a forecasts ...... 52

15. Relative areal shifts of modeled sister-species B2a forecasts...... 52

16. Displacement of forecast distribution centroids ...... 54 ix LIST OF FIGURES

Figure Page

1. Study area and species ...... 23

2. Modeling domains and modeled current distributions...... 31

3. Distribution forecasts A2a ...... 44

4. Distribution forecasts B2a...... 45

5. Comparison between species and sister-species models ...... 46

6. Areal gain and loss plots...... 53

7. Distribution forecasts of richness and centroids ...... 55

BACKGROUND

Overview

The objective of this research is to model the current and forecast 21st century

changes in the distributions of two piñons (Pinus edulis, P. monophylla) and four junipers

(Juniperus deppeana var. deppeana, J. monosperma, J. occidentalis, J. osteosperma) under two scenarios of projected climate change. First, a brief summary of piñon and juniper phylogeny is presented. The ancestral development of traits that are shared and derived among piñons and junipers is then reviewed in the context of their life cycles and the coevolutionary relationships associated with them. An emphasis is placed on how these relationships collectively scale up to become the climate-driven statistical behavior exhibited by the integrated distribution of these two groups of species. Finally, the present day distribution of piñons and junipers are described in the context of their adjustments to the current interglacial period.

Phylogeny of piñon pines and sabinoid junipers

Conifers were among the first plants to colonize the terrestrial environment, and the Pinaceae were among the earliest conifers (Chaw et al. 2000). The pines (genus

Pinus) are the most ancestral member of the Pinaceae and their emergence is associated with the spread of aridity at mid-latitudes during the Mesozoic. The earliest known fossil is from ~130 million years ago (mya) and corresponds to interior continental mountains, the ancestral Appalachian Mountains where North America and Europe collided as

Pangaea formed (Millar 1993). Pines diverged into three subgenera early on. The three 2 groups are differentiated by the number of vascular bundles, and characteristics of their

cone scales and seed wings. Subgenera Pinus and Strobus are the most distinctive,

whereas Ducampopinus is intermediate and is thought to be ancestral.

Ducampopinus has one fibrovascular bundle per leaf and cone scales that do not seal with resin, like Strobus, but has articulate seed wings like Pinus. Subgenus Pinus has two fibrovascular bundles and inhabits more mesic sites where fire is an important element of their life cycle. The resin-sealed cone scales respond to temperature and humidity. Seeds are generally small and are dispersed by wind. The articulated wing allows the seed to detach as it interacts with the ground. This same strategy is employed by some members of Ducampopinus, but in piñons the wing membrane has become specialized to hold seeds onto the cone scale, distinguishing them as the subsection

Cembroides. The cone scales of Ducampopinus, like Strobus, do not seal with resin.

Strobus, however, is unique in having wings that are rigidly fixed to the seed. Both

Ducampopinus and Strobus inhabit more extreme temperature and moisture regimes

where animal dispersal is an essential force.

Piñons diverged from Ducampopinus by the early Cenozoic (~50 mya), and were associated with extensive mountain building in Mexico (Malusa 1992). They consist of

eleven species (10% of all extant pines, Gernandt et al. 2005), ranging across the arid

interior west from their center of diversity in the Mexican highlands to their northern limits in the Colorado Plateau and the Great Basin (Lanner and Van Devender 1998). The monophyllous mutation of P. monophylla is the most derived trait of the piñons,

emerging ~5 mya in association with the orogenisis of the Sierra-Nevada. It has been

suggested that the single needle is an adaption to the winter dominated precipitation 3 regime consequent of the Sierra-Nevada Mountains (Malusa 1992). The nearest relative of P. monophylla, P. quadrifolia, has four-needles per sheath and inhabits a more arid and hot climatic regime. A hybrid zone occurs where these two species are sympatric, along the zero-degree annual minimum temperature isotherm where hybrids may express either one or four needles. A hybrid zone also exists along the overlap of the two-needle piñon, P. edulis, and P. monophylla, where seasonal moisture transitions from winter- dominated to summer-dominated (Lanner 1974, Cole et al. 2008a).

Junipers (Juniperus) are a much younger genus than Pinus. They diverged from

Cupressaceae in Asia during the global cooling and drying of the Paleocene (~60 mya), and subsequently radiated to Europe and North America (Mao et al. 2010). The divergence of section Sabina from Juniperus occurred in the Mediterranean ~30 mya, with the two main clades diverging into warmer versus colder conditions, respectively.

The serrate margin junipers are a derived group of Sabina which first arrived in N

America soon after their emergence in Eurasia ~25 mya (Mao et al. 2010). The serrate margin junipers radiated during the global cooling and drying of the Miocene. J. occidentalis diverged from J. osteosperma and J. deppeana from J. monosperma during the Late Miocene ~10 mya, both of these divergences were associated with the mountain building events throughout western North America at the time (Mao et al. 2010).

Life cycle and coevolutionary relations of piñons and junipers

A single piñon or juniper tree will produce millions of seeds during an average lifespan span of ~200 years (piñons) to ~500 years (junipers). Above average seed production exceeds >5,000 seeds per tree annually (Chambers, Vander Wall, and Schupp 4 1999) over life spans >800 years (piñons) to >3,000 years (junipers). What, then, limits

piñons and junipers from exponentially spreading across the planet? The answer to this

question emerges from an examination of the life cycle of piñons and junipers, which is

presented in the following four stages: (i) fertilization; (ii) dispersal; (iii) establishment;

and (iv) persistence.

Fertilization. - Fertilization is a central phase in the life cycle of piñons and junipers and facilitates the widespread mixing of genes among a lineage. The pollen grain has the most conserved traits expressed in any phase of the life cycle. The ancestral condition of pollen is reflective of the earliest adaptations plants made in colonizing terrestrial environments, and the chloroplast genome transmitted through pollen carries instructions for the enzymatic catalyst at the core of photosynthesis ancestral to all oxygenic phototrophs (Dismukes et al. 2001).

Pollen was among the first adaptations to terrestrial conditions, as it protected the gametophyte from desiccation and aided in the gravitational orientation of the pollen tube

(Fernando, Lazzaro, and Owens 2005). The ancestral state of pollen is thought to be saccate, with two ‘sacs’ of air oriented in a plane, a trait only retained in the Pinaceae and

Podocarpaceae (Doyle 2010). The dominance of the sporophyte generation marks the initial divergence of vascular plants from their sister taxa, the liverworts and hornworts

(Qiu 2008). The separation of seed cones and pollen cones is an early evolutionary development of conifers (Fernando, Lazzaro, and Owens 2005).

Pollen is shed in huge quantities over a matter of days and the fate of the pollen is contingent upon climatic conditions during this time. The timing of pollen shed is thought to track suitable climatic conditions, and is shed in synchrony with ovule receptivity 5 (Gelbart and Aderkas 2002). Once captured by the cone and absorbed by ovular

secretions, pollen germinates and grows a tube through which a flagellate sperm reaches

the archegonium (Fernando, Lazzaro, and Owens 2005). Development of the fertilized

embryo into a viable seed takes two to three years for both piñons and junipers

(Chambers, Vander Wall, and Schupp 1999).

Dispersal- Piñons and junipers both have seeds adapted to vertebrate dispersal,

although their strategies are quite different. Piñons are highly reliant upon avian

dispersal, whereas junipers have a broader range of dispersal vectors. Coevolutionary

relationships have been important in shaping the cones of both groups as well as the

morphology and behavior of their avian dispersers. Dispersal distances are central to

distribution behavior through time and are discussed in greater detail in the Methods

section.

White pines developed larger seeds as an initial adaptation to the greater

nutritional requirements of seedlings in abiotically stressful conditions, and were

therefore predisposed to develop coevolutionary relationships with avian dispersers

(Tomback and Linhart 1990). Piñon seeds are perhaps the most derived of the white pines

as they have completely lost their seed wing and are among the largest of pine seeds

(Lanner 1998). The large scales open perpendicular to the stem and expose the seeds, which are held in place by a membrane that derives from the same tissue that forms wings on other pine seeds (Malusa 1992). The oily nuts are contained in a thin but hard shell that is light colored when mature but dark when aborted, acting as a visual queue that makes the corvids task more efficient (Ligon 1974, Lanner 1998). 6 Numerous rodents are important in local piñon dispersal (Lanner 1998), but the

morphology of piñon cones and seeds are particularly adapted to dispersal by members of

the Corvidae (Vander Wall and Balda 1977). Corvids share a common ancestor that was

a moderate cacher (de Kort and Clayton 2006). Convergent evolution has subsequently

resulted in analogous adaptations by the Clarke's Nutcracker (Nucifraga columbiana) and piñon jay (Gymnorhinnus cyanocephalus) who are specialist cachers (de Kort and

Clayton 2006). The Clarke's nutcracker utilizes all white pines, and has developed a sublingual pouch for seed transport (Vander Wall and Balda 1977). The piñon jay specializes on piñons and has an expandable esophagus to transport seeds (Vander Wall and Balda 1981).

Unlike piñon seeds which are destroyed by consumption, juniper seeds are encased in fleshy berry-like cone scales. The seed is protected in a hard shell and passes through the digestive tract of their dispersal vector. The evolution of the juniper berry reflects an overall trend in the Cupressaceae towards reduced seed cones (Schulz, Jagel, and Stutzel 2003). Juniper berries range in color from blue to red when they mature, acting as a visual cue to birds. The berries remain on the tree for months or longer and are highly nutritious (Salomonson 1978). In addition to being dispersed by a wide range of vertebrates, juniper berries are dispersed by surface runoff which likely plays an important role in range expansions into lower elevations (Chambers, Vander Wall, and

Schupp 1999). Of all the avian frugivores which utilize juniper berries, the Townsend's

Solitaire (Myadestes townsendi) relies on the juniper berry nearly exclusively as a winter food source (Poddar and Lederer 1982). 7 The interaction between junipers and avian dispersal is strongly influenced by

mistletoes (Phorodendron spp.). Junipers host mistletoes, which are parasitic in a

nutritional sense but may be mutalistic because the mistletoe also attracts avian

dispersers. Regeneration is higher in woodlands which host mistletoe and lower in those

without (van Ommeren and Whitham 2002). Mistletoes also grow on piñons, as they do

on most woodland and forest trees throughout the world (Watson 2001), but their

relationship with piñon dispersal has not been studied.

Establishment. - Piñon seeds are viable for one year (Lanner 1998), whereas

juniper seeds may be viable for two or more years. Juniper regeneration is thus assisted

by a seed bank (Van Auken, Jackson, and Jurena 2004). The establishment of piñons and

junipers is contingent upon temperature, moisture, nutrients, and predators (Chambers,

Vander Wall, and Schupp 1999; piñons: Chambers 2001; junipers: Johnsen 1962). These

factors are moderated by other plants, but the relationship appears to be general to

physiognomy and not specific to taxa. Piñons and junipers must track soil moisture that

varies both vertically and horizontally in the soil through time (Breshears et al. 2009). In

arid woodlands, temperatures are lower, soil moisture higher, and extractable nutrients

greater under shrubs and trees than in open areas (Chambers 2001). Unlike the light

demanding colonizing strategy common to the hard pines, piñon pines are relatively more shade tolerant and successful establishment of piñons is generally greater under shrubs and trees than in exposed sites (Chambers 2001). Similarly, junipers tend to establish more successfully under shrubs and trees than in the open (Johnsen 1962).

Germination of piñons and junipers, where they occur together, has been observed to be equally distributed throughout the woodland (Martens, Breshears, and Barnes 2001) 8 and extend into neighboring shrublands and forests (Chambers 2001; Salomonson 1978;

Poddar and Lederer 1982; Vander Wall and Balda 1977). Differential mortality among

seedlings and juveniles along moisture gradients results in a partitioning of piñons to

moist areas in the woodland and junipers to drier areas in the woodland (Martens,

Breshears, and Barnes 2001). Establishment is often greatest under tree canopies, but

growth rates are greatest at the canopy edges (Van Auken, Jackson, and Jurena 2004).

Junipers have been observed germinating in the open when conditions are suitable

(Salomonson 1978). Junipers are widely observed to act as nurse trees for piñons on

coarse soils (Gascho Landis and Bailey 2005; 2006). This relationship is also noted for

oaks (Quercus spp.) that facilitate piñon establishment (Floyd 1982)

Mycorrhizal mutualisms originated with the early colonization of land by the first

land plants during the early Paleozoic ~380 mya (Wang and Qiu 2006, Strullu-Derrien

and Strullu 2007). Mycorrhyzae assist in acquiring nutrients and water, and have been

found to be critical in the establishment of junipers (Bush 2008) and piñons (reviewed in

Haskins and Gehring 2004). Piñons and junipers have hydrophobic mycorrhizae that funnel surface water into a concentrated depth in the soil profile, which sustains them between precipitation events (Robinson et al. 2010). Arubuscular mycorrhyzae (AM) was the ancestral condition and remains so with many plants including Juniperus, whereas piñons form more derived ectomycchorizal (EM) relationships (Hibbett and Matheny

2009). AM perform better in drought and give junipers a competitive advantage over piñons under moisture stress (Haskins and Gehring 2004). EM, however, is better at acquiring nutrients and give piñons a competitive advantage when moisture is less limiting (Allen et al. 2010). 9 Persistence. - Once established, the growth and persistence of an individual tree

is determined by the ability of the tree to tolerate local fluctuations in soil moisture, and to defend itself against herbivorous insects and microbial pathogens. The ability to

tolerate moisture fluctuations is determined by the transpiration strategy of the tree and

the root-soil interactions. The carbon budget, and consequent persistence of piñons and junipers, is intricately linked with insects and microbes.

Piñons maintain relatively isohydric pressure in their vascular tissue by closing

their stomata as soil and air become dry, whereas junipers allow anisohydric pressures to

develop (McDowell et al. 2008). This allows piñons to have higher pre-dawn pressure

during summer drought (Linton, Sperry, and Williams 1998). The key implication of

these contrasting strategies is that piñons avoid structural damage and recover from

drought better than junipers, but can become carbon starved in extended droughts and are

unable to maintain their defenses, thus succumbing to insects and disease. Junipers, in

contrast, continue to transpire in drought which allows them to fix carbon but at the risk

of greater structural damage (West et al. 2008).

Within P. edulis populations in N. Arizona, the degree of vascular responsiveness

varies annually, with the most responsive trees growing faster when conditions are good

but dying faster when conditions are bad (Ogle, Whitman, and Cobb 2000). Temperature

and moisture variation along an elevation gradient in eastern Nevada resulted in different

growing season lengths experienced by individual P. monophylla. At the moister higher

elevations individual trees expressed two-needles, whereas at the more arid low

elevations individuals expressed one-needle (Jaindl, Eddleman, and Doescher 1995).

Genetic variation in P. edulis is associated with landscape gradients in summer 10 precipitation and is reflective of the broader, regional genetic variation along the continental summer precipitation gradient (Mitton and Duran 2004).

Soils affect the growth form of piñons and junipers primarily through moisture holding capacity (Landis and Bailey 2006). Piñon root activity is hypothesized to be

temperature dependant. Shallow roots have been observed to be dormant in the summer

and active in the fall, whereas junipers were observed to absorb shallow soil water

throughout the growing season (Williams and Ehleringer 2000). However, as a whole

tree, piñons are observed to utilize summer rains more than junipers (West et al. 2007).

The ability of the tree to acquire sufficient moisture largely determines its

susceptibility to insects, although insects may also influence moisture uptake. EM was

found to be diminished by scale insects (Matsucoccus acalyptus) on juvenile piñons, and by boring moths (Dioryctia albovittella) on adult piñons (Del Vecchio et al. 1993,

Gehring, Cobb, and Whitham 1997). Insects quickly followed plants in the terrestrial colonization of the Paleozoic, and every morphological feature, from the cuticle to the xylem, has become habitat for herbivorous insects (Sequeira, Normark, and Farrell 2010).

A wide assemblage of herbivorous insects feed on the leaves, seeds, cambium, and roots of piñons and junipers (Rogers 1993). Each of these morphological traits are partitioned among herbivorous insects (e.g. Turgeon, Roques, and Groot 1994; Rouault et al. 2004).

Herbivorous insect clades are taxon-specific with both generalists and specialists.

Generalists provide a constant source of speciation for specific plant species (Nyman

2010). For example, the gall forming midges of the genus Oligotrophus are specific to

Juniperus and contain species both general to the genus and specific to individual species

(Harris, Sato, and Yukawa 2006). This trend is further exemplified by a variety of the 11 needle scale Matsucoccus acalyptus found to be limited to the hybrid zone of P. edulis and P. monophylla (Christensen, Whitham, and Keim, 1995).

Many of the interactions of herbivorous insects with piñons and junipers have become highly intricate and involve microbes (Janson et al. 2008). Microbial vascular diseases, such as root stain (Ophiostoma spp.), are transmitted by bark beetles (family

Scolytidae) and facilitate their mutual colonization through weakening of the host tree

(Lim et al. 2004). Further, nematode species in the genus Bursaphelenchus are transmitted by the bark beetles. The nematodes have two life phases, the first feeds on the tree and the second feeds on a fungal pathogen also transmitted by the beetle (Wingfield

1987).

A large variety of defensive biochemical compounds have evolved among conifers in coevolution with insects and pathogens (Keeling and Bohlman 2006). Such compounds are found in fossils of early conifers ~168 mya (Marynowski et al. 2007).

The terpenoids are of particular importance and are reflective of the overall phylogeny of both Pinaceae (Otto, Simoneit, and Wilde 2007) and Cupressaceae (Adams 2008).

Terpenoids number in the 1,000s and have intricate relationships with both insects and fungi. However, there is often just primarily one terpenoid found in relationships between a conifer and a given coevolutionary insect. Many insects have evolved metabolic pathways which transform these compounds into pheromones used to attract members of their species who gather to overwhelm a tree's defense (Keeling and

Bohlman 2006). Defense mechanisms may be mechanical, where the infested tissue is shed. Defense mechanisms may also include biochemical signals the tree produces to attract predatory insects. Survivorship of a xylophagous insect, Styloxux bicolor, 12 infecting J. monosperma was found to be reduced by ~33% from branch shedding, by

~16% from sap, and by ~16% from natural enemies (Itami and Craig 1989).

Finally, the production of seeds is also influenced by a wide range of insects. For example, boring moths (Dioryctria albovittella) were found to reduce viable seed production on infested P. edulis by ~96% (Mueller et al. 2005). This is reflective of the overall compounding influence insects and microbes have on persistence and regeneration.

The sum effect of herbivorous insects and microbial disease is to accelerate mortality when moisture and temperature conditions become unfavorable. This is particularly true for piñons, whose responsive transpiration strategy leads to carbon starvation under drought conditions, and consequently an inability to defend against herbivory. Population increases of phytophagous insects on overwhelmed trees further magnifies mortality when conditions become less than suitable, and scale up to cause distribution contractions. However, under severe drought, mortality predominantly reflects abiotic gradients (Floyd et al. 2009).

The integrated distribution

Distributions of piñons and junipers are primarily constrained by climatic gradients that are moderated by latitude and elevation. However, every aspect of an individual tree's life cycle is locked into coevolutionary relationships which magnify the effect of climate on the individual. This biotic amplification on the individual organism collectively scales up to become the emergent behavior of distributions. Specifically, the 13 mortality and dispersal dynamics of distributions as a whole arise from biotic interactions within the bounds of climatic conditions.

Climatic limitations represent thresholds of biotic interaction. The very water and nutrient balance of piñons and junipers is dependent on coevolved mycchorizae. Insects are the principle amplifying force in mortality and distribution contraction. Mortality may also be amplified by competition from co-occurring plants, but this competition more generally defines transitions to grassland, shrubland, and montane forest life zones. Co- occurring plants can also play a facilitative, structure-based role as nurse plants (shade of shrub or tree). Coevolved plants (mistletoe) are even integrated into the dispersal system which is highly coevolved with avian mutualists.

Distributions track ancestral conditions as an integrated assemblage of coevolved organisms. However, distributions of species behave in the context of their lineage and are influenced by con-generic competition and hybridization. Sister taxa share traits and coevolved relationships, facilitating their own distribution shifts and generating daughter species. Hybridizations among sister species and high variability in trait expression reflect active evolution in both piñons and junipers in western North America, but even the youngest of these species (P. monophylla, 5 mya) is much older than Quaternary glacial cycles (1.8 mya). Distributions of these species, therefore, are tracking climatic conditions they adapted to during the Miocene (~24 - 5 mya) through the unprecedented abruptness of climatic changes during the glacial cycles. 14 Holocene and current distribution of piñons and junipers

The distributions of piñons and serrate margin junipers during glacial conditions were generally displaced to the south where the present day Mojave and Sonoran deserts are found. They grew throughout the basins in mixed conifer communities with little contemporary analogue, including species now confined to high elevations, such as bristlecone and limber pines, and with species now occurring in the Mexican highlands, including some piñons (Betancourt, Van Devender, and Martin 1990). Packrat middens along the present day USA-Mexico borderlands indicate piñon-juniper woodlands were common during the last ice age and were replaced by desert scrub following the development of the Holocene interglacial conditions ~10,000 years ago (Holmgren et al.

2003). There are some places within each species range that have remained largely in place during both glacial and interglacial conditions. In the Sierra Nevada Mountains near Lake Tahoe, for example, J. occidentalis macrofossils dating into the last glacial period are found ~1,500 meters lower in elevation from where the species still occurs

(Nowak et al. 1994). Similarly, piñons and junipers were found throughout the southern

Colorado Plateau at lower elevations during the last glacial period (Anderson et al. 2000).

In general, distributions were 500 to 1,000 km farther to the south and 1,000- 1,500 meters lower in elevation (Miller and Wigand 1994, Jackson et al. 2005, Cole 1990).

Distribution shifts following the development of interglacial conditions were episodic (Jackson et al. 2005). Junipers typically expanded their ranges earlier than piñons following the retreat of the last ice age. This was in part due to their persistence in northerly latitudes. This transition occurred in northeastern Utah between 10,500 and 15 7,500 years ago. In this area piñons did not become a significant component until ~800

years ago (Betancourt et al. 1991). Similarly, in the northwestern Great Basin, J.

osteosperma has been present for at least 30,000 years, adjusting to climatic changes by

shifting elevation, whereas P. monophylla had much greater latitudinal adjustments and

did not reach its current northern limits until the last two thousand years (Spaulding

1990). The demographic structure of northern populations combined with the more recent

appearance of piñons in the fossil record indicates they are still expanding northward.

Currently, piñon pines and sabinoid junipers both inhabit arid climates and form a

coniferous woodland zone between grasslands/shrublands and coniferous montane forests

(Merriam 1898). On a regional scale, the relative composition of piñons and junipers are

structured by the amount and seasonality of precipitation. The juniper component

increases as the precipitation regime becomes dominated by winter snow and summer

rain becomes minimal, eventually giving way to shrublands. The piñon component increases as the precipitation regime becomes dominated by summer rains, and eventually gives way to grasslands. At both ends of the seasonality spectrum the trees become increasingly sparse and transition from woodlands to savannas (West et al.

2008).

The woodland zone is bounded between its neighboring zones by disturbance regimes. As moisture increases and temperature decreases the woodland zone gives way

to montane forests. Although it is common for young piñons and junipers to grow in the ecotone, the more productive montane forests have a fire return interval that precludes the persistence of piñons and junipers. Woodlands give way to shrublands as aridity 16 increases, and give way to grasslands as summer rain increases. Colonization of forests

and grasslands is primarily limited by high fire return intervals.

In summary, the western and eastern distribution boundaries of piñon and juniper

species exist where there is a transition to an annual moisture deficit. The southern

boundary runs along the annual minimum zero-degree isotherm. The northern boundary

is a transition between asynchronous wet and dry decadal-scale phases in combination

with the winter position of the polar air mass. Within this ~3,000,000 ha area a strong

seasonal precipitation gradient runs from the winter dominated north-western Great Basin

to the summer-monsoon of the south-eastern Colorado Plateau. The physiographic regions of western North America structure the regional climate gradients and their

unique topographies structure the spatial configuration of piñon-juniper woodlands. 17 INTRODUCTION

Piñon and juniper woodlands form an expansive vegetation zone throughout

North America’s arid interior west. Two piñons (Pinus edulis, P. monophylla) and four junipers (Juniperus deppeana var. deppeana, J. monosperma, J. occidentalis, J. osteosperma) are endemic to this region. They co-dominate the arid woodlands in a mosaic of overlapping distributions structured along climatic gradients (Cole et al. 2008a) driven by the seasonal positioning of regional air masses (Neilson 1987). Piñon and juniper populations have been highly dynamic over the last two centuries, during which time they have undergone an overall expansion into neighboring shrublands, grasslands and forests (Tausch, West, and Nabi 1981, Miller and Wigand 1994, Romme et al. 2009).

Accumulating demographic studies across the interior west indicate the drivers of this expansion vary regionally, but have an underlying climatic driver (Appendix A). The expansions into shrublands, primarily throughout the Great Basin, were compounded by a reoccupation of large areas that had been clear-cut for the mining industry in the late

1800s (Young and Budy 1979). Expansion into grasslands, primarily in Texas, has been accelerated by alteration to grazing regimes (Van Auken 2009). In the Colorado Plateau, however, it has become evident that piñon and juniper populations are responding primarily to climatic drivers, with changes in grazing regimes having localized effects, if any at all (Weisberg et al. 2008, Barger et al. 2009).

The 20th century climate of the Northern Hemisphere has changed at a rate

exceeding any other over the last two millennia (Mann and Jones 2003), and projected

21st century temperature increases are of a magnitude that has not occurred since the end 18 of the last ice age ~10,000 years ago (IPCC 2001). Mortality rates among all dominant

western forest tree species over the last few decades have increased to doubling periods

of less than 30 years (van Mantgem et al. 2009), and are expected to affect all current tree

species. Recent and projected climatic changes in North America's interior west are

among the most severe on the continent (Overpeck and Udall 2010). Differential

mortality rates among piñons and junipers during the last decade indicate distribution-

wide shifts in co-occurrence have already began in response to contemporary climate

change (Breshears et al. 2005, Breshears et al. 2009, McDowell et al. 2008).

Piñons and junipers are a model system for understanding climate change effects

and have received intensive study on physiological (Linton, Sperry, and Williams 1998,

West et al. 2008), and population levels (Jaindl, Eddleman, and Doescher 1995, Ogle,

Whitham, and Cobb 2000). Every phase in the life cycle of both piñons and junipers is intricately linked in coevolutionary relationships which have also received detailed study

(e.g. insect and microbial pests: Christensen, Whitham, and Klein 1995, Mueller et al.

2005, Del Vecchio et al. 1993, mycorrhizae: Haskins and Gehring 2004, Allen et al.

2010, Robinson et al. 2010, dispersal: Vander Wall and Balda 1977, Poddar and Lederer

1982, Lanner 1998). Additionally, the presence of packrat (Neotoma spp.) middens

across the interior west has provided a window into the response of piñon and juniper

distributions to past climatic changes reaching deep into the last ice age (Betancourt, Van

Devender, and Martin 1990, Lyford et al. 2003, Coats, Cole, and Mead 2008). The emergent picture indicates both establishment and mortality, though moderated by coevolutionary relationships, are filtered by climatic gradients (Martens, Breshears, and

Barnes 2001, Floyd et al. 2009). However, it is also clear that dispersal is a critical 19 element of piñon and juniper distribution dynamics, particularly for piñons which are

apparently still adjusting to interglacial conditions (Madsen and Rhode 1990, Betancourt

et al. 1991, Gray et al. 2006).

Given the over-arching constraint of climate on the historical and current

distributions of piñons and junipers, we investigated potential shifts in their individual

distributions and collective co-occurrence under forecast climate change scenarios of the

21st century. The response of woody plant distributions to forecast climate change has

been, and continues to be, investigated by numerous researchers (Iverson et al. 2008,

Zimmerman et al. 2009, Cole et al. 2008b, Refheldt, Ferguson, and Crookston 2008). The general approach is to empirically relate current species distributions to current climatic variables, producing an n-dimensional model of the classic realized niche (sensu

Hutchinson). Often referred to as species distribution models (SDMs, Guisan, Thuiller,

and Gotelli 2005) these are then applied to forecast future climate conditions.

This process is subject to legitimate criticism. SDMs are not equivalent to either

the fundamental or the realized niche (e.g., Pearson and Dawson 2003), representing

either potential or realized distributions at best (Soberón 2007). Environmental variables

define the dimensions of the SDMs and determine the limits on model interpretation

(Doorman 2007). Environmental variables are distal and can only attempt to reflect

underlying physiological mechanisms (e.g., Helmuth, Kingsolver, and Carrington 2005).

Biotic-interactions may exert important influences on a species geographic distribution,

but are difficult to incorporate into SDMs (e.g., Araújo and Luototo 2007, Elith and

Leathwick 2009). Forecasting SDMs is complicated further by differential life-stage

specific requirements (e.g. juvenile vs. adult, Jackson et al. 2009), by genetic variation 20 within each respective species (Grulke 2010), and by particular dispersal

characteristics (e.g., Ibanez et al. 2006, Thuiller et al. 2008). The appearance of no- analogue future climates (after Williams, Jackson, and Kutzbach 2007), for which no empirical relationships exist for modeling purposes, further confounds modeling attempts.

Many of the concerns surrounding the process of correlative distribution modeling

may be minimized by explicitly aligning the model to reflect life cycle dynamics within

the spatial and temporal resolution of the modeling variables, the modeling methods, and

model interpretation. We build on this basic idea to investigate the implications of non-

equilibrium distributions through a dispersal-based method for defining geographic

domains to limit model building and projection. We next address the implications of

potential hybridization and competition between sister species by modeling distributions

at two levels of taxonomic resolution. First, we develop models for each of the six

species. These six species occur in pairs of hybridizing sister species, which is the second

level of taxonomic resolution that is modeled.

We use the Hadley Center Coupled Atmosphere-Ocean Global Circulation Model

(HadCM3 GCM) as the basis of the climate projections (Aritt, Goering and Anderson

2000, Williams, Senior, and Mitchell 2001). Two climate change scenarios were applied,

following the Intergovernmental Panel on Climate Change (IPCC) story lines A2a and

B2a (Hijmans et al. 2005). Climate change scenarios were applied at three, 30 year time

steps over the 21st century to incorporate the influence of multi-decadal climate variation

around the century-scale change. We employ a procedure for modeling climate change

effects within the context of species-specific dispersal-defined spatial domains for the 21 entire geographic range of each taxon. For each taxon, probability of occurrence maps

were generated across its respective domain for each time step, reclassified as presence or

absence, and linked together with a simple expansion threshold. The resulting distribution maps depict, for each taxon, areas under pressure of expansion, contraction, or persistence under each climate change scenario. All thresholds were kept constant for all models to provide a consistent comparison. The implications of these pressures, i.e. the

manifestation of actual expanding or contracting populations in the landscape, must be

considered in context of the spatial and temporal resolution of variables and the life

histories unique to each species.

We gauge the magnitude and spatial patterning of climate-driven distribution

changes in terms of elevation, latitude, and overall area. Elevation shifts are simply

gauged as changes in mean, variation, minimum, and maximum. Latitudinal shifts are

gauged as changes in the centroid of the geographic distributions. Changes in areal

distributions are used to gauge potential shifts in co-occurrence on the species level. We

assess the potential role of hybridization through the geographic discordance between the

sister-species models and the respective species models.

We investigated numerous environmental predictor variables hypothesized as

related to the species distributions, and chose the simplest expressions of annual and

seasonal temperature and moisture in order to avoid problems associated with forecast climatic conditions having no contemporary analogue. We used Random Forests (RF,

Brieman 2001, Cutler et al. 2007), a commonly employed statistical tool for distribution forecasting (Lawler et al. 2006, Iverson et al. 2008, Refheldt, Ferguson, and Crookston

2009). Observations of occurrence were derived from the USDA Forest Service 22 Inventory and Analysis Data (FIA, McRoberts et al. 2005). Problems associated with modeling incomplete geographic distributions (see Thuiller et al. 2004, Randin et al.

2006) were avoided by focusing on species endemic to our study area.

23 METHODS

Study area

Our study area is in the mid-latitudes (31º N to 49º N) of North America’s Dry

Domain, an area approximately 3,000,000 km2 in size (Fig. 1c, 1d). The Dry Domain is

defined by an annual moisture deficit (Bailey 1989) resulting from a combination of the

mid-latitude Hadley cell circulation and the orographic configuration of western North

America. Sea surface temperatures of the Pacific and Atlantic oceans generally drive the

annual and decadal climatic variation within the Dry Domain (Cayan et al. 1998).

Fig. 1. Distribution of FIA observations for study piñons (a) and junipers (b) within their species-specific modeling domains. Distribution of all piñons (c) and section III sabinoid junipers (d) (Little 1978). 24 There are two important climatic divisions within the Dry Domain that

influence the overall northern and southern extent of our species distributions. The south

is generally bounded by the zero degree annual minimum temperature isotherm. The

north generally corresponds to a transition of asynchronous decadal-scale wet and dry

phases driven by oceanic circulation (Jackson et al. 2005), and runs east-west between

40ºN and 45ºN. This boundary is also the extent of the polar air mass during winter months (Mitchell 1976). A strong seasonal precipitation gradient runs from the winter dominated north-western Great Basin to the summer-monsoon of the south-eastern

Colorado Plateau (Rodwell and Hoskins 2001).

Within the landscape, piñons and junipers are limited in low elevations by aridity

in the winter-dominated regimes, where they give way to shrubs. In the summer-

dominated regimes they are out-competed by grass and limited by higher fire-return intervals. The upper elevations also correspond to a higher fire frequency that occurs in the more productive montane forests as conditions become more mesic.

Physiographic provinces within the Dry Domain structure regional climate and also have characteristic topographies influencing the spatial configuration of woodlands.

The Great Basin is composed of hundreds of narrow basins and ranges trending north- south. Woodlands grow in the foothills from where they extend through gullies into the basin shrublands and reach into the higher elevation forest along exposed mountain slopes. The Colorado-Plateau is composed of numerous plateaus at various elevations dissected by deep canyons. Here, expansive woodlands cover middle elevation plateaus and extend into the higher and lower elevation plateaus through deep canyons. The 25 woodland zone along the interior Coastal Range, Wasatch Mountains and Southern

Rockies is generally in the foothills like that of the smaller Great Basin ranges.

Study Species

We focus on two piñons and four junipers that occur exclusively in our study area

(Figs. 1a, 1b). Both are the northern-most reaching members of their respective groups

(Figs. 1c, 1d) who have undergone speciation in western North America during the

Neogene uplift ~15- 3 mya (Neogene uplift described in Wilson and Pitts 2010).

Delineation of species within these groups is highly problematic; there is widespread

hybridization and genetic classifications are at odds with morphological classifications

(Gernandt, Liston, and Piñero 2003, Adams et al. 2007). We developed models for each

species individually, and for the three sister-species groups where hybridization occurs:

P. edulis- P. monophylla; J. occidentalis- J. osteosperma; and J. deppeana- J.

monosperma.

Piñon pines, as a group, are endemic to North America. Fossils indicate this group

originated in Mexico during the Laramide orogeny ~50 mya and radiated during the

Neogene orogeny ~10 mya (Malusa 1992). Two species, P. edulis and P. monophylla, occur entirely in the midlatitudes and are the northern most extension of this group. P. monophyla is the most derived member, whose monophyllous trait is unique among all

pines and arose ~5 mya in association with the orogenesis of the Sierra Nevada

Mountains and consequent development of winter-dominated precipitation in the Great

Basin (Malusa 1992). These two piñons hybridize wherever they overlap (Lanner 1974), but inhabit distinct precipitation regimes (Cole et al. 2008a). P. edulis requires the 26 summer rains that currently characterize the Colorado Plateau, whereas P. monophylla

inhabits the winter dominated precipitation regime of the Great Basin.

The junipers modeled all belong to the serrate-margin junipers (Adams 2008).

This group is distinguished by serrated leaf margins, thought to be an adaptation to arid

conditions, and their overall phylogeny reflects adaptation to drought-resistance (Wilson,

Manos, and Jackson 2008). This group originated in Eurasia ~45 mya and radiated in

western North America during the Neogene uplift ~10 mya (Mao et al. 2010). J. occidentalis has two varieties; however, recent genetic analysis indicates J. occidentalis var. australis is a separate species from J. occidentalis var. occidentalis, which is more closely related to J. osteosperma (Adams 2008). Both varieties of J. occidentalis hybridize with J. osteosperma, facilitated by a shared timing of pollen shed in the late

spring. We model J. occidentalis sensu latu. The chloroplast genome of J. occidentalis,

associated with a foliar gland, is transmitted by pollen and found to be widely dispersed

into western populations of J. osteosperma. Together they are found in varying levels of

introgression across the entire Great Basin, with the likely implication being a more

widespread distribution of J. occidentalis during the glacial conditions of the Pleistocene

(Terry, Nowak, and Tausch 2000). Where they occur together, J. occidentalis occurs at

higher elevations.

There are numerous allopatric varieties of J. deppeana which extend from

Mexico, and are thought to have formed a contiguous distribution during the Pleistocene

(Adams et al. 2007). We limit our analysis to J. deppeana var. deppeana, which occurs

entirely within the midlatitudes. J. monosperma is also a member of closely related

varieties who hybridize widely, but which extend into Texas (Adams et al. 2007). Of this 27 clade, we limit our analysis to J. monosperma which occurs in the Colorado Plateau

(Adams 2008). J. deppeana and J. monosperma occur together throughout most of their

range, with J. deppeanna growing in higher elevations. Although they both are members

of different hybridizing groups, the phylogenetic divergence is of similar depth to J.

occidentalis and J. osteosperma (see Mao et al. 2010). Hybridization between J. deppeana and J. monosperma has been reported in a cultivated setting (Hall 1961).

Hybridization may potentially occur under wild conditions as they both shed pollen in the late winter/early spring (c.f. Adams 2008).

Dispersal

Junipers and piñons both have seeds adapted to animal dispersal, although with

markedly different strategies. The piñon nut is relatively large and composed of rich fats

and carbohydrates (Vander Wall 1988). Piñon nuts are a primary food source for a

number of corvid species (Lanner 1998) as well as some human cultures, and both are

documented long distance dispersers of piñons. Corvids stash the nuts purposefully

throughout the landscape according to their winter feeding habits. The piñon jay,

Gymnorhinus cyanocepjalus, is most closely tied with the piñon pine, to such an extent

that mast seed crops can actually trigger piñon jays to reproduce in the fall (Ligon 1974,

1978). Piñon jays inhabit the woodlands year round and distribute seeds within the local

woodlands (<10km) near their colonial breeding site (Marzluff and Balda 1992). The

Clark’s nutcracker, Nucifraga Columbiana, is likely the most important long-distance

dispersal agent (up to 22 km, Vander Wall and Balda 1977) because it lives in the 28 montane coniferous forests, harvesting the nuts from the woodlands but caching them in the higher elevation forests (Vander Wall 1988).

The juniper seed is encased in a fleshy berry-like cone which a wide array of mammals and birds disperse. Mammals include rodents, lagomorphs, foxes and coyotes

(Schupp et al. 1997; Chambers, Vander Wall, and Schupp, 1999). Coyotes and foxes may play an important role in long distance dispersal (Schupp et al. 1997). In the Interior West there are at least fifteen species of over-wintering avian furgivores that rely on juniper berries to some extent, with the thrushes (Turdidae) and waxwings (Bombycillidae) the most important (Chavez-Ramirez and Slack 1994). Townsend's solitaires Myadestes townsendi, in particular, rely on juniper berries for their winter diet nearly exclusively

(Poddar and Lederer 1982). Juniper berries are consumed whole by avian frugivores resulting in a spatial deposition pattern reflecting the daily habits of the over-wintering frugivores (Salomonson 1978, Chavez-Rameriz and Slack 1994). Transport by surface runoff is also likely to be an important dispersal vector for colonization of lower elevations (Chambers, Vander Wall, and Schupp, 1999).

Dispersal exhibits paradoxical, scale-dependent behavior (Clark 1998). The furthest observed dispersal distance for P. edulis is 22 km (Vander Wall and Balda 1977) but isolated northern populations of P. edulis are inferred to have dispersed from 40 km

(Gray et al. 2006) to 200 km (Betancourt et al. 1991). Similarly, observed distances for juniper dispersal by thrushes and waxwings are reported in hundreds of meters (Chavez-

Rameriz and Slack 1994), whereas isolated northern populations of J. osteosperma are estimated to have dispersed 135 km (Lyford 2003). Dispersal is predominantly localized and regeneration is typically pulsed on decadal time spans, influencing population 29 distributions within a landscape. Yet for time spans of centuries to millennia the rare long-distance dispersal events appear to drive overall continental distributions.

We compiled published accounts of dispersal distances (Table 1) for our species to aid in two objectives. First, to define a modeling domain for the entire distribution that, for each species, better reflects suitable and unsuitable conditions rather than the idiosyncrasies of historical biogeography (Soberon and Nakamura 2009). The second objective was to determine an informative dispersal rate to parameterize movement of forecast distributions.

Table 1. Observed and inferred dispersal distances of piñon and juniper species

Distance Species (km) Vector Reference

P. edulis 200 unknown Bentancourt et al.1991 P. edulis 40 unknown Gray et al. 2006 P. edulis 22 Nucifraga columbiana Vander Wall and Balda 1977 P. edulis 12 Gymnorhinus cyanocephala Balda Kamil 1998 P. monophylla 10 Nucifraga columbiana Vander Wall 1981

P. monophylla 38.9 scatter-hoarding rodents Vander Wall 1997 J. osteosperma 135 unknown Lyford et al. 2003 J. ashii .044 Turdus migratorius Chavez-Ramirez and Slack 1994 J. ashii .012 Bombycilla cedrorum Chavez-Ramirez and Slack 1994

30 Modeling domains

A modeling domain was generated for each of the six species and for each of the three sister-species group by buffering all respective presences by a dispersal-based distance (Fig. 2). Using the entire geographic range of a species allows for a more complete estimation of the realized distribution, which has important consequences in forecasting (Barbet-Massin, Thuiller, and Jiguet 2010). Limiting the modeling domain to areas within dispersal range of the respective species reduces uncertainty associated with the degree to which each distribution is at equilibrium with the current climate (e.g.

Guisan, Thuiller, and Gotelli 2005), but introduces the risk of not capturing the entire tolerance-response curve of the species (Thuiller et al. 2004). In principle, the modeling domain is a balance between being small enough to experience frequent dispersal, but large enough to encompass the majority of long distance dispersal events which have occurred over the last several centuries. We compared the range of conditions inhabited by each species to that covered by their respective modeling domain (Appendix B). With this consideration we choose a distance of 100 km to buffer present distributions to become the species- specific modeling domains. This same procedure was performed for the sister-species groups.

The modeling domains define the sampling frame to develop initial models and to bound subsequent model forecasts. The distribution of presences for all study species are contained by the mid-latitude Dry Domain, but for J. occidentalis and P. monophylla are 31 in places less than 100 km from the western boundary. The modeling domain of these species was restricted to the Dry Domain. Similarly, presences for the other four species

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Figure 2. Modeling domains and modeled current distributions for species and sister- species. Distribution models are shown for the whole mid-latitude Dry Domain, but subsequent analysis is limited to the modeling domains. J. deppeana var. deppeana.(a); J. monosperma (b); J. deppeana var. deppeana - J. monosperma (c); J. occidentalis (d); J. 32 osteosperma (e); J. occidentalis - J. osteosperma (f); P. edulis (g); P. monophylla (h); P. edulis - P. monophylla (i).

were in places less than 100 km from Mexico. In these instances, the modeling domains

were limited to the U.S. to which the FIA is confined. This decision was made with

support from distribution maps which indicate the study species, or variety in the case of

J. deppeana var. deppeana, are entirely endemic to the U.S. (see Adams 2008 for junipers, Cole et al. 2008a for piñons).

Data Structure

Presence and absence for each study species came from the United States Forest

Service Inventory and Analysis program (FIA). The FIA recently established locations

for permanent plots with a consistent spatial intensity (McRoberts et al. 2005). However,

the spatial intensity of FIA plots currently varies geographically. To ensure a probabilistic sample we took a random subset of one plot per 10 km2, which is the lowest

sampling intensity in our area. We collapsed plot observations to either presence or

absence for each of our species. True plot coordinates were used throughout all model

building.

Observations of presence were the basis for generating the 100 km buffer

modeling domain for each species. Within these sample frames we randomly selected, for

each species, a number of absences equal to the number of presences (Table 2) in order to

simplify issues of prevalence (Manel, Williams, and Ormerod 2001). To ensure this subset of absences was representative of climatic conditions within the modeling domain we drew 101 random samples. The first 100 samples were used to generate a distribution 33 of sample variation against which the last sample was tested before becoming the

training absences. The range of our species is contained by the area covered by the FIA,

and so allows for a geographically complete model to be built (after Guisan, Thuiller, and

Gotelli, 2005, Randin et al. 2006).

Table 2. Prevalence of six piñon and juniper species from FIA observations within modeling domains.

JUDE JUMO JUOC JUOS PIED PIMO All FIA plots within species 100 km dispersal envelope A 7,557 10,110 8,660 23,688 13,546 10,618 766 1,377 691 3,444 2,449 1,216 P (9%) (12%) (7%) (13%) (15%) (11%) Random selection of one FIA plot/ 10km² A 7,410 11,544 3,913 24,823 13,924 10,546 729 1,440 515 3,341 29,66 1,172 P (8.9%) (11.1%) (12.6%) (12.8%) (17.6%) (10%) JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla A = species absent; P = species presence

Predictor variables

Although detailed physiological (e.g., West et al. 2008) and germination studies

(e.g., Chambers 2001) of piñons and junipers have determined thresholds which may eventually parameterize mechanistic models (e.g., Chown et al. 2010) of mortality and

colonization, the resolution of available environmental variables currently precludes this

approach. Instead, we follow the approach common to distribution modeling efforts

which rely on the empirical relationship of contemporary distributions to relevant 34 environmental variables.Environmental variables were chosen to correspond with the known variables that constrain the life cycle of our study species and thereby structure and limit their distribution (Table 3, Appendix B). The resolution of all predictor variables is 1 km2, which has been found to adequately portray distributions in the size

range of our study species (Seo et al. 2008). At this resolution populations of piñons and

junipers generally have both old and young trees (Tausch, West, and Nabi 1981).

Seasonal and annual climate statistics were generated from monthly statistics

acquired from WorldClim (http://www.worldclim.org/, Hijmans et al. 2005). Seasonal

variables included degree sums, temperature extremes, and precipitation sums. Degree

sums were calculated for both above and below freezing temperatures by adding all

maximum temperatures above freezing and all temperatures below freezing, respectively.

Annual variables included temperature extremes and variation, heat sums above and

below freezing, precipitation totals above and below freezing, and precipitation

seasonality.

Topographic variables were all generated from a digital elevation model also

acquired from WorldClim. Slope was calculated in degrees. Aspect was transformed to

range from 1º at due north to 0º at due south. Finally, a topographic scalar was generated

through a focal analysis which assigned to each cell the standard deviation of elevation

values in a 3x3 neighborhood.

We choose the Hadley Center GCM for its ability to capture the regional

dynamics of midlatitude climates, particularly the North American monsoon (Kim,

Wang, and Ding 2008). The Intergovernmental Panel on Climate Change (IPCC)

scenarios A2a and B2a model runs, obtained from WorldClim, were applied to the 35 distribution models. The A2a scenario is a high emissions scenario with the most severe climatic changes. The B2a scenario is a lower emissions scenario with comparatively moderate climatic changes. Each forecast is applied at a thirty year time step and represents the averaged conditions during each time step.

Table 3. Predictor variables used to model piñon and juniper distributions within their respective domains. Data sources are described in text.

Variable class / Code Description Units

Climate sum of monthly maximum temperatures DGRS_AFMX °C×10 above zero sum of monthly minimum temperatures DGRS_BFMN °C×10 below zero sum of maximum temperatures for mar, apr, DGRS_SPRG °C×10 may sum of maximum temperatures for sep, oct, DGRS_FALL °C×10 nov PREC_WNTR sum of precipitation for dec, jan, feb mm PREC_SPRG sum of precipitation for mar, apr, may mm PREC_SUMR sum of precipitation for jun, jul, aug mm PREC_FALL sum of precipitation for sep, oct, nov mm TMAX_ANUL annual maximum temperature °C×10 TMIN_ANUL annual minimum temperature °C×10 TAVE_TRNG annual range in temperature °C×10 Topography ZTASP Transformed aspect derived from ZAVE 0=south, 1=north ZSLOPE Derived from ZAVE percent Standard deviation from 3×3 km ZSTD n/a neighborhood

36 Distribution model

Distribution models were built using Random Forests (RF) (Brieman 2001, Cutler et al. 2007). Ecological interpretations of the models are based on variable importance plots, which report the relative importance of each variable in classification accuracy and class purity. Model accuracies were evaluated in terms of percent correctly classified, sensitivity, specificity, kappa (Landis and Koch 1977), and area under the receiver curve

(AUC, Fielding and Bell 1997), all generated through a ten-fold cross-validation. Models were generated in R (R Development Core Team 2008) through the randomForest package (Liaw and Wiener 2002) and applied to predictor variable grids with yaImpute

(Crookston and Finley 2007), yielding a probability of occurrence surface for each taxon

at each time step by scenario.

Three probability thresholds, in conjunction with dispersal parameters, were used

to generate forecast distributions for each species (Table 4). The first threshold of 50%

defines presence and absence for the initial distribution maps, and is based on the

prevalence of presence observations in the training data (Real et al. 2006). The second

and third thresholds define subsequent transitions to absence or presence, respectively,

and are based on the concept of the juvenile niche being nested within a broader adult

niche (e.g., Jackson et al. 2009). Although these latter two thresholds correspond to

mortality and colonization, they are not absolute predictions of such at the resolution of

the individual tree. Rather, they are considered indicative benchmarks on the spectrum of conditions which populations of piñons and junipers may exhibit within a given square kilometer. With this precaution, presence transitioned to absence for subsequent 37 probabilities less than 40% and absence transitioned to presence for subsequent

probabilities equal to or greater than 60% within the expansion constraints explained

below.

The modeled rate of expansion into newly suitable areas is balanced between

observed and inferred dispersal distances and is weighted toward the observed (Table 1).

We chose a dispersal rate of 30 km/30 year time step. The life cycle of piñons and junipers are nested within this time frame as they are known to become reproductively active within this period (Lanner 1998, Adams 2008). Additional support for this dispersal rate comes from a study of J. occidentalis colonization which found that stands without old trees were established an average of 24 years later than those with a local

seed source provided by old trees (Johnson and Miller 2006). Although such observations

of dispersal are used to inform this modeled dispersal rate, this approach simplifies the very complex and idiosyncratic behavior of dispersal as an initial step to examine distribution shifts on the continental scale. The expansion rate is held constant for all species to provide a consistent measure of the magnitude of change that is forecast for each distribution.

Table 4. Thresholds for modeled presence and absence Time step probability spatial condition (for each species) initial distribution = 1 ≥ 50% within modeling domain i i+1 1 1 ≥ 40% N/A 1 0 < 40% N/A 0 1 ≥ 60% within dispersal threshold 38 RESULTS

Model performance

The modeling domains and sampling techniques were found to have climatic

ranges encompassing the conditions of their respective species (Appendix C). Variable

importance plots indicate all seasonal precipitation variables are important in all the

models except J. deppeana var. deppeana, which is less sensitive to spring precipitation

(Appendix D). In contrast, seasonal degree sums were generally less important, with

spring being the most consistently important variable across all models.

Accuracy metrics for the species and sister-species models are reported in Table

5. The percent correctly classified (PCC) for species models ranged from 93% for J. deppeana to 80% for P. edulis with the other four models near the overall average of

85%. Errors of omission for species models were consistently less than errors of commission, with an overall average sensitivity of 89% and specificity of 80%. The kappa statistic was lowest for J. deppeana var. deppeana (0.58) which otherwise had the highest accuracies, and was highest for J. osteosperma (0.71). Kappa averaged 0.65 for all models. The AUC constituted our only threshold independent metric and had an average of 0.91 for all models. Accuracies metrics for sister-species were similar to the lowest accuracies of their component species except for J. deppeana-J. monosperma, which was roughly 5% higher.

Although each species and sister-species model was analyzed within its respective modeling domain, it is interesting to note the predicted distribution of current distributions applied across the entire mid-latitude Dry Domain (Fig. 2). For each species, 39 some likelihood of presence is predicted to occur outside the respective modeling domain and current, known distribution. In all cases a closely related species is currently present. For example, J, monosperma is predicted to occur in the southwestern Great

Basin where J. occidentalis and other sabinoid junipers occur (Adams 2008). In turn, J. occidentalis has predicted suitable habitat co-mingled with the distributions of J. monosperma and the other study junipers. Both P. edulis and P. monophylla are predicted to extend into each others distribution along their hybridization zone. P. monophylla has extensive suitable conditions extending into the northwestern Great Basin, which is in agreement with the general observation that the distribution of this species is still adjusting to interglacial conditions. In all these cases the dispersal-based modeling domain limited the overestimation of current distributions through the exclusion of distant, but possibly suitable areas.

Elevation shifts in distributions

Elevation shifts of species groups are generally less than their component species

(Table 6, 7). Average elevation increases for all species under the a2 scenario is 350 m, with J. osteosperma shifting the least at 143 m, and J .occidentalis the most at 1,023 m.

Average elevation increases under the b2 scenario average 300 m, with P. monophylla shifting the least at 139 m, and J. occidentalis the most at 895 m. The overall standard deviation of elevations increases an average of 2 m - 17m (a2, b2), but varies from J. occidentalis with the greatest increase of 54m - 92m to a decrease of 28 m for J. deppeana var. deppeana. The average minimum elevation is shifting faster (288 m, 363 m) than the average maximum elevation (248 m, 360 m). 40 Table 5. 10-fold cross validation accuracy metrics for Random Forest models predicting distributions of six piñons and junipers, and their three sister groups.

Model PCC Specificity Sensitivity kappa AUC

Single species JUDE 92.73 90.26 95.2 0.58 0.97 JUMO 83.67 79.74 87.58 0.67 0.9 JUOC 82.32 76.26 88.38 0.65 0.9 JUOS 85.71 83.09 88.33 0.71 0.92 PIED 80.13 75.7 84.56 0.6 0.87 PIMO 83.66 79.74 87.58 0.67 0.9 Sister species JUDE-JUMO 78.01 74.83 81.19 0.56 0.85 JUOC-JUOS 82.26 79.23 85.3 0.65 0.89 PIED-PIMO 80.3 76 84.7 0.61 0.87 JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma; PIED = P. edulis; PIMO = P. monophylla; JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla.

Table 6. Elevation shifts of modeled A2a forecasts (m.a.s.)

Mean (Standard Deviation) Minimum Maximum Model current 2080 change current 2080 change current 2080 change JUDE 2001 (317) 2325 (288) 323 (-28) 1220 1399 179 2908 3421 513 JUMO 1927 (313) 2142 (312) 215 (-1) 696 1154 458 3134 3385 251 JUOC 1626 (486) 2648 (540) 1022 (54) 518 961 443 3345 3916 571 JUOS 1922 (276) 2065 (282) 143 (6) 733 1107 374 2880 3320 440 PIED 2060 (269) 2290 (262) 230 (-7) 1053 1391 338 3047 3157 110 PIMO 1942 (329) 2091 (322) 149 (-7) 720 1104 384 3197 3472 275 JUOC-JUOS 1966 (305) 2175 (335) 209 (30) 709 1022 313 3150 3117 -33 JUOC- JUOS 1805 (370) 2007 (326) 202 (-44) 226 519 293 2869 3284 415 PIED- PIMO 2040 (285) 2261 (299) 221 (14) 720 1285 565 3153 3470 317 Species Mean 1913 (331) 2260 (334) 347 (3) 823 1186 363 3085 3445 360 Sister-groups 1937 (320) 2148 (320) 210 (0) 552 942 390 3057 3290 233 Mean JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla; JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla.

41

Table 7. Elevation shifts of modeled B2a forecasts (m.a.s.)

Mean (Standard Deviation) Minimum Maximum Model current 2080 change current 2080 change current 2080 change JUDE 2001 (317) 2249 (280) 248 (-37) 1220 1522 302 2908 3154 246 JUMO 1927 (313) 2089 (162) 162 (2) 696 865 169 3134 3357 223 JUOC 1626 (486) 2521 (578) 895 (92) 518 804 286 3345 3788 443 JUOS 1922 (276) 2064 (300) 142 (24) 733 1032 299 2880 3151 271 PIED 2060 (269) 2261 (270) 201 (1) 1053 1415 362 3047 3133 86 PIMO 1942 (329) 2081 (351) 139 (22) 720 1031 311 3197 3418 221 JUOC-JUOS 1966 (305) 2131 (340) 164 (35) 709 748 39 3150 3078 -72 JUOC- JUOS 1805 (370) 2020 (330) 215 (40) 226 419 195 2869 3239 370 PIED- PIMO 2040 (285) 2246 (305) 206 (20) 720 1059 339 3153 3468 315 Species Mean 1913 (331) 2211 (349) 298 (17) 823 1112 288 3085 3334 248 Sister-groups 1937 (320) 2132 (325) 195 (5) 552 742 190 3057 3262 205 Mean JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla; JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla. 42

43 Areal shits in individual distributions and their co-occurrence

Final distribution maps for each species and species group are shown in Fig. 3

(A2a scenario) and Fig.4 (B2a scenario). Geographic discordance between species and

sister-species models is shown in Fig. 5. The greatest difference between species models

and sister species models are for J. occidentalis and J. osteosperma, where the sister species model predicts greater occurrence in the northwestern Great Basin while the species model predicts greater occurrence in the Sierra Nevada Mountains. Compared to individual species forecasts, species-groups are forecast to lose less area and gain more area. Individual species are forecast to lose an average of 62% to 52% of their modeled current areas for the A2a and B2a scenarios, respectively, while only gaining 14% for both scenarios (Tables 8 - 11). This trend is carried into relative areal shifts of co- occurrence which all exhibit greater loss than gain under both scenarios (Tables 12 - 15,

Fig. 7). The areal shifts are greatest for the Great Basin species, J. occidentalis and P. monophylla. Modeled areal loss is greatest for J. occidentalis which, for both scenarios, is forecast to lose 90% of its current distribution, remaining only in the high Sierra

Nevada mountains and gaining an area of only 3% relative to its current distribution. J. osteosperma, which spans the Great Basin and Colorado Plateau, has the least relative areal loss 31% (B2a) and 39% (A2a) while having the second highest relative areal gain of 21% (B2a) and 19% (A2a). J. deppeana var. deppeana has the highest relative areal gain of 25% (B2a) and 21% (A2a), and the second lowest loss of 43% (B2a) and 48%

(A2a). The other two Colorado Plateau species, P. edulis and J. monosperma, also lose

50% of their current modeled areas while gaining 15%.

44

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Figure 3. Forecast distributions for A2a scenario. J. deppeana var. deppeana.(a); J. monosperma (b); J. deppeana var. deppeana - J. monosperma (c); J. occidentalis (d); J. osteosperma (e); J. occidentalis - J. osteosperma (f); P. edulis (g); P. monophylla (h); P. edulis - P. monophylla (i). The color ramp indicates presence (1) and absence (0) over the four modeled time steps. For example, the far left column, in blue, represents areas that were modeled as suitable for each time step. The far right column, in red, represents areas that were modeled as suitable in the first time step but not in any of the subsequent ones.

45

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Figure 4. Forecast distributions for B2a scenario. J. deppeana var. deppeana.(a); J. monosperma (b); J. deppeana var. deppeana - J. monosperma (c); J. occidentalis (d); J. osteosperma (e); J. occidentalis - J. osteosperma (f); P. edulis (g); P. monophylla (h); P. edulis - P. monophylla (i). The color ramp indicates presence (1) and absence (0) over the four modeled time steps. For example, the far left column, in blue, represents areas that were modeled as suitable for each time step. The far right column, in red, represents areas that were modeled as suitable in the first time step but not in any of the subsequent ones.

46 Current 2080 A2a 2080 B2a

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Figure 5. Comparison of sister-species models to species models for current and forecast 2080 distributions. Current J. deppeana var. deppeana - J. monosperma (a); forecast A2a J. deppeana var. deppeana - J. monosperma (b); forecast B2a J. deppeana var. deppeana - J. monosperma (c); current J. occidentalis - J. osteosperma (d); forecast A2a J. occidentalis - J. osteosperma (e); forecast B2a J. occidentalis - J. osteosperma (f); current P. edulis - P. monophylla (g); forecast A2a P. edulis - P. monophylla (h); forecast B2a P. edulis - P. monophylla (i).

47 Table 8. Absolute areal shifts of modeled species A2a forecasts (km² (%)) JUDE JUMO JUOC JUOS PIED PIMO Absolute areal loss 59,129 54,837 36,784 57,685 14,377 0 JUDE (48.6) (66.5) (73.9) (68.6) (88) 21,1908 87,349 16,3615 20,762 0 JUMO (65.9) (63.6) (77.6) (88.2) 205,617 12,285 14,664 0 JUOC (92.4) (99.9) (96.8) 229,160 148,781 125,827

JUOS (38.6) (59.1) (70) 234,313 31,857

PIED (59.9) (75.6) 152,739

PIMO (67) Absolute areal gain 30,865 21,476 11,821 14,663 3,415 JUDE (25.4) (26) 0 (23.8) (17.4) (20.9) 44,483 26,669 29,938 5,160 JUMO (13.8) 0 (19.4) (14.2) (21.9) 6,426 9 3,262 JUOC (2.9) (0.1) 0 (21.5) 113,445 45,223 11,928 JUOS (19.1) (18) (6.6) 54,341 5,279 PIED (13.9) (12.5) 18,924 PIMO (8.3) Absolute equivalence 62,589 27,619 12,978 26,424 1,958 JUDE (51.4) (33.5) 0 (26.1) (31.4) (12) 109,816 49,982 47,100 2,783 JUMO (34.1) 0 (36.4) (22.4) (11.8) 16,987 480 JUOC (7.6) 1 (0) 0 (3.2) 365,267 102,876 53,882 JUOS (61.4) (40.9) (30) 157,003 10,285 PIED (40.1) (24.4) 75,386 PIMO (33) JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla.

48 Table 9. Absolute areal shifts of modeled species B2a forecasts (km² (%)) JUDE JUMO JUOC JUOS PIED PIMO Absolute areal loss 52,231 44,440 28,099 40,868 11,205 JUDE (42.9) (53.9) 0 (56.5) (48.6) (68.6) 149,307 57,831 124,871 14,991 JUMO (46.4) 0 (42.1) (59.3) (63.7) 197,223 12,247 13,830 JUOC (89.6) (99.7) 0 (91.7) 186,040 120,933 115,539 JUOS (31.3) (48.1) (64.3) 183,287 27,887 PIED (46.8) (66.2) 133,193 PIMO (59.6) Absolute areal gain 25,672 25,994 22,050 19,100 4,409 JUDE (21.1) (31.5) 0 (44.3) (22.7) (27) 53,834 42,222 41,697 7,725 JUMO (16.7) 0 (30.7) (19.8) (32.8) 6,424 143 5,680 JUOC (2.9) (1.2) 0 (37.7) 127,285 62,792 11,005 JUOS (21.4) (25) (6.1) 62,932 6,948 PIED (16.1) (16.5) 17,925 PIMO (8) Absolute equivalence 69,487 38,016 21,663 43,241 5,130 JUDE (57.1) (46.1) 0 (43.5) (51.4) (31.4) 172,417 79,500 8,844 8,554 JUMO (53.6) 0 (57.9) (40.7) (36.3) 22809 39 1,253 JUOC (10.4) (0.3) 0 (8.3) 408,387 130,724 64,170 JUOS (68.7) (51.9) (35.7) 208,029 14,255 PIED (53.2) (33.8) 90,419 PIMO (40.4) JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla.

49

Table 10. Absolute areal shifts of modeled sister-species A2a forecasts (km² (%)) JUDE- JUMO JUOC- JUOS PIED- PIMO Absolute areal loss JUDE- JUMO 141,211 (51.8) 61,850 (52.7) 137,587 (69.9) JUOC- JUOS 294,134 (42.7) 200,647 (54.8) PIED- PIMO 311,831 (57.6) Absolute areal gain JUDE- JUMO 55,105 (20.2) 27,450 (23.4) 44,209 (22.5) JUOC- JUOS 127,060 (18.4) 51,667 (14.1) PIED- PIMO 76,265 (14.1) Absolute areal equivalance JUDE- JUMO 131,633 (48.2) 55,612 (47.3) 59,295 (30.1) JUOC- JUOS 394,760 (57.3) 165,636 (45.2) PIED- PIMO 229,398 (42.4) JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla.

Table 11. Absolute areal shifts of modeled sister-species B2a forecasts (km² (%)) JUDE- JUMO JUOC- JUOS PIED- PIMO Absolute areal loss JUDE- JUMO 141,211 (51.8) 49,909 (42.5) 137,587 (69.9) JUOC- JUOS 286,012 (41.5) 202,172 (55.2) PIED- PIMO 311,831 (57.6) Absolute areal gain JUDE- JUMO 55105 (20.2) 31,510 (26.8) 44,209 (22.5) JUOC- JUOS 110,293 (16) 164,111 (44.8) PIED- PIMO 76,265 (14.1) Absolute areal equivalence JUDE- JUMO 131,633 (48.2) 67,553 (57.5) 59,295 (30.1) JUOC- JUOS 402,882 (58.5) 164,111 (44.8) PIED- PIMO 229,398 (42.4) JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla.

50

Table 12. Relative areal shifts of modeled species A2a forecasts (%) JUDE JUMO JUOC JUOS PIED PIMO Relative areal loss JUDE 48.6 45.1 0.0 30.2 47.4 11.8 JUMO 17.0 65.9 0.0 27.2 50.9 6.5 JUOC 0.0 0.0 92.4 5.5 0.0 6.6 JUOS 6.2 14.7 2.1 38.6 25.0 21.2 PIED 14.7 41.8 0.0 38.0 59.9 8.1 PIMO 6.3 9.1 6.4 55.2 14.0 67.0 Relative areal gain JUDE 25.4 17.6 0.0 9.7 12.0 2.8 JUMO 6.7 13.8 0.0 8.3 9.3 1.6 JUOC 0.0 0.0 2.9 0.0 0.0 1.5 JUOS 2.0 4.5 0.0 19.1 7.6 2.0 PIED 3.7 7.7 0.0 11.6 13.9 1.3 PIMO 1.5 2.3 1.4 5.2 2.3 8.3 Relative areal equivalence JUDE 51.4 22.7 0.0 10.7 21.7 1.6 JUMO 8.6 34.1 0.0 15.5 14.6 0.9 JUOC 0.0 0.0 7.6 0.0 0.0 0.2 JUOS 2.2 8.4 0.0 61.4 17.3 9.1 PIED 6.8 12.0 0.0 26.3 40.1 2.6 PIMO 0.9 1.2 0.2 23.6 4.5 33.0 JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla

51 Table 13. Relative areal shift of modeled species B2a forecast to current distributions (%) JUDE JUMO JUOC JUOS PIED PIMO Relative areal loss JUDE 42.9 36.5 0.0 23.1 33.6 9.2 JUMO 13.8 46.4 0.0 18.0 38.8 4.7 JUOC 0.0 0.0 89.6 5.6 0.0 6.3 JUOS 4.7 9.7 2.1 31.3 20.3 19.4 PIED 10.4 31.9 0.0 30.9 46.8 7.1 PIMO 5.0 6.7 6.2 51.7 12.5 59.6 Relative areal gain JUDE 21.1 21.4 0.0 18.1 15.7 3.6 JUMO 8.1 16.7 0.0 13.1 13.0 2.4 JUOC 0.0 0.0 2.9 0.1 0.0 2.6 JUOS 3.7 7.1 0.0 21.4 10.6 1.9 PIED 4.9 10.7 0.0 16.0 16.1 1.8 PIMO 2.0 3.5 2.5 4.9 3.1 8.0 Relative persistence JUDE 57.1 31.2 0.0 17.8 35.5 4.2 JUMO 11.8 53.6 0.0 24.7 26.7 2.7 JUOC 0.0 0.0 10.4 0.0 0.0 0.6 JUOS 3.6 13.4 0.0 68.7 22.0 10.8 PIED 11.1 21.9 0.0 33.4 53.2 3.6 PIMO 2.3 3.8 0.6 28.7 6.4 40.4 JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla

52 Table 14. Relative areal shifts of modeled sister-species A2a forecasts (%) JUDE- JUMO JUOC- JUOS PIED- PIMO Relative areal loss JUDE- JUMO 51.76 22.67 50.43 JUOC- JUOS 8.98 42.70 29.13 PIED- PIMO 25.42 37.07 57.62 Relative areal gain JUDE- JUMO 20.20 10.06 16.20 JUOC- JUOS 3.98 18.44 7.50 PIED- PIMO 8.17 9.55 14.09 Relative areal equivalence JUDE- JUMO 48.24 20.38 21.73 JUOC- JUOS 8.07 57.30 24.04 PIED- PIMO 10.96 30.60 42.38 JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla.

Table 15. Relative areal shifts of modeled sister-species B2a forecasts (%) JUDE- JUMO JUOC- JUOS PIED- PIMO Relative areal loss JUDE- JUMO 51.8 18.3 50.4 JUOC- JUOS 7.2 41.5 29.3 PIED- PIMO 20.0 37.4 57.6 Relative areal gain JUDE- JUMO 20.2 11.5 16.2 JUOC- JUOS 4.6 16.0 8.6 PIED- PIMO 8.2 10.9 14.1 Relative areal equivalence JUDE- JUMO 48.2 24.8 21.7 JUOC- JUOS 9.8 58.5 23.8 PIED- PIMO 11.0 30.3 42.4 JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla.

53

Figure 6. Relative percent areal gain and loss for individual and co-occurring species.

54 Geographic shifts in distribution centroids

Geographic displacement of species distribution centroids is approximately double the displacement in sister species groups (Table 16, Fig.7). Species averages are

238 km (A2a) and 118 km (B2a). J. occidentalis has the greatest shift (415 km for the

A2a scenario and 372 km for the B2a scenario) which obscures the overall average.

Shifts are generally twice as great for the A2a scenario as the B2a scenario, except for J.

osteosperma who has the most similar magnitude of change between the two scenarios,

113 km (A2a) and 93 km (B2a).

Table 16. Displacement of forecast distribution centroids, distance (km) and direction ( º ) A2a B2a distance direction distance direction JUOC 795 162 367 180 JUOS 113 24 93 61 JUDE 167 25 84 15 JUMO 117 3 49 315 PIED 164 7 94 4 PIMO 74 334 13 27 JUOC- JUOS 144 85 178 98 JUDE-JUMO 61 354 49 336 PIED- PIMO 98 38 93 80 Species Mean 238 117 Sister-group Mean 101 53 JUDE = J. deppeana var. deppeana; JUMO = J. monosperma; JUOC = J. occidentalis; JUOS = J. osteosperma, PIED = P. edulis; PIMO = P. monophylla; JUDE-JUMO = J. deppeana var. deppeana + J. monosperma; JUOC-JUOS = J. occidentalis + J. osteosperma; PIED-PIMO = P. edulis + P. monophylla.

55

(a) (b)

Figure 7. Shifts in geographic centroids and changes in richness for A2a (a) and B2a (b) scenarios. Arrows indicate shifts in centroid J. deppeana var. deppeana.(a); J. monosperma (b); J. occidentalis (c); J. osteosperma (d); P. edulis (e); P. monophylla (f); J. deppeana var. deppeana - J. monosperma (G); J. occidentalis - J. osteosperma (H); P. edulis - P. monophylla (I).

Although changes in the mean geographic center of distribution measure the overall distribution shift, there are subtleties which are more particular to each species

(Fig. 3 and 4). Some exhibit latitudinal expansions along topographically complex corridors that are aerially small but have a long reach. P. monophylla, for example, has an overall southeast shift in geographic mean center of 75 km, but is forecast to have a network of suitable conditions that extends up the western Great Basin to 300 km north of the current distribution. The southward shift in mean center results from persistence and expansion into higher elevations across the southern Great Basin, and from an overall retraction from the east central Great Basin. Here the summer rains are forecast to

56 increase, becoming suitable for P. edulis, replacing P. monophylla throughout the eastern Great Basin.

57 DISCUSSION

The accumulated observations of fossil and current piñon and juniper distributions, and their well documented link to climate, provide a rich environment to develop and contextualize models of climate-induced distribution shifts. Climate is clearly the primary driver of piñon and juniper distributions, and climate forecasts are available at temporal and spatial scales relevant to the life cycles of piñons and junipers; however, these forecast distributions are contingent on the GCM forecast behavior of the complex North American monsoon system. The two climate forecasts we modeled with include moderate and extreme scenarios that result in similar trending distributional

shifts, but which vary two-fold in the magnitude of geographic centroid displacement.

Variables driving distribution shifts

Directional shifts in the geographic centroids (Fig. 7) are driven by changes in seasonal air mass position and trend from southwards across the western Great Basin to northwards across the Colorado Plateau (Appendix D). The regional shift of distributional mean centers reflects multi-decadal changes in the average seasonal position of continental air masses as moderated by topography, which acts to amplify or dampen areal shifts. Distribution displacement in the Colorado Plateau corresponds to a northward shift in the summer monsoon, whereas displacement in the western Great

Basin corresponds to increasing fall temperatures compounded by decreasing fall precipitation.

58 The Hadley climate model (HadCM3) predicts the summer monsoon will shift northwards from its current position centered over the Colorado Plateau and center over the northeastern Colorado Plateau. It will extend into the eastern Great Basin, and north into the Colorado Rockies and Wasatch Mountains. P. edulis tracks this movement by contracting from the lower elevations and southern rim of the Colorado Plateau, while expanding in the foothills of the Colorado Rockies, Wasatch Mountains, and the eastern ranges of the Great Basin. The hybridization zone of P. edulis and P. monophylla shifts in tandem. P. monophylla contracts from the eastern Great Basin where P. edulis expands, and expands into the southern rim of the Colorado Plateau where P. edulis retracts. The

Colorado Plateau junipers, J. deppeana and J. monosperma also contract from their lower elevations in the Colorado Plateau but expand into an elevation range that is similar in overall area. The northward shift in these two junipers results from their expansion into the northern Colorado Plateau and surrounding regions.

The mean distribution center of J. osteosperma has a northward shift of ~100 km, persisting throughout the eastern Great Basin as the two piñons shift their overlap up to

300 km west. The northward expansion results from conditions becoming suitable across wide portions of the northern Great Basin and Wasatch Mountains. Contraction from lower elevations in the Colorado Plateau is largely balanced by expansion into higher elevations. The hybridization zone of J. osteosperma and J. occidentalis also shifts in tandem, but their overlapping suitable area retracts greatly in the A2a scenario, with suitable conditions for J. osteosperma expanding into only roughly a quarter of J. occidentalis’ marginal areas. Throughout the northwestern Great Basin J. occidentalis loses most suitable habitat across its entire northern half under both scenarios. The

59 resulting displacement of mean geographic center moves the greatest distance of ~400 km south into the southern half of its current distribution along the interior Coastal Range where it has persisted since at least the Pleistocene (Nowak et al. 1994).

Forecast distribution shifts in relation to recent and historic shifts

The most severe scenario indicates distribution shifts are of a smaller magnitude than Early Holocene shifts, but likely the most rapid in several millennia. Elevation ranges of dominant plant species near the south-west corner of our study area are observed to have shifted an average of 65 meters over the last 30 years (Kelly and

Goulden 2008), whereas the modeled rate of elevation migration is an average of ~100 m/30 years. A global meta-analysis found an average latitudinal shift in a wide array of species distributions of 6.1 km / 10 years (Parmesan andYohe 2003). The models produced in this study indicate a shift in geographic centroids of ~15 km/ 10 years.

Forecast shifts in elevation and geographic centroids are ~1/3 to ~1/5 of early Holocene shifts, which are ~1,000 to 1,500 meters in elevation and ~500 - 1,000 km during the

Holocene (Miller and Wigand 1994, Jackson et al. 2005, Cole 1990).

The three main vegetation zones which piñon-juniper woodlands are forecast to colonize are sagebrush-steppe, ponderosa pine forests, and aspen forests (based on comparison of forecast maps with Landfire). Shifts into all of these neighboring vegetation zones have been observed throughout the 20th century. Sagebrush can facilitate the establishment of both piñons and junipers (Chambers 2001), and their expansion into sage-steppe has been observed to be phenomenally rapid when climatic

60 conditions become more favorable, typified by the 90% increase of J. occidentalis in the north-western Great Basin during the 20th century (Miller and Rose 1995).

The ecotone between montane forests and piñon-juniper woodlands has also been observed to be highly responsive to climatic changes. Piñons generally transition to ponderosa pine throughout the Colorado Plateau and neighboring Wasatch Mountains and Colorado Rockies. This ecotone was observed to shift a distance of 2 km within a decade during a severe drought in the 1940s (Allen and Breshears 1998). Where junipers extend into more northerly latitudes than piñons they generally transition to aspen/mixed conifer forests. Junipers have been observed to rapidly colonize aspen forests (Miller and

Rose 1995) whose sensitivity to drought has led to widespread and sudden decline (e.g.,

Worrall et al. 2008).

Investigations by Rehfeldt et al. (2009) indicate a rapid retreat of aspen from lower elevations in the midlatitude west is likely to be prevalent in the coming century.

The establishment of piñons and junipers is facilitated by both ponderosa and aspen.

Furthermore, the seeds of junipers are preferentially deposited along this ecotone by their avian dispersal vectors (Salomonson 1978, Poddar and Lederer 1982, Vander Wall and

Balda 1977). Forecasted latitudinal shifts, an average of 60 km/ 30 years, are much faster than observed shifts in colonization but are congruent with the widespread drought induced mortality of the last decade (e.g., Gitlen et al. 2006).

Interpretation of forecast distribution models

Models indicate an average of 53% and 62%, for the A2a and B2a scenarios, of each species current distribution will have less than a 40% probability of occurrence. In

61 these areas the species may lose dominance but may well persist in microhabitats

within our modeling resolution. Such “old trees” are very common components of many

populations on this scale and are important seed sources for infrequent establishment

opportunities. As a group, the piñon-juniper woodlands are becoming marginalized in the

lower elevations of the Colorado Plateau and the southern and western Great Basin.

Species richness is constricted to higher elevations of the Colorado Plateau and increases

in the neighboring Great Basin, Wasatch Mountains, and Colorado Rockies. Within the

group, transitions between distributions of closely related species shift in tandem but each

species is unique in overall geographic direction.

Sister species models forecast greater areal gain and less areal loss than their component species along hybrid zones for J. occidentalis-J. osteosperma and for P. edulis-P. monophylla, but forecast greater loss along the periphery. The implications of hybridization may be particularly important for J. occidentalis, which is forecast to lose

nearly all suitable area in the northern half of its distribution based on the species model,

but is forecast to loss much less area and gain some area based on the sister-species

model. The sister species model for J. deppeana var. deppeana-J. monosperma underestimates the areal gain compared to the component species model, and likely reflects the distinctness of these two sympatric species.

62 CONCLUSION

The large overlap between the distributions of piñons and junipers is a result of millions of years of adaptation to similar selective pressures, primarily arid conditions and avian seed dispersal (Malusa 1992, Adams 2008). Their co-dominance across vast areas illustrates their success. But for all their similarities, their consequent response to climatic change is varied between the two groups, with piñons responding primarily to summer precipitation and junipers to changes in winter precipitation. Within each group the individual species have adapted to distinct conditions and also respond uniquely to climatic changes. Model results show an overall shift of the woodland zone to higher elevations and latitudes. Within the woodland zone species vary in their specific adjustments, but show an overall shift towards the north and west with a resulting shift in the piñon component throughout the central Great Basin and the introduction of Colorado

Plateau junipers into the south eastern Great Basin. Comparison of the sister-species models with the species models indicates hybridization may be an important element in distribution adjustments.

63 LITERATURE CITED

Adams, R. A., Schwarzbach A. E., Nguyen S., and Morris J. A. 2007. Geographic variation in Juniperus deppeana. Phytologia 89:132-149.

Adams, R. P. 2008. Junipers of the World: The Genus Juniperus, 2nd edition. Trafford Publishing Co., Vancouver.

Allen, C. D., and D. D. Breshears. 1998. Drought-induced shift of a forest-woodland ecotone: rapid response to climate variation. Proceedings of the National Academy of Science 95:14839-14842.

Allen, M. F., E. B. Allen, J. L. Lansing, K. S. Pregitzer, R. L. Hendrick, R. W. Ruess, and S. L. Collins. 2010. Responses to chronic N fertilization of ectomycorrhizal piñon but not arbuscular mycorrhizal juniper in a piñon-juniper woodland. Journal of Arid Environments, in press.

Anderson R. S., J. L. Betancourt, J. I. Mead, R. H. Hevley, and D. P. Adam. 2000. Middle- and late-Wisconsin paleobotanic and paleoclimatic records from the southern Colorado Plateau. Palaeogeography, Palaeoclimatology, Palaecology 155:31-57.

Araujo, M. B., and M. Luoto. 2007 The importance of biotic interactions for modeling species distributions under climate change. Global Ecology and Biogeography 16:743- 753.

Arritt, R. W., D. C. Goering, and C. J. Anderson. 2000. The North American monsson system in the Hadley Centre coupled ocean-atmosphere GCM. Geophysical Research Letters 27:565-568.

Bailey, R. G. 1989. Explanatory supplement to ecoregions map of the continents. Environmental Conservation 16:307-309.

Barbet-Massin, M., W. Thuiller, and F. Jiguet. 2010. How much do we overestimate future local extinction rates when restricting the range of occurrence data in climate suitability models? Ecography Early View

Barger, N. N., H. D. Adams, C. Woodhouse, J. C. Neff, and G. P. Asner. 2009. Influence of livestock grazing and climate on pinyon pine (Pinus edulis) dynamics. Rangeland Ecology and Management 62:531-639.

Betancourt, J.L., T.R. Van Devender, and P.S. Martin. 1990. Packrat middens: the last 40,000 years of biotic change. University of Arizona Press, Tucson.

64 Betancourt, J. L., W. S. Schuster, J. B. Mitton, and R. S. Anderson. 1991. Fossil and genetic history of a pinyon pine (Pinus edulis) isolate. Ecology 72:1685-1697.

Bradley, B. A., and E. Fleishman. 2008. Relationship between expanding pinyon-juniper cover and topography in the central Great Basin, Nevada. Journal of Biogeography 35:951-964.

Breiman, L. 2001. Random forests. Machine Learning 45:5-32.

Breshears, D. D., N. S. Cobb, P. M. Rich, K. P. Price, C. D. Allen, R. G. Balice, W. H. Romme, J. H. Kastens, M. L. Floyd, J. Belnap, J. J. Anderson, O. B. Myers, and C. W. Meyer. 2005. Regional vegetation die-off in response to global-change-type drought. Proceedings of the National Academy of Sciences 102:15144-15148.

Breshears, D. D., O. B. Myers, C. W. Meyer, F. J. Barnes, C. B. Zou, C. D. Allen, N. G. McDowell, and W. T. Pockman. 2009. Tree die-off in response to global change-type drought: mortality insights from a decade of plant potential measurements. Frontiers in Ecology and the Environment 7:185-189.

Bush, J. K. 2008. The potential role of mycorrhizae in the growth and establishment of Juniperus seedlings. Pages 111-130 in Western North American Juniperus communities: A dynamic vegetation type. Springer 2008

Cayan, D. R., M. D. Dettinger, H. F. Diaz, and N. E. Graham.1998. Decadal variability of precipitation over western North America. Journal of Climate 11:3148-3166.

Chambers, J. C., S. B. Vander Wall, E. W. Schupp. 1999. Seed and seedling ecology of piñon and juniper species in the pygmy woodlands of western North America. The Botanical Review 65:1-38.

Chambers, J. C. 2001. Pinus monophylla establishment in an expanding pinus-juniperus woodland: Environmental conditions, facilitation, and interacting factors. Journal of Vegetation Science 12:27-40.

Chavez-Rameriz, F., and R. D. Slack. 1994. Effects of foraging and post-foraging behavior on seed dispersal patterns of Ashes juniper. The Southwestern Naturalist 38:141-145.

Chaw, S. M., C. L. Parkinson, Y. Cheng, T. M. Vincent, and J. D. Palmer. 2000. Seed plant phylogeny inferred from all three plant genomes: monophyly of extant gymnosperms and origin of Gnetales from conifers. Proceedings of the National Academy of Sciences 97:4086-4091.

65 Chown, S. L., A. A. Hoffmann, T. N. Kristensen, M. J. Angilletta Jr., N. C. Stenseth, and C. Pertoldi. 2010. Adapting to climate change, a perspective from evolutionary physiology. Climate Research 43:3-15.

Christensen, K. M., T. G. Whitham, and P. Keim. 1995. Herbivory and tree mortality across a pinyon pine hybrid zone. Oecologia 101:29-36.

Clark, J. S. 1998. Why trees migrate so fast: confronting theory with dispersal biology and the paleorecord. The American Nauralist 152: 204-224.

Coats, L. L., K. L. Cole, and J. I. Mead. 2008. 50,000 years of vegetation and climate history on the Colorado Plateau, Utah and Arizona, USA. Quaternary Research 70:322- 338.

Cole, K. L. 1990. Reconstruction of past desert vegetation along the Colorado River using packrat middens. Paleogeography, Paleoclimatology, Paleoecology 76:349-366.

Cole, K. L., J. Fisher, S. T. Arundel, J. Cannella, and S. Swift. 2008a. Geographical and climatic limits of needle types of one- and two-needled pinyon pines. Journal of Biogeography 35:257-269.

Cole, K. L., K. Ironside, S. Arundel, P. Duffy, and J. Shaw 2008b. Modeling future plant distributions on the Colorado Plateau: an example using Pinus edulis. Pages 319–330 in C. van Riper III M. Sogge, editors. The Colorado Plateau III; integrating research and resources management for effective conservation. The University of Arizona Press, Tucson, Arizona.

Crookston, N. L., and A. O. Finley. 2007. yaImpute: An R Package for k-NN Imputation. Journal of Statistical Software. 23(10):1-16.

Cutler, R. C. T. C. Edwards, Jr., K. H. Beard, A. Cutler, K. T. Hess, J. Gibson, and J. J. Lawler. 2007. Random Forests for classification in ecology. Ecology 88:2783-2792.

Del Vecchio, T. A., C. A. Gehring, N. S. Cobb, and T. G. Whitham. 1993. Negative effects of scale insect herbivory on the ectomycorrhiza of junile pinyon pine. Ecology 74:2297-2302.

Dismukes, G. C., V. V. Kilmov, S. V. Baranov, Y. N. Kozlov, J. DasGupta, and A. Tyryshkin. 2001. The origin of atmospheric oxygen on Earth: the innovation of oxygenic photosynthesis. Proceeding of the National Academy of Science 98:2170-2175.

Doorman, C. F. 2007. Promising the future? Global change projections of species distributions. Basic and Applied Ecology 8:387-397.

Doyle, J. A. 2010. Function and evolution of saccate pollen. New Phytologist 188:6-9.

66

Elith, J., and J. R. Leathiwick. 2009. Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics 40: 577-697.

Fielding, A. H., and J. F. Bell. 1997. A review of methods for the assessment of prediction errors in conservation presence/ absence models. Environmental Conservation 24:38-49.

Fernando, D. D., M. D. Lazzaro, and J. N. Owens. 2005. Growth and development of conifer pollen tubes. Sexual Plant Reproduction 18:149-162.

Floyd, M. E. 1982. The interaction of piñon pines and gambel oak in plant succession near Dolores, Colorado. The Southwestern Naturalist 27:143-147.

Floyd, M. L., M. Clifford, N. S. Cobb, D. Hanna, R. Delph, P. Ford, and D. Turner. 2009. Relationship of stand characteristics to drought-induced mortality in three Southwestern piñon-juniper woodlands. Ecological Applications 19:1223-1230.

Gascho Landis, A. M., and J. D. Bailey. 2005. Reconstruction of age structure and spatial arrangement of piñon-juniper woodlands and savannas of Anderson Mesa, northern Arizona. Forest Ecology and Management 204:221-236.

Gascho Landis, A. M., and J. D. Bailey. 2006. Predicting age of piñon and juniper using allometric relationships. Western Journal of Applied Foresty 21:203-206.

Gehring, C. A., N. S. Cobb, and T. G. Whitham. 1997. Three-way interactions among ectomycorrhizal mutualists, scale insects, and resistant and susceptible pinyon pines. The American Naturalist 149:824-841.

Gelbart, G., and P. von Aderkas. 2002. Ovular secretions as part of pollination mechanisms in conifers. Annals of Forest Science 59:345-357.

Gernandt, D. S., A. Liston, and D. Piñero. 2003 Phylogenetics of Pinus Subsections Cembroides and Nelsoniae inferred from cpDNA sequences. Systematic Botany 28:657- 673.

Gernandt, D. S., G. G. López, S. O. García, and A. Liston. 2005. Phylogeny and classification of Pinus. Taxon 54:1-14

Gitlin, A. R., C. M. Sthultz, A. A. Bowker, S. Stumpf, K. L. Paxton, K. Kennedy, A. Muñoz, J. K. Bailey, and T G. Whitham. 2006. Mortality gradients within and among dominant plant populations as barometers of ecosystem change during extreme drought. Conservation Biology 20:1477-1486.

67 Gray, S. T., J. L. Betancourt, S. T. Jackson, and R. G. Eddy. 2006. Role of multidecadal climate variability in a range extension of pinyon pine. Ecology 87:1124- 1130.

Grulke, N. E. 2010. Plasticity in physiological traits in conifers: implications for response to climate change in the western U.S. Environmental Pollution 158:2032-2042.

Guisan, A., W. Thuiller, and N. Gotelli. 2005. Predicting species distribution: offering more than simple habitat models. Ecology Letters 8:993-1009.

Hall, M. T. 1961. Notes on cultivated junipers. Butler University Botanical Studies 39:1- 64.

Harris, K. M., S. Sato, N. Uechi, and J. Yukawa. 2006. Redefinition of Oligotophus (Diptera: Cecidomyiidae) based on morphological and molecular attributes of species from galls on Juniperus (Cupressaceae) in Britain and Japan. Entomological Science 9:411-421.

Haskins, K. E., and C. A. Gehring. 2004. Interactions with juniper alter pinyon pine ectomycorrhizal fungal communities. Ecology 85:2687-2692.

Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. Jarvis. 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25:1965-1978.

Helmuth, B., J. G. Kingsolver, and E. Carrington. 2005. Biophysics, physiological ecology, and climate change: does mechanism matter? Annual Review of Physiology 67:177-201.

Hibbet, D. S., and P. B. Matheny. 2009. The relative ages of ectomycorrhizal mushrooms and their plant hosts using Bayesian relaxed molecular clock analyses. BioMed Central Biology 7:1-13.

Holmgren, C. A., M. C. Peñalba, K. A. Rylander, and J. L. Betancourt. 2003. A 16,000 14C yr B.P. packrat midden series from the USA-Mexico borderlands. Quaternary Research 60:319-329.

Ibanez, I., J. S. Clark, M. C. Dietze, K. Feeley, M. Hersh, S. LaDeau, A. McBride, N. E. Welch, and M. S. Wolosin. 2006. Predicting biodiversity change: outside the climate envelope, beyond the species-area curve. Ecology 87:1896-1906.

IPCC. 2000. Emissions Scenarios. Cambridge University Press, Cambridge, UK.

IPCC. 2001. Climate Change 2001: The Scientific Basis. Cambridge University Press, Cambridge, UK.

68

Itami, J.K. and T. P. Craig. 1989. Life history of Styloxus bicolor (Coleoptera: Cerambycidae) on Juniperus monosperma in Northern Arizona. Annals of the Entomological Society of America 82:5827-587.

Iverson, L. R., A. M. Prasad, S. N. Matthews, and M. Peters. 2008. Estimating potential habitat for 134 eastern US tree species under six climate change scenarios. Forest Ecology and Management 254:390-406.

Jackson, S. T., J. T. Betancourt, M. E. Lyford, S. T. Grey, and K. A. Rylander. 2005. A 40,000-year woodrat-midden record of vegetational and biogeographical dynamics in north-eastern Utah, USA. Journal of Biogeography 32:1085-1106.

Jackson, S. T., J. L. Bentancourt, R. K. Booth, and S. T. Gray. 2009. Ecology and the ratchet of events: Climate variability, niche dimensions, and species distributions. Proceedings of the National Academy of Science 106:19685-19692.

Jaindl, R. G., L. E. Eddleman, and P. S. Doescher. 1995. Influence of an environmental gradient on physiology of singleleaf pinyon. Journal of Range Management 48:224-231.

Janson, E. M., J. O. Stireman III, M. S. Singer, and P. Abbot. 2008. Phytophagous insect- microbe mutualisms and adaptive evolutionary diversification. Evolution 62:997-1012.

Johnsen, T. N., Jr. 1962. One seed juniper invasion of northern Arizona. Ecological Monographs 32:187-207.

Johnson, D. D., and R. F. Miller. 2006. Structure and development of expanding western juniper woodlands as influenced by two topographic variables. Forest Ecology and Management 229:7-15.

Keeling, C. I., and Bohlmann. 2006. Genes, enzymes and chemicals of terpenoid diversity in the constitutive and induced defence of conifers against insects and pathogens. New Phytologist 170:657-675.

Kelly, A. E., and M. L. Goulden. 2008. Rapid shifts in plant distribution with recent climate change. Proceedings of the National Academy of Sciences 105:11823-11826.

Kim, H. J., B. Wang, and Q. Ding. 2008. The global monsoon variability simulated by CMIP3 coupled climate models. Journal of Climate 21:5271-5294 de Kort, S., and N. S. Clayton. 2006. An evolutionary perspective on caching by corvids. Proceedings of the Royal Society Biological Sciences 273:417-423.

Landis, J. R., and G. G. Koch. 1977. The measurement of observer agreement for categorical data. Biometrics 33:159-174.

69

Landis, A. G., and J. D. Bailey. 2005. Reconstruction of age structure and spatial arrangement of piñon-juniper woodlands and savannas of Anderson Mesa, northern Arizona. Forest Ecology and Management 204:221-236.

Landis, G., and J. D. Bailey. 2006. Predicting age of pinyon and juniper using allometrice relationships. Western Journal of Applied Forestry 21:203-206.

Lanner, R. M. 1974. Natural hybridization between Pinus edulis and Pinus monophylla in the American southwest. Silvae genetica 23:108-116.

Lanner, R. M., and T. R. Van Devender. 1998. The recent history of pinyon pines in the American southwest. Pages 171–182 in Richardson, D. M. Ecology and biogeography of Pinus. Cambridge University Press, Cambridge, UK.

Lanner, R. M. 1998. Seed dispersal in Pinus. Pages 281–293 in Richardson, D. M. Ecology and biogeography of Pinus. Cambridge University Press, Cambridge, UK.

Lawler, J. J., D. White, R. P. Neilson, and A. R. Blaustein. 2006. Predicting climate- induced range shifts: model differences and model reliability. Global Change Biology 12:1568-1584.

Liaw, A., and M. Wiener. 2002. Classification and regression by randomForest. R News 2:18-22.

Ligon, J. D. 1974. Green cones of the piñon pine stimulate late summer breeding in the piñon jay. Nature 250:80-82.

Ligon, J. D. 1978. Reproductive interdependence of piñon jays and piñon pines. Ecological Monographs 48:111-126.

Lim, Y. W., S. M. Alamouti, J. Kim, S. Lee, and C. Breuil. 2004. Multigene phylogenies of Ophiostoma clavigerum and closely related species from bark beetle-attacked Pinus in North America. Microbiology Letters 237:89-96.

Linton, M. J., J. S. Sperry, and D. G. Williams. 1998. Limits to water transport in Juniperus osteosperma and Pinus edulis: implications for drought tolerance and regulation of transpiration. Functional Ecology 12:906-911.

Little, E.L., Jr. 1978. Atlas of United States trees. Volume 5. Florida, U.S. Department of Agriculature Miscellaneous Publication 1361.

Lyford, M. E., S. T. Jackson, J. L. Betancourt, and S. T. Gray. 2003. Influence of landscape structure and climate variability on a late Holocene plant migration. Ecological Monographs 73:567-583.

70

Madsen, D. B., and D. Rhode. 1990. Early Holocene pinyon (Pinus monophylla) in the northeastern Great Basin. Quaternary Research 33:94-101.

Malusa, J. 1992. Phylogeny and biogeography of the pinyon pines (Pinus subsect. Cembroides). 17:42-66.

Manel, S., H. C. Williams, and S. J. Ormerord. 2001. Evaluating presence-absence models in ecology: the need to account for prevalence. Journal of Applied Ecology 38:921-931.

Mann, M. E., and P. D. Jones. 2003. Global surface temperatures over the past two millennia. Geophysical Research Letters 30:1820-1824.

Mao, K. G. Hao, J. Liu, R. P. Adams, and R. I. Milne. 2010. Diversification and biogeography of Juniperus (Cupressacea): variable diversification rates and multiple intercontinental dispersals. New Phytologist 188:254-272.

Martens, S. N., D. D. Breshears, and F J. Barnes. 2001. Development of species dominance along an elevational gradient: population dynamics of Pinus edulis and Juniperus monosperma. International Journal of Plant Sciences 162:777-783.

Marzluff, J. M., and R. P Balda. 1992. The Pinyon Jay: Behavioral Ecology of a Colonial and Cooperative Corvid. T. and A. D. Poyser, London.

Marynowski, L., A. Otto, M. Zaton, M. Philippe, and B. R. T. Simoneit. 2007 Biomolecules preserved in ca. 168 million year old fossil conifer wood. Naturwissenschaften 94:228-236.

McDowell N., W. T. Pockman, C. D. Allen, D. D. Breshears, N. Cobb, T. Kolb, J. Plaut, J. Sperry, A. West, D. G. Williams, and E. A. Yepez. 2008. Mechanisms of plant survival and mortality during drought: why do some plants survive while others succumb to drought? New Phytologist 178:719-739.

McRoberts, R. E., W. A. Bechtold, P. L. Patterson, C. T. Scott, and G. A. Reams. 2005. The enhanced Forest Inventory and Analysis program of the USDA Forest Service: historical perspective and announcement of statistical documentation. Journal of Forestry 103: 304-308.

Merriam, C. H. 1898. Life zones and crop zones of the United States. USDA Bull. 10. Washington, DC

Mitchell, V. L. 1976. The regionalization of climate in the western United States. Journal of Applied Meteorology 15:920-927.

71 Millar, C. I. 1993. Impact of the Eocene on the evolution of Pinus L. Annals of the Missouri Botanical Garden. 80:471-498.

Miller, R. F., and R. J. Tausch 2001. The role of fire in pinyon and juniper woodlands: a descriptive analysis. In: Galley, K.E.M., Wilson, T.P., eds. Proceedings of the invasive species workshop: the role of fire in the control and spread of invasive species. Fire Conference 2000: the first national congress on fire ecology, prevention, and management. Miscellaneous Publication No. 11. Tallahassee, FL: Tall Timbers Research Station: 15-30.

Miller, R. F., and P. E. Wigand. 1994. Holocene changes in semiarid pinyon-juniper woodlands. BioScience 44:465-474.

Miller, R. F., and J. A. Rose. 1995, Historic expansion of Juniperus occidentalis in southeastern Oregon. Great Basin Naturalist 55:37-45.

Mishler, B. D. 2009. Three centuries of paradigm changes in biological classification: Is there an end in sight? Taxon 58:61-67.

Mitton, J. B., and K. L. Duran. 2004. Genetic variation in piñon pine, Pinus edulis, associated with summer precipitation. Molecular Ecology 13:1259-1264.

Mueller, R. C., B. D. Wade, C. A. Gehring, and T. G. Whitham. 2005. Chronic herbivory negatively impacts cone and seed production, seed quality and seedling growth of susceptible pinyon pines. Oecologia 143:558-565.

Neilson, R. P. 1987. Biotic regionalization and climatic controls in western North America. Vegetatio 70:135-147.

Nowak, C. L., R. S. Nowak, R. J. Tausch, and P. E. Wigand. 1994 Tree and shrub dynamics in northwestern Great Basin woodland and shrub steppe during the Late- Pleistocene and Holocene. American Journal of Botany 81:265-277.

Nyman, T. 2010. To speciate, or not to speciate? Resource heterogeneity, the subjectivity of similarity, and the macroevolutionary consequences of nich-width shifts in plant- feeding insects. Biological Reviews 85:393-411.

Ogle, K., T. G. Whitham, and N. S. Cobb. 2000. Tree-ring variability in pinyon predicts likelihood of death following severe drought. Ecology 81:3237-3243.

Otto A., B. R. T. Simoneit, and V. Wilde. 2007. Terenoids as chemosystematic markers in selected fossil and extamt species of pine (Pinus, Pinaceae). Botanical Journal of the Linnean Society 154:129-140.

Overpeck, J., and B. Udall. 2010. Dry times ahead. Science 328:1642-1643.

72

Parmesan, C., G. Yohe. 2003. A globally coherent fingerprint of climate change impacts across natural systems. Nature 421:37-42.

Pearson, R. G., and T. P. Dawson. 2003. Predicting the impacts of climate change on the distribution of species: area bioclimate envelope models useful? Global Ecology and Biogeography 12:361-371.

Pitterman, J. 2010. The evolution of water transport in plants: an integrated approach. Geobiology 8:112-139.

Poddar, S., and R. J. Lederer. 1982. Juniper berries as exclusive winter forage for Townsend's solitaires. American Midland Naturalist 108:34-40.

R Development Core Team 2008. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07- 0, URL http://www.R-project.org.

Randin, C. F., T. Dimböck, S. Dullinger, N. E. Zimmermann, M. Zappa, and A. Guisan. 2006. Are niche-based species distribution models transferable in space? Journal of Biogeography 33:1689-1703.

Real, R., A. M. Barbosa, and J. M. Vargas. 2006. Obtaining environmental faviurability functions from logistic regression. Environmental and Ecological Statistics 13 237-245.

Rehfeldt, G. E., D. E. Ferguson, and N. L. Crookston. 2008. Quantifying the abundance of co-occurring conifers along the inland northwest (USA) climate gradients. Ecology 89:2127-2139.

Rehfeldt, G. E., D. E. Ferguson, and N. L. Crookston. 2009. Aspen, climate, and sudden decline in western USA. Forest Ecology and Management 258:2353-2364.

Robinson, D. A., I. Lebron, R. J. Ryel, and S. B. Jones. 2010 Soil water repellency: A method of soil-moisture sequestration in pinyon-juniper woodland. Forest, Range and Wildland Soils 74:624-634.

Rodwell, M. J., and B. J. Hoskins. 2001. Subtropical anticyclones and summer monsoons. Journal of Climate 14:3192-3211.

Romme, W. H., C. D. Allen, J. D. Bailey, W. L. Baker, B. T. Bestelmeyer, P. M. Brown, K. S. Eisenhart, M. L. Floyd, D. W. Huffmann, B. F. Jacobs, R. F. Miller, E. H. Muldavin, T. W. Swetnam, R. J. Tausch, and P. J. Weisberg. 2009. Historical and modern disturbance regimes, stand structures, and landscape dynamics in piñon-juniper vegetation of the western United States. Rangeland Ecology and Management 62:203- 222.

73

Rouault, G., J. Turgeon, J.-N. Candau, A. Roques, and P. von Aderkas. 2004. Oviposition strategies of conifer seed chalcids in relation to host phenology. Naturwissenschaften 91:472-480.

Rowland, M. M., L. H. Suring, R. J. Tausch, S. Geer, and M. J. Wisdom. 2008. Characteristics of western juniper encroachment into sagebrush communities in central Oregon. USDA Forest Service Forestry and Range Sciences Laboratory, La Grande, Oregon 97850, USA. 23 pp.

Ruse, M. 1978. What kind of revolution occurred in geology? Proceedings of the Biennial Meeting of the Philosophy of Science Association 2:240-273.

Salomonson, M. G. 1978. Adaptations for animal dispersal of one-seed juniper seeds. Oecologia 32:333–339.

Schulz, C. A. Jagel, and T. Stutzel. 2003. Cone morphology in Juniperus in light of cone evolution in Cupressaceae s.l. Flora 198:161-177.

Schupp, E. W., J. M. Gómez, J. E. Jiměnez, and M. Fuentes. 1997. Dispersal of Juniperus occidentalis (western juniper) seeds by frugivorous mammals on Juniper Mountain, Southeastern Oregon. Great Basin Naturalist 57:74-78.

Seo C., J. H. Thorne, L. Hannah, and W. Thuiller. 2008. Scale effects in species distribution models: implications for conservation planning under climate change. Biology Letters 5: 39-43.

Sequeira, A. S., B. B. Normark, and B. D. Farrell. 2010. Evolutionary assembly of the conifer fauna: distinguishing ancient from recent associations in bark beetles. Proceedings of the Royal Society of London 267:2359-2366.

Soberon, J. 2007. Grinellian and Eltonian niches and geographic distribution of species. Ecology Letters 10: 1115-1123. Fedded Repertorium 117:146-157.

Soberon, J., and M. Nakamura. 2009. Niches and distributional areas: concepts, methods, and assumptions. Proceedings of the National Academy of Sciences 106:19644-19650.

Soulė, P. T., and P. A. Knapp. 2006. Radial growth increases in naturally occurring ponderosa pine trees: a late-20th century CO2 fertilization effect? New Phytologist 171:379-390.

Spaulding, W. G. 1990. Vegetation dynamics during the last deglaciation, southeastern Great Basin, U.S.A. Quaternary Research 33:188-203.

74 Stephenson, F. R., and L. V. Morrison. 1984. Long-term rotation of the Earth: 700 B.C. to A.D. 1980. Philosophical Transactions of the Royal Society of London 313:47- 70.

Strullu-Derrien, C., and D.-G. Strullu. 2007. Mycorrhization of fossil and living plants. Systematic Paleontology 6:483-494.

Swenson, N. G., and B. J. Enquist. 2007. Ecological and evolutionary determinants of a key plant functional trait: wood density and its community-wide variation across latitude and elevation. American Journal of Botany 94:451-459.

Tausch, R. J., N.E. West, and A. A. Nabi. 1981. Tree age and dominance patterns in Great Basin pinyon-juniper woodlands. Journal of Range Management 34:259-264.

Terry, R.G., R. S. Nowak, and R. J Tausch. 2000. Genetic variation in chloroplast and nuclear ribosomal DNA in Utah juniper (Juniperus osteosperma, Cupressaceae): evidence for interspecific gene flow. Botanical Society of America, 87:250-258.

Thuiller, T., L. Brutons, M. B. Arújo, and Sandra Lavorel. 2004. Effects of restricting environmental range of data to project current and future species distributions. Ecography 27:165-172.

Thuiller, W., C. Albert, M. B. Araujo, P. M. Berry, M. Cabeza, A. Guisan, T. Hickler, G. F. Midgley, J. Paterson, F. M. Schurr, M. T. Sykes, and N. E. Zimmerman. 2008. Predicting global change impacts on plant species distributions: future challenges. Perspectives in Plant Ecology, Evolution and Systematics 9:137-152.

Tomback, D. F., and Y. B. Linhart. 1990. The evolution of bird-dispersed pines. Evolutionary Ecology 4:185-219.

Turgeon, J. J., A. Roques, and P. de Groot. 1994. Insect fauna of coniferous seed cones: diversity, host plant interactions, and management. Annual Reviews Entomology 39:179- 212.

Van Auken, O. W., J. T. Jackson, and P. N. Jurena. 2004. Survival and establishment of Juniperus seedlings in Juniperus woodlands. Plant Ecology 175:245-257.

Van Auken, O. W. 2009. Causes and consequences of woody plant encroachment into western North American grasslands. Journal of Environmental Management 90:2931- 2942.

van Mantgem, P. J, N. L. Stephenson, J. C. Byrne, L. D. Daniels., J. F. Franklin, P. Z. Fule, M. E. Harmon, A. J. Larson, J. M. Smith, A. H. Taylor, and T. T. Veblen. 2009. Widespread increase of tree mortality rates in the western United States. Science 323: 521-524.

75

van Ommeren, R. J., and T. G. Whitman. 2002. Changes in interactions between juniper and mistletoe mediated by shared avian frugivores: parasitism to potential mutualism. Oecologia 130:281-288.

Vander Wall, S. B., and R. P. Balda. 1977. Coadaptions of the Clark's nutcracker and the piñon pine for efficient seed harvest and dispersal. Ecological Monographs 47:89-111.

Vander Wall, S. B., and R. P. Balda. 1981. Ecology and evolution of food-storage behavior in conifer-seed-caching corvids. Zeitschrift für Tierpsychologie 56:217-42.

Vander Wall, S. B. 1988. Foraging of Clark's nutcrackers on rapidly changing pine seed resources. Condor 90:621-631.

Wang, B., and Y.-L. Qiu. 2006. Phylogentic distribution and evolution of mycorrhizas in land plants. Mycorrhiza 16:299-363.

Watson, D. M. 2001. Mistletoe- a keystone resource in forests and woodlands worldwide. Annual Review Ecology and Systematics 32219-249.

Weisberg, P. J., K. Dongwook, C. Py, and J. M. Bauer. 2008. Modeling fire and landform influences on the distribution of old-growth piñon-juniper woodland. Landscape Ecology 23:931-943.

West, A. G., K. R.Hultine, K. G. Burtch, and J. R. Ehleringer. 2007. Seasonal variations in moisture use in a piñon-juniper woodland. Oecologia 153:787-798.

West A.G., K. R. Hultine, J. S. Sperry, S. E. Bush, and J. R. Ehleringer. 2008. Transpiration and hydraulic strategies in a piñon-juniper woodland. Ecological Applications 18:911-927.

Williams, D. G., and J. R. Ehleringer. 2000. Intra- and interspecific variation for summer precipitation use in pinyon-juniper woodlands. Ecological Monographs 70:517-537.

Williams, K. D., C. A. Senior, and J. F. B. Mitchell. 2001. Transient climate change in the Hadley Centre models: the role of physical processes. Journal of Climate 14:2659- 2674.

Williams, J. W., S. T. Jackson, and J. E. Kutzbacht. 2007. Projected distributions of novel and disappearing climates by 2100 AD. Proceedings of the National Academy of Science USA 104:5738-5742.

Wilson, C. J., P. S. Manos, and R. B. Jackson. 2008. Hydraulic traits are influenced by phylogentic history in the drought-resistant, invasive genus Juniperus (Cupressaceae). American Journal of Botany 95:299-314.

76

Wilson J. S., and J. P. Pitts. 2010. Illuminating the lack of consensus among descriptions of earth history data in the North American deserts: a resource for biologists. Progress in Physical Geography 34:419-441.

Wingfield, M. J. 1987. Fungi associated with the pine wood nematode, Bursaphelenchus xylophilus, and Cerambycid beetles in Wisconsin.

Worrall, J. J., L. Egeland, T. Eager, R. A. Mask, E. W. Johnson, P. A. Kemp, and W. D. Shepperd. 2008. Rapid mortality of Populus tremuloides in southwestern Colorado, USA. Forest Ecology and Management 255:686-696.

Young, J. A., and J. D. Budy. 1979. Historical use of Nevada's pinyon-juniper woodlands. Journal of Forest History 23:112-121.

Zimmerman, N. E., N. G. Yoccoz, T. C. Edwards, Jr, E. S. Meier, W. Thuiller, A. Guisan, D. R. Schmatz, and P. B. Pearman. 2009. Climatic extremes improve predictions of spatial patterns of tree species. Proceedings of the National Academy of Science USA 106:19723-19728.

77

APPENDICES

78

APPENDIX A

DEMOGRAPHIC LITERATURE

79

Figure A.1. Distribution of piñons and junipers and precipitation seasonality in the mid- latitude Dry Domain, and demographic studies reported in the literature (Table A.1)

80

Table A.1. Recent (1800-2010) distribution shifts reported in the literature. Species and distribution behavior Location Hypothesized/ References observed driver

2 km shift in p-j / ponderosa ecotone N. Arizona regional drought Allen and during 1953- 1958 Breshears 1998 expansion/ densification of p-j into lower Great Basin increased precipitation Bradley and elevations and s. facing slopes and atmospheric CO2 Fleishman 2008 differential mortality of piñons (41.4%) 80 km radius Drought of 2002-2003 Gitlen et al. 2006 and junipers (3.3%) around Flagstaff, Arizona P. edulis densification, 4/5 of population Colorado periods of increased Barger et al. 2009 young trees, expansion/ densification Plateau precipitation during the 19th and 20th century, expansion from old trees p-j expansion/ densification Great Basin expansion from old Jacobs et al. 2006 trees J. occidentalis expansion across exposed/ SE Oregon, expansion from old Johnson and rocky sites SW Idaho trees Miller 2006 Juniperus spp. expansion Wyoming drought Lyford 2003 pervasive increase in mortality across all western US drought van Mantgem et western tree species al. 2009 p-j expansion Central Utah little ice age (mid- Miller and 1800s) Wigand 1994 J. occidentalis expansion into aspen and E Oregon little ice age (late Miller and Rose sagebrush 1800s) 1995 p-j 10-fold increase Great Basin Miller and Tausch 2001 differential mortality of piñons (2-6 times N Arizona Drought of 2002-2003 Mueller et al. greater) than junipers (3.3%), primarily 2005 large trees effected with seedlings surviving under nurse plants

p-j Great Basin Interaction of climatic Ogle et al. 2000 variability and topographic/ edaphic variability P. monophylla and J. osteosperma central and warmer and wetter Tausch et al. expansion into both higher (gradual) and eastern Great conditions 1981 lower (rapid) elevations Basin SYNTHESIS PAPER western US Romme et al. 2009 J. occidentalis 75% young trees with John Day Rowland et al. scattered old trees in rocky outcroppings watershed, 2008 Oregon

81

APPENDIX B

CLIMATE VARIABLES

82

(a) (b)

(c) (d)

Figure B.1 Current and forecast (A2a 2080) degree sums for months with above freezing average temperatures (a, b) and below freezing average temperatures (c, d).

83

(a) (b)

(c) (d)

Figure B.2 Current and forecast (A2a 2080) degree sums for spring (a, b) and fall (c, d).

84

(a) (b)

(d) (e)

Figure B.3 Current and forecast (A2a 2080) precipitation sum for winter (a, b) and summer (c, d).

(c) (d)

85

(a) (b)

(c) (d)

Figure B.4 Current and forecast (A2a 2080) precipitation sum for spring (a, b) and fall (c, d).

(c) (d)

86

APPENDIX C

RANGE OF SPECIES CLIMATIC VARIABLES WITHIN MODELING DOMAINS

87

(a) (b)

(c) (d)

Figure C.1 Distribution of J. deppeana var. deppeana in climate space, presence (red), absence within modeling domain (blue), and absences from all of the mid-latitude Dry Domain (light grey) based on FIA plots. Variable codes defined in Table 3.

88

(a) (b)

(c) (d)

Figure C.2 Distribution of J. monosperma in climate space, presence (red), absence within modeling domain (blue), and absences from all of the mid-latitude Dry Domain (light grey) based on FIA plots. Variable codes defined in Table 3.

89

(a) (b)

(c) (d)

Figure C.3 Distribution of J. occidentalis in climate space, presence (red), absence within modeling domain (blue), and absences from all of the mid-latitude Dry Domain (light grey) based on FIA plots. Variable codes defined in Table 3.

90

(a) (b)

(c) (d)

Figure C.4 Distribution of J. osteosperma in climate space, presence (red), absence within modeling domain (blue), and absences from all of the mid-latitude Dry Domain (light grey) based on FIA plots. Variable codes defined in Table 3.

91

(a) (b)

(c) (d)

Figure C.5. Distribution of P. edulis in climate space, presence (red), absence within modeling domain (blue), and absences from all of the mid-latitude Dry Domain (light grey) based on FIA plots. Variable codes defined in Table 3.

92

(a) (b)

(c) (d)

Figure C.6 Distribution of P. monophylla in climate space, presence (red), absence within modeling domain (blue), and absences from all of the mid-latitude Dry Domain (light grey) based on FIA plots. Variable codes defined in Table 3.

93

APPENDIX D

VARIABLE IMPORTANCE PLOTS

94

J. deppeana var. deppeana J. monosperma

J. occidentalis J. osteosperma

Figure D.1. Variable importance plots for juniper species.

95

P. edulis P. monophylla

Figure D.2. Variable importance plots for piñons.